Philips CD-i

[terse flare]

Yes, all unstable Mains after 20251111 have this exact behavior. It was discussed a long time ago... at that day... that this break is accepted by this community.

Yes. Because the file type .chd or .cue are no longer part of the save file as well. This was also a wish by the users.
Your saves will not be deleted. They just need to be renamed.

At the same time, all CDs in /media/fat/games/CD-i have the same NvRAM. Like the PSX does that too. A sub folder is required to have a unique NvRAM.

I need to go to bed now. I will look at this during the next days, since this might be bigger issue. This also means that there is no overlap between SAM users with CD-i VCD and those that use unstable Main. Because if that was the case, we would have noticed this earlier.

[eager pier]

Night, and thanks for all your work.

[west pagoda]

*all the fish.

[terse flare]

I've crafted an idea which needs further testing and analysis before I put that even into unstable. After some dialog with the prime ministers of dutch discord interactive media, this is the suggestion:

- The machine is reset, when the tray is closed and the image was already used to launch the application and a new image has been mounted. This should fix SAM as well.
- The machine is not reset, when the tray is open and another image with the same NvRAM is inserted, which in our case can only occur when a multi disc application requests the next disc to be inserted. The eject button is missing on the MiSTer.

There are still small issues

• I can't test multi disc change with commercial software, because I don't know of a single title and have no save game to start the scenario.
• According to CD-i Fan there might be a chance that the tray must be opened manually when instructed to do so to change the disc with a multi disc title. This goes against the Green Book but I need to be sure this is true. IF that is true, the core needs to detect this condition somehow to not force a reset.

Since most of you probably don't own a physical CD-i: Like a CD32, a CD-i is always performing a reset when the tray is opened. To avoid the reset, the application must advertise this to the OS before the tray is ejected automatically.

[eager pier]

Is this reset only for multi-disk games? The other cores handle that the same way they handle save files, they don't reset if you are loading an image file from the same folder as the previous one

[terse flare]

Well that behavior still exists if not disabled via Hardware Config -> NvRAM live update. If a different NvRAM is select, it will also perform a reset. The problem is just that the current unstable just continues to read the disc from another image, causing some issues.

I wasn't even aware of that problem until now.

[signal jay]

@terse flare I have a very minor bug report: when unplugging the mouse during a game (at least in Inca), the game audio starts to stutter. I didn't really investigate where it comes from and what the Linux side is doing when this happens. I know you have more interesting stuff to work on now 🙂

plugging the mouse back in fixes the issue

[terse flare]

That issue is still caused by the data stalling problem. I don't have a solution for that right now and the problem is big.
I could make the sector cache even larger but that only makes the issue less likely. PSX solves this by halting the whole machine when this condition is detected. The CD-i core already had a sector cache added. But MiSTer Main is a broken mess. 300ms of cache aren't enough because it is stalled incredibly long. And even if all RAM from the FPGA is used, it would still be insufficient to fix making a screenshot.

[signal jay]

oh, sounds like a deep rabbit hole 🙂

[terse flare]

It is. There are solutions that would require a rewrite. A large 3 second buffer in DDR would fix this. But this is just a patch for something that is incredibly broken with the software running on the Linux side. The MiSTer is supposed to be a low latency machine. 300ms is a very long time and I consider this a small disaster. NEVER ever should system() be used in a fast paced control loop that should be called at least every 5ms.
If MiSTer 2 will ever be a thing, Main should be rewritten.

[eager pier]

Tell sorg that

But yeah, I am guessing it's putting way too much work in the USB side?

[terse flare]

I'm not 100% sure what is taking so long. We have profiling functionality in Main which helps catching long running functions. But I've failed to get a quick understanding of what is going on.

[eager pier]

Is this with a basic setup? There are many Scripts one can load that continue running things in the background on the CPU that might make this harder to track down

Even bluetooth, which would be over usb

[terse flare]

According to my notes, the function int input_test(int getchar)was to blame.

We can of course always make a sector cache bigger. But this will lead to an inrush situation after seeking, putting unnecessary burden on MiSTer Main. Another solution would be to make the core stable against this stalling. The MPEG parts are especially angry about this. The audio and video streams would require a precise stop on - maybe - the video frame boundary.

[signal jay]

I don't mind taking a look at Main, unless you think that's useless or has already been done.

[young grotto]

do you mean it stutters the entire time the mouse is unplugged?

stutters when adding or removing devices usually only persist for like, a few hundred milliseconds tops

[spring gate]

That is a weird issue, is it just a USB mouse or does it happen with anything USB like a controller?

Have we seen that on any other cores before?

[young grotto]

stuttering like, permenantly after you unplug something? no

maybe main is in some kind of a panic if it can't find any input device or something?

[terse flare]

Nah, it is the CD stalling problem again some users seem to still have with the CD-i core. Probably...

[young grotto]

thats usually only one stall

it sounded like they were saying the entire time the mouse was unplugged it kept doing it

[signal jay]

For me it kept stuttering. I plugged back in after 10 seconds or so.

I'm on lastest master though, haven't tried the latest unstable Main or CD-i builds.

[terse flare]

Please don't tell me you are running CDi_20250626.rbf. That would explain a lot

[signal jay]

haha, oops 😄

[terse flare]

[signal jay]

I'm looking at that date and how long ago that is...

[terse flare]

Too be honest, we didn't have another stable release for quite some time now since DVC development has started. One can see here how long it took so far. Please I BEG YOU. Don't use CDi_20250626. It has >>>NO<<< sector cache

[signal jay]

ok, since you are begging, I will think about it 😛

[terse flare]

In case the audience really wants it, we can make a custom release with the latest non DVC stable and fix it with a sector cache. But since we hadn't complains about stability for a long time, I've assumed everyone was just using the unstable one since DVC was a long requested feature. And MPEG is clearly not stable ^^ yet

[tranquil leaf]

Does someone knows how to patch the bios to change the cyan color to black ? It would be better for SAM. We could even use a special core.

By the way, i just tested Mister_SAM with latest Mister and CD-i core from today. I think it's a bit more stable, but it always end up crashing on the menu screen after approx 10/15 videos. Don't know if i can do something to log the problem.

[terse flare]

You are working on the same issue as @river idol right now. A team up might help. I'm not sure if this is related to MiSTer Main, the CD-i core or SAM.

[bright zinc]

We try not to say the dreaded D word here 😄

[eager pier]

https://tenor.com/view/developers-gif-13292051

[terse flare]

Sometimes the solution is almost in reach but one does need to know: Debug Options -> Force Video Plane -> DVC
||Yeah I know, I said the word again. But in my defense, I had to to communicate it.||
I also suggest overclocking the CPU and disable Seeking time for maximum speed.
You will find that in the Hardware config

[young grotto]

idk why but i find it incredibly awesome that cdi can be kind of dual purposed as a videoplayer like this

[terse flare]

That... was the original purpose of this machine. The game stuff only came later because Philips thought it might be a good idea. The CD-i is terribly underpowered for fast action.

[young grotto]

yeah its poetic or something

[terse flare]

I do have to say that I'm not entirely happy with so much stuff happening at the same time. I don't want to be responsible for projects like SAM having issues. Especially because updates of the CD-i core could break parts of SAM if I'm not careful. Maybe a fork would be a good idea.

[young grotto]

they could always just have like their own build they grab from whenever its stable

i wouldn't worry about that kind of thing until you at least have DVC in the main core, anyone toying with it now should expect there be dragons

[stiff flare]

Do I pin this?

[terse flare]

I'm not really sure. It was supposed to be an emergency fix for the way SAM makes use of this core in an attempt to get possible SAM users a hot fix. Sometimes I don't know if I want to roll out a change to everyone or just a small set of users. It does seem that since November, no user has launched a second title after the core was loaded, triggering the bug in the first place. Or they were unphased by chaos that unfolded and just reset the machine.

I mean it already is creating fruit. It seems I've made a mistake. With Main 2025 06 25 I've fixed a buffer overflow in Main, breaking the table of contents. With 2025 09 30 that bug was fixed but the buffer overflow was reintroduced. @river idol has performed a check for me on his machine and yeah, Main is broken and can cause a segmentation fault. It seems I have been too hasty with those changes. Too reactive perhaps.
SAM has a high chance of triggering this, because the CD is constantly spinning and image data is loaded.

[gray badge]

I can test this tomorrow (later today?) while I work, Slamy

The idea is to just let sam run? I have noticed that sam will occasionally crash now that it plays videos in the CDI core, but I hadn’t found a reproducible path for that. Maybe that was this

[tranquil leaf]

A fork would be a great idea. Just a freeze of the core with a new name a soon as SAM works well.

[tranquil leaf]

I did and had the symptom of video switching to the new one without a reset (it works !). But i didn't know if it was the intended behaviour or not...

[tranquil leaf]

Perfect !!

[terse flare]

No, we don't want that. It is a bug that went unnoticed for too long.

SAM Folks, I'm unable to reproduce this Heisenbug. I keep staring at the MiSTer and we both are heavily sweating by now. But I think I've made a blunder when reviewing my own code that might have resulted into the same buffer overflow that I've tried to fix once. In Germany we call that brown meadow programming. It is code from another person that you now maintain. Since the CD-i ODE is just the one of the PSX, because it works, I've might not fully understood what the original intent of some functionality was.
This MiSTer binary is the unstable of today with a small fix applied. Please use it with the CD-i core from #1103404843512631357 message and let's see if this fixes the issue as expected.

[gray badge]

Just so I'm clear, what is the bug? Just the crash?

@fervent urchin which core does SAM use? CDi_dvc_rc3.rbf or CDi_unstable_20260222_1329c1.rbf

[fervent urchin]

it' pulling the latest unstable if no cd-i core is found in _Unstable folder: https://github.com/mrchrisster/MiSTer_SAM/blob/5461329e8048b7c9f805aa4b6977a960cbb1e688/MiSTer_SAM_on.sh#L4141

[solar mirage]

local cdi_url=$(curl -k -s "https://raw.githubusercontent.com/MiSTer-unstable-nightlies/Unstable_Folder_MiSTer/main/db_unstable_nightlies_folder.json" | \

[gray badge]

ok, so if I delete the unstable and dvc core, and replace it with [slamy's test core](#1103404843512631357 message), it will be fine because it will see that as a CDI core

[fervent urchin]

that's right, it will use whatever cd-i core is in _Unstable folder

[gray badge]

Needed to rename to CDi_unstable_20260222_1329c1.rbf otherwise I would get an RBF not found

[gray badge]

Dunno if this was a download error or the bug, but the heart of darkness ad didn't play [2026-02-24 16:39:53] Picked core (samvideo): psx (likelihood: 31.61%) [2026-02-24 16:39:53] nextcore = psx [2026-02-24 16:39:53] /tmp/.SAM_List/psx_tvc_vcd.json found. [2026-02-24 16:39:53] Removing PSX-Heart_of_Darkness.chd from /tmp/.SAM_List/psx_tvc_vcd.json [2026-02-24 16:39:53] Searching for psx\|playstation[2026-02-24 16:39:53] Picked PSX-Heart_of_Darkness.chd[2026-02-24 16:39:53] Video selected via TVC: PSX-Heart_of_Darkness.chd[2026-02-24 16:39:53] Checking availability of http://archive.org/download/mister-sam-cdi-vcd-commercials/PSX-Heart_of_Darkness.chd... [2026-02-24 16:39:54] URL is available: http://archive.org/download/mister-sam-cdi-vcd-commercials/PSX-Heart_of_Darkness.chd Preloading PSX-Heart_of_Darkness.chd from archive.org for smooth playback [2026-02-24 16:39:56] Saved local copy of video: /media/fat/video/PSX-Heart_of_Darkness.chd [2026-02-24 16:39:56] Found VCD JSON: /media/fat/Scripts/.MiSTer_SAM/tvc/psx_tvc_vcd.json [2026-02-24 16:39:56] Duration set to 47 (from JSON: 31)Now playing: PSX Commercial - Heart of Darkness[2026-02-24 16:39:59] CD-i core is loaded [2026-02-24 16:40:00] Timer synced to video: 47 seconds [2026-02-24 16:40:46] Filters for 'psx' already applied this session. Skipping.[2026-02-24 16:40:46] Filtered list already exists.[2026-02-24 16:40:46] Searching for game matching string: Heart of DarknessERROR: File not found - /media/fat/games/PSX/1 USA/H-M/Heart of Darkness (USA) (Disc 1)/Heart of Darkness (USA) (Disc 1).cue [2026-02-24 16:40:46] Gamelist for 'psx' not found in '/media/fat/Scripts/.MiSTer_SAM/SAM_Gamelists'. Building... [2026-02-24 16:40:46] Building gamelist for psx in /media/fat/Scripts/.MiSTer_SAM/SAM_Gamelists ROM check failed. Picking a new ROM to try again (1/3)... [2026-02-24 16:42:01] Filtered list already exists. [2026-02-24 16:42:01] Searching for game matching string: Heart of Darkness Starting now on the Sony Playstation: Heart of Darkness (USA) (Disc 1)

[fervent urchin]

can you play /media/fat/video/PSX-Heart_of_Darkness.chd with the CD-i core?

[tranquil leaf]

Video works for me on latest everything (Test Mister file and Test Core RBF) when played in the CD-i core

Now, i'm currently testing Mister SAM with the same config.

Force Video Plane -> DVC, Overclocking and Disable Seek time make Mister SAM very nice. I hope it'll be stable (still running after approx 10 videos)

Seems solid...

[tranquil leaf]

Still running. I think we can say we're good

[tranquil leaf]

I just stopped now after 6 hours without any problem. problem is fixed for good. I think now the core must be forked with a new name... VCD Player ?

[terse flare]

Thank you for this feedback. Together with the testing time told me via DM, I'm now slightly more confident that this is worth a PR to Main.

I probably will combine this PR with something else. My idea is to start the CD-i core with the standard root folder NvRAM. Is there some argument against it? The nice thing about this is that even without "Forced NvRAM live update" a reset is not required when the image is not in a sub folder.

[terse flare]

I need an opinion. This core has too many permutations in the options, potentially affecting behavior of resets. We now have

Autoplay
Hardware Config -> NvRAM live update -> allows core to rug pull the NvRAM without a reset, the OS usually is ok with that
Debug Options -> CD image live update -> allows changing the image without opening the tray (very dirty, don't do it)

The only normal use case might be Autoplay which uses a custom kernel module for the OS to start the application.
If I go through with the state of #1103404843512631357 message, the core will not perform a reset when the first image is selected, even if auto play is activated. So, if the system menu is visible, auto play will not work. Play needs to be pressed.
So, the current implementation applies auto play only on reset.

If this is not wanted, I need to add a reset when Auto play is activated. There was a time where this was always the case, because the first load of a CD image always loaded an NvRAM as well. So Autoplay didn't need an extra reset handling. With the new suggested Main, the CD-i will always have at least the root folder NvRAM after launch. Because this would spare us one reset.
I hope I got my points along. I think I might be overcomplicating things. It's just... I don't like how the PSX core behaves. I want the CD-i core to act more natural :3

If there is a voice of reason against an initial non empty NvRAM (like cdiemu or MAME do), this voice must now speak up.☝️

[gray badge]

Just to piggy back on this, SAM has been running for close to 24 hours here. No crashes

[terse flare]

This means 3 users have tested it. Very good

Another thing, and the question already scares me. Does SAM feed a custom configuration into the CD-i core? Without affecting the normal operation? Or is it the same?

[gray badge]

I think this was just a hiccup. While I have you, feature request: a commercials only mode 😄

(if that's not already a thing)

[radiant minnow]

@terse flare MadDog MCree seems to look perfect now! No more strange parts in frames! 🙂

[terse flare]

🥴 I wish you were right but I had one macroblock type problem with an Undertaker cutscene after dying. The problem is now that I can't reproduce it. Also it seems that there is a problem with audio video sync in this core. I forgot who mentioned that. The problem is slightly random and I'm not sure what is causing that.

[radiant minnow]

Ooops, I just had a play and (as far as I got, which was a duel after completing the Saloon and Bank scenes) everything worked apparently well! 🙁

So I ran here to tell you

Mr Undertaker has always been nice to me in regards to errors and sync

[fervent urchin]

There is an option for commercial only https://github.com/mrchrisster/MiSTer_SAM/blob/1cb62812793e7166b8afbd597184b8e694b8a624/MiSTer_SAM.ini#L179 - set to only

[solar mirage]

samvideo_freq="alternate"

[fervent urchin]

It does modify the cfg file to set Autoplay to yes automatically

[gray badge]

I should have known it was in there already

[fervent urchin]

no worries, havent actually tested it in the test brnach

[tranquil leaf]

I honestly thinks Mister_SAM needs a way to have its own options without impacting normal use of CDi core... e.g : I would want DVC plane, overclock & seek time disabled for SAM and not for normal CDI.
Is there a way without forking and changing core name before recompiling ? I did it in the past for other cores, it's not so hard and would need to be done once. But i think there also would be caveats to not follow the cdi core progress.

[restive jay]

I'd love to see the audio-visual components of the CDi core eventually forked off

I could envision this core also allowing you to play like, mp3s and flac, and such. Maybe also supporting midis and some game-specific audios like NSF files.

[fiery forge]

https://github.com/ajgowans/wip/pull/16

Where is this dual ram repo 😮

[gray badge]

https://github.com/TheJesusFish/Dual-Ram-Console-Cores

[fiery forge]

oh, totally missed that one. will add right away! thanks! 😄

[eager pier]

Yeah, I would love it if SAM could run in some sort of more sandboxed mode where it uses it's own settings, can't see your own save files, etc. Not sure how possible that is though with the way mister is currently designed though

Maybe some kind of whitelist people can setup of what games are even good for it to load, pretty boring if it keeps randomly loading games that just stay at the title screen

(I remember when I was setting up that NES screensaver, same issue, but it supported downloading from a database of replay files and would load those so you aren't just seeing title screens, usually from a random location too so it's not all from the start of a game)

[young grotto]

can it just use an mgl?

mgls offer exactly that

[lavish crow]

I didn’t realize there was a dual ram PSX core. Are there any noticeable improvements to the stock core?

[gray badge]

The sound is imperceptively better and the gpu is imperceptively slower

[young grotto]

i'd just stick to the single-ram for psx, there's no real benefit and the dualram core is much less tested

[wintry dust]

putting the psx core in there makes me want to revive those monkey gifs of them beating people

[young grotto]

its good for testing your dual ram is working ok? you'll hear it right away if not

[eager pier]

Would an MGL prevent it from accessing your save data?

[young grotto]

yeah you can have an mgl basically instance the entire core with a new name

you might not even realize but the GB/GBC "cores" are one core with 2 MGLs

[tranquil leaf]

Errr... yeah... dumb me ! Why didn't i thought about this option...

[eager pier]

I assumed the Game Boy core was just simply the Game Boy Color core

The Atari 2600 core is just the 7800 core after all

[ionic island]

vice versa

gameboy color is gameboy

it uses a mgl so you can have different screen settings

like filters and whatnot

[stiff flare]

Whoever put those MGL files is probably handsome and very strong

[eager pier]

Ok, well, that rules out Robby. Who was it?

[young grotto]

clicks stopwatch over two hours from MGL mentioned to robby commenting?

[ionic island]

I think it was moondandy

[fervent urchin]

SAM automatically filters static games. it's not perfect but it works pretty well! I created short 90s videos of every game for most consoles and then used ffmpegs scene detection. the results are the blacklist files in the game list folder: https://github.com/mrchrisster/MiSTer_SAM/tree/main/.MiSTer_SAM/SAM_Gamelists

[young grotto]

its impressive the amount of effort you've put into SAM at this point

[fervent urchin]

lol, yeah .. probably spent too much time on it but then again it's been a hobby for the last 5 years or so at this point

[remote rapids]

have you had a chance to see if SAM can detect the user selecting a game via the remote script? To this day if I have SAM running in the background and access a game via remote - SAM must still think the system has been left idle

[terse flare]

Due to the nature of how unstable builds work, it wasn't listed, but #unstable-nightlies message from today contains the latest fixes for CD-i as well:

* CD-i: Fixed buffer overflow
* CD-i: Mount root folder NvRAM after core load
  Instead of starting with an empty NvRAM, "saves/CD-i/CD-i.sav" will be loaded instead.
  Avoids machine reset by NvRAM change in case a root folder CD image shall be used

I've only noticed one upsetting side effect. Since the NvRAM image is now always a valid one, the initial NvRAM writing of the ROM (yes, it sadly does that) will now result into an NvRAM write to SD card with a freshly loaded core.
This could be solved by not using those early writes as a "pending NvRAM state". But I'm not yet sure. In general I hope that this change is not hated.

I'm noticing still some after effects of my cold on my work, so maybe I should stop integrating code in that state. It might have had a bad effect on my latest changes perhaps, I worry.

[fervent urchin]

I got wizzo to support SAM in zaparoo since I'm using that, haven't looked into supporting remote, but I'm sure that could be achieved easily

[strange crater]

I know I have a whole thread on this, but this is part of why I would love to see someone tackle a DVD player core on MiSTer.

Aside from the games & video playback capabilities (which would also likely add the potential for video at twice the resolution for SAM), the base MPEG2 capabilities would be great as a basis for a lot of other projects - LaserDisc-based systems being a big one.

[strange crater]

As it stands, I’d also be curious if the existing MPEG1 implementation for the CD-i core might be possible to use in a ReelMagic card implementation with ao486. There are some quite neat titles which use that card (most notably the MPEG version of Return to Zork).

[restive jay]

Would be cool to see Homebrew CDi ports of Arcade laserdisc titles

[eager pier]

I use remote frequently so that would help, I recall last time I tried to use SAM it kept going off even when I was actively doing something, not sure if it was becuase of remote.

Though sadly remote itself appears to be dead, it hasen't been updated in like three years 🙁

And still has some bugs

[fervent urchin]

just chatting with wizzo about it. support for turning off SAM on core launches is easy, not sure if keyboard detection will work

[fervent urchin]

Thanks to @river idol we now have CD-i commercials playing in SAM. Check it out, it's quite awesome! If you are on test branch, just update and launch MiSTer_SAM_on.sh cdi from ssh

[fervent urchin]

I did start porting a MPEG2 FPGA project, but I'm currently hitting a roadblock. It does produce output on the MiSTer but only with VGA timings currently: https://github.com/mrchrisster/MiSTer_MPEG2

[young grotto]

thats awesome, is there a thread anywhere for mpeg2 stuff? i thought i saw one a long time ago

[eager pier]

I forgot, what was the reason it's better to create a core on the FPGA that can decode MPEG2 video instead of just making the ARM CPU do it?

[young grotto]

the arm cpu isn't actually that strong and its basically stuck trying to software render everything too

it cant even get help with basic scaling or colorspace ops

[eager pier]

I know it's no monster but surely it would be capapible of SD resolution video? My Pentium3 could easily handle that

[young grotto]

your pentium 3 had a graphics card to help it draw

[vagrant shale]

In the early days of dvd they sold mpeg-2 accelerator cards and some dvd-rom drives were bundled with them. Of course, before long, software decoding was fine because CPUs got faster.

[radiant minnow]

Doing it in hardware on a predictable and smooth way instead of some crappy solution drawing to fbdev with broken frame pacing and tearing.
If you can't tell how good videos look on the CD-i core and why that is, then, well...

[restive jay]

Hell in the early days they decoded music in hardware by making CD-ROM drives be actual CD players

[eager pier]

I remember many early CD drives having a headphone jack...

[restive jay]

https://youtu.be/TUdtqtIgzQ0

[valid onyx]

with a internal extra cable to pass audio to the soundcard, so you could hear music at the same time as windows 3.11 / 95 sounds 'effects' 🙂

[terse flare]

This project is an example for the importance of automated integration tests, or their lack of, to be more exact. The most recent core is not even able to play Addams Family without issues. I've noticed that pausing our "Rick Roll Joke Video" is causing audio video desyncing. Wasn't really sure what to expect or if it worked at some point at all and tried Addams Family. Then I've noticed that chapter 3 (Thorny Issues) is having random corruption on the macroblock level directly at the start of said chapter. It's almost we are running in circles around a good solution. A frequent side step. I just wanted to let you know this. I'm not sure which of the recent core releases had the least amount of problems. This reminds me a bit of that time when we had a debug setting for deactivating the MODE2 filter in the CDIC because it's function wasn't understood by me.

[fervent urchin]

Latest SAM test build supports remote now

[young grotto]

is it actually plausible to have automated integration tests for something like a mister core?

[wintry dust]

jotego kinda has one where he runs a simulation and captures some outputs/samples from it

[eager pier]

Nice, thanks

[terse flare]

100%. I do have a fully automated integration test for the video chip, since the output mostly depends on the content of RAM. With a library of RAM dumps and a library of reference outputs, I can check for regressions rather quickly. I just don't have something like this for the CDIC or the MPEG stuff yet.

But I do have a respectable collection of MPEG CD-i software with various use cases :3
https://github.com/Slamy/CDi_FMVTest

[restive jay]

What's the state of CDi Homebrew? I think it'd be cool to make ports of other FMV games

[spring gate]

I think it is basically non existent. I don't think there are the tools to do really do it

[mossy steppe]

there was an independent fmv game on cdi 2 years ago lol i think

https://youtu.be/mEf0jKwBPfQ?si=luy4dDvxPi2xLn07

by this Johnny depp impersonator?

[spring gate]

Hah, well fair enough, this I did not know about!

[mossy steppe]

i'm not saying that makes it a "thing" but someone out there knows how evidently

there's two others i know, that zeldalike one and uhhh the one that's like the 3d racing levels from old snes platformers

[terse flare]

The original tools for authoring discs on MS-DOS still exist and were uploaded by TwBurn once, the author of the CD-i homebrew title Nobelia. So, you need dosbox to generate disc images. A C compiler for targeting Microware OS9 is available as Win32 binary. They work on Windows, but also work on Wine for development on Linux. This is my way of developing for the CD-i. Tools for converting image and audio data to Green Book standard (ICUshell and ACUshell) are still available, run horribly on Wine, but are required to get XA audio tracks. Image conversion should be done by modern tools, written in Python (or something that you like). The CD-i is happy with chunky bitmaps and a color lookup table.

https://twburn.itch.io/nobelia

If you are a programmer and fluent in C, you will hate the original C compilers for the CD-i. It is K&R style. Very archaic.

Compared to the Amiga and C64 development scene, the CD-i tooling landscape was never upgraded to the modern era. Not enough developers. It might be possible that it is one of the smallest homebrew scenes out there.

[wintry dust]

holy K&R

[restive jay]

I wonder how easy it'd be to make an FMV engine sort of deal

[terse flare]

There is only one way to find out. You want it ?
It's yours my friend, as long as you invest some time and effort. Rubies not required.

[strange crater]

I believe this is actually a Video CD with interactive menus/game elements, not a true CD-i game (also runs on other VCD players & systems with VCD adapters).

Believe TWBurn made Nobelia and started work on a port of Skyroads using the MPEG-1 card. There’s also a pretty simple game called Frog Feast, and a SupraKillMinds clone called Super Quartet - believe that’s currently the extent of the homebrew scene.

[mossy steppe]

skyroads, exactly the one i was thinking of

and yes, frog feast is epic

i forgot not everyone knows and loves frog feast

and yes perhaps that is true, i'm unsure of the true nature of that pirates thing and it looks lazy

frog feast is actually one of the games i've never been able to find a rom for in my life for snes, unrelated to this but it seems like no one has ever ripped it and the copies on ebay are insanely expensive

[strange crater]

It’s not an impersonator - they actually commissioned Johnny Depp to do clips for the game in character as an off-brand Jack Sparrow on Cameo.

And yes, that is as awkward as it sounds like it would be - he didn’t record lines for it; it’s just generic congratulations messages that they cut out of Cameos.

[mossy steppe]

i'm glad you know things about this i just only know the bare minimum and it sounds ridiculous

ever since the tim and eric movie i have to assume johnny depp looking guys are impersonators

i've wanted to try skyroads on cdi and couldn't remember the name of the game

[strange crater]

There’s also a pair of 0MHz packs for the original release and the Christmas Special version on DOS, if you want to try the original versions.

[topaz hamlet]

This must be preserved: https://www.ebay.com/itm/285866532720

[quasi viper]

I noticed a lack of color depth in one title at least.

in 'normal' gfx mode, not video

[quasi viper]

I use vmware for stuff that relies on older os'es

[terse flare]

Would you please elaborate on this? Which title? Is this a bug report? Some accuracy issue?

[terse flare]

I could understand this with tools that require Windows 95,98,ME. But with MS-DOS, Dosbox has a certain advantage. There is no boot up time and it is nicely scriptable. I want my work flow to require only one click to compile and link the application and author a disc. Optionally even an integration of a copy operation to MiSTer.

[quasi viper]

I wanted to reproduce it using cd-i emu (i stupidly sold my real cd-i machines) - but cannot for the love of dog figure out how to enable dvc there

[terse flare]

DVC mode in cdiemu is only enabled when one of the possible ROMs was found during launch. Either GMPEG or VMPEG. I think, IMPEG is not supported. But I'm not sure.

[quasi viper]

check. thanks. I will search for those

[strange crater]

I’ve never had luck getting MPEG to work in cdiemu, either.

I got it to detect the VMPEG ROM, and it reported it in the top of the window, but it wouldn’t load any MPEG games, instead just running like a console without the DVC.

[restive jay]

Only two CDi discs use index 02 or higher

Philips CDI Format Test Disc 1 (Europe).cue
Zeneca Pharmaceuticals Group - Experts in Anti-Infectives - Pathogen Pursuit - An Interactive Medical Program (USA).cue

[restive jay]

I also checked the PhotoCD cues just to be safe

[eager pier]

Well, you never know if somebody wants to play CDI Format Test Disk #1 or Zeneca's 30 year old medical program disk

[terse flare]

I'm not sure if anyone has experimented with those yet. I've just tried the Format Test Disc and it seems to not do anything at all. The application is not booting. So something is terribly broken here.
I haven't burned the image yet to try on a real machine but MAME also has troubles booting. The cyan screen never goes away while the MiSTer decided to perform a system reset.
cdiemu even just refuses to load the image because it is "not a CD-i disc".
So I used a hex editor and looked at the image data just to check whether it would make sense to burn the image.
It turns out, the CD sectors are filled with bytes that are counting upwards. So this is NOT a CD-i disc but instead a medium with a probably well defined purpose.

Just to avoid any further hassle, I've made a comment in the testing sheet.

[restive jay]

Redump has a note:

Does not follow Green Book standard completely, will not work in a standard CD-i Player.

For the test disc

WTF is the disc testing then? My patience?

[terse flare]

I tried to find answers but the blog possibly containing them was removed. It seems the disc is older than the CD-i itself

[restive jay]

Why does this have the energy of Cosmic horror?

"Some say the disc is older than the CDi itself"

[terse flare]

Do you have played the video game "Control" ? It's all about SCPs. Maybe it is like the urban legend of Polybius. Maybe this disc was authored in a distant realm around Gravity Falls.

Maybe it is a time traveling medium. It came from the future, affecting the release of the CD-i

[young grotto]

it was recovered from the crashed alien ship at roswell and all cdrom technology is derived from it

[terse flare]

Before or after Megatron was found to establish laser technology?

[terse flare]

Maybe to further elaborate on this topic of integration tests. A machine can be surprisingly self aware. I do recall that we had a recent event for the NES core which was failing at some automated tests. The same is possible for the CD-i as well. Of course you can't check what is not measurable. So the actual video output would only be testable with a "Hardware in the loop" test harness that records the output and analyzes it using some algorithms. This is not new, at the company where I work, we've done this.

[young grotto]

like capture the hdmi output and verify it?

[terse flare]

yes^^ at our company it was LVDS using an industrial capture card. But yes, that would be possible. A video grabber via USB, a script stimulating the MiSTer to do something and some analysis to compare the output against "ground truth".

[young grotto]

hmm

[terse flare]

On the other hand, we do have more possibilities here. Since the CD-i is implemented inside an FPGA, the analysis algorithms could be embedded inside the same circuit.

[young grotto]

could a testrom transmit it results out the jaglink port or something?

[terse flare]

Yes^^ or even better. I need some time to write it down. It seems, it is story time again.

Over the last days I've checked the accuracy of some registers that have influence on the timing of the MPEG video playback. We know that something is wrong and it seems I've really made a blunder at some point on the last release.
There is a register containing the number of pictures in a FIFO. But it was unclear to me WHICH FIFO it is. The output with the decoded pictures? That was my assumption so far. There is also a register containing the DTS, which is a timestamp from the "system stream pack header". It was not clear to me, when this register is written with WHICH pack header.
Both are very important, because the FMV driver synchronizes the video to the audio based on these values.

So, I've created a tool to get a better picture.
To the FPGA community, gtkwave is a nice viewer for visualizing the simulation results from Verilator or other sims.
I've used it differently now. A small CD-i application that plays MPEG streams and records the internal state of both the MPEG Audio and the video driver and reads some registers at the same time.
Whenever some state changes, it is recorded as one entry with all the current values. I think, I've already talked about experiments like these earlier.
Lucky for me, the Audio DSP in the DVC is having a monotonic 45kHz time stamp always readable via register.
The CD-i base case offers no way of reading high precision timers from "user mode applications". Only the OS has access to the timer.
After collecting everything, the output is presented via 19200 baud on the CD-i serial port like this

  0  00000000 00000000 00000020 00000000 00000000 01800118 00000000 000ee54b
  1  00000000 00000000 00000021 00000000 00000000 01800118 00000000 00000153
  2  00000135 0000000a 00000021 00000000 00000000 01800118 00000000 0000499d
  3  00000135 0000000a 00000021 0000c014 00000000 01800118 00000000 000003ca
  4  00000135 0000000a 00000021 00000000 0000b200 01800118 00000000 0000029f
  5  00000135 0000000a 0000000e 00000000 00000000 00100080 00000000 00000052
  6  00000135 00000009 0000000e 00000000 00000000 00100080 00000000 0000016d
  7  00000135 00000008 0000000e 00000000 00000000 00100080 00000000 0000071c

Every column is a value. That goes into a python script which converts this to a VCD for viewing in gtkwave.
And that also is capable of showing real VMPEG data to compare to.

For the keen eye, it is noticable that the number of pictures is increased from 0 to 0xa in an instant. So it can't be the output FIFO. It is actually the input. So, the VMPEG driver keeps track on potential pictures to be decoded.

My VMPEG implementation is different at one point. I give the MPEG decoder 25 frames of output storage. This is something real hardware seems not to have. MPEG1 requires 3 frame buffers. I assume a real MCD251 only has those 3... and not more.
So, pictures are indeed decoded on the fly only milliseconds before they are displayed. My core doesn't replicate that behavior. It just starts decoding as soon as data arrives.

This picture is actually like an MRI. Only a doctor knows what to look for. The next topic I totally got wrong is the concept of a "hanging MCD251". So far, the core suffered from weird artifacts whenever an MPEG buffer underflow occurred. This is because I didn't really knew how to proceed here. The MPEG standard has no concept of package lengths in the elementary stream. The CD-i however has one covered. The driver adds a small code into the MPEG data to tell the MPEG decoder that it has reached the end of the stream.
I'm currently trying to add this concept to the pl_mpeg library, so I also get this finally right.

@river idol has created some videos using their scripts. And one commercial for Litil Divil has caused problems like these. The MPEG stream had an underflow but just halting the decoding and returning that incomplete frame was the wrong approach. A real VMPEG just hangs in there. In general, feeding broken MPEG streams with artificial bugs has helped a lot.

If you look at the waveform again, you might notice B002(picture about to display) and B00E(fresh sequence about to display).
This is an actually malfunctioning MPEG stream that halts the VMPEG card. If it was correctly authored, the distance between those signals would have been 40 ms / 25 Hz for PAL. This is not the case and the CD-i is still happy. Nonetheless, the frame rate became unstable. Future releases of the core will replicate this behavior.

Maybe some more context. This is a broken MPEG stream made up from 2 CD sectors. The second sector does come too late, resulting into this underflow condition. Usually, the stream should be crafted to always have some frames in the buffer.
This is why good tools for MPEG authoring are so important and why some of the newly made discs are failing.

[valid onyx]

I can point you to a company that claims to be experts in videocd encoding, and one of our members here worked there:

#1103404843512631357 message

But: another great debugging session there, @terse flare ! Very nicely done!!

[terse flare]

Changes based on these findings will fix a lot. But the recent problem of missing MPEG audio is not yet solved by this approach. Slow motion on VCDs is also still creating sync problems. The latter is very low prio for me, because it is an edge case. But MPEG audio should be stable...

I just wanted to give an update after a long time, so you don't start to think, that I'm no longer working on this xD

[valid onyx]

I only saw the missing audio once, but I havent been trying the core a lot after our conversation about mad dog

[valid onyx]

haha 🙂 You can take all the time in the world and even if you don't work on it any more, you've given us a great gift still!

[terse flare]

You can replicate it by pausing and continuing Addams Family frequently.

[fervent urchin]

Thanks for the detailed writeup, super interesting!

[restive jay]

No idea if this is the case on actual hardware or not, but I noticed that PhotoCD behavior is extremely weird

IDK how it's supposed to work exactly, but it seems like Photo CDs can save certain settings almost like a bookmark, and it seems like this isn't saving?

[spring gate]

What discs did you try and how did you do the saving?

[young grotto]

thanks for the update slamy!

[stiff flare]

There’s a new update?

[restive jay]

1. The two Patlabor PhotoCDs
2. I tried saving the "High Quality" flag, and possibly bookmark my location (the menu system on this isn't exactly clear)

If the "high quality" flag isn't actually supposed to save (it keeps reverting on me), perhaps there could be a core-level option to always display high quality for Photo CDs?

[spring gate]

Ah, so I also assumed that setting saved and it seemingly doesn't on real hardware which seems a real product fail

Why they didn't let you default to high Res is beyond me

[restive jay]

Also the "autoplay" setting seems to not save

Which, I'm pretty sure that should save, seeing as how I'm not sure how it would do anything otherwise

(There's an Autoplay setting on PhotoCDs, and I'm pretty sure that just bypasses the pre-system menu)

[terse flare]

Oh sorry, Robby. It was only some written text to give some insight. The oven is still too hot for some pie.

[stiff flare]

Nah my bad, thank you though.

[terse flare]

Note in case someone is experiencing it: There is another regression in the core. Starting with 260222, "The Addams Family - Disc 2" is again crashing when the playback controls are opened. That was something that was fixed back in November with the release CDi_dvc_rc3 but now is occurring again. The problem seems to be the timecode of the GOPs that is handled by the MPEG soft core now, instead of hardware state machines.
In general 2 additional problems can be experienced with Addams Family (Disc 1) by pausing and continuing frequently.

1. The movie player hangs and overwrites the OSD graphics with rainbow colors while the movie still plays.
2. The already known problem that audio is lacking that can be fixed by pausing and playing again.
   I'm not sure when these problems actually appeared. They might have been always there or started to occur at some point.

[young grotto]

is the soft core handling the timecodes wrong?

or is it just like, nature of a soft core its not great at timing things

[terse flare]

I'm sorry for miscalculating the shift of the hours and minutes by a margin of 4 bits. I've tested it once with a disc but I should have tested it with a timecode later on a medium. This is why it breaks.
Before the change it was extracted directly from the MPEG stream in Verilog. But that was wrong. It should be matched to the pictures. This is why I've added GOP parsing support into pl_mpeg but I've made a blunder. Sorry for having not tested it fully.

[young grotto]

you dont have to apologize, thats awesome !

[terse flare]

The time code is very coarse since it is only updated once per GOP. So only every 15 frames or so

[young grotto]

oh weird

so it could fail but only if the video was longer than an hour?

[terse flare]

That is the interesting part about the timecode. It is not related to the stream but can be arbitrary and should follow the time code of the movie.

MPEG1 reserves 5 bits for the hours. That is a lot. 🙂

[tame juniper]

dropping in to say go slamy go your work is awesome 🙌

[spring gate]

God a technical question for Slamy: so on Jag CD various settings for the CD/VLM player are saved on an eeprom (volume set etc.). How is similar settings savings done on CDi consoles?

[terse flare]

If you go to settings on the system shell, you will notice some options. The volume is probably the most problematic one, because the default is not maximum. All of them are stored on the NvRAM. There are also hidden settings like the number of controllers attached to the machine and whether it is PAL or NTSC. Those hidden settings are only configured by reset. This is why Hardware Config -> NvRAM live update is a problematic setting. You are able to start into a weirdly configured machine. A CD-i always writes the NvRAM during bootup. This is also the reason why the newest unstable MiSTer Main causes a pending NvRAM write, because it always starts at least with the root NvRAM.

[spring gate]

Ah, so there is a ram chip on the board that this stuff is saved to that gets accessed and settings pulled from on boot?

[terse flare]

Yes. I wouldn't call it a RAM chip. That would be too friendly. It is a diabolic combination of 8kB SRAM, a coin cell battery, a crystal oscillator in one enclosed package. Next to the drive mechanism it is a main cause for failing CD-i players because it happens to be also soldered to the board.

[spring gate]

That makes sense, it would need a battery to store that info when powered off. Same battery that saves the RTC time?

[terse flare]

Also the same battery that stores the save games. A CD-i is very limited when it comes to capacity. This is a problem with games like Burn:Cycle that take up 11% per save file.

[spring gate]

Got you

Seems odd that they had this functionality and didn't use it to save the Photo CD player settings as well, they really dropped the ball there with that player

[terse flare]

Maybe there were other priorities. The CD-i design allows for continuous writing. A VCD - for example - is frequently writing the time code there during playback. So if you watch a movie, you just turn off the player without a proper shutdown and it will continue like a stopped video tape.

If it was a CD-i 450 you wouldn't even need to press a button to resume. Just give the player power.

[spring gate]

I still just find it baffling you can set thing like high resolution, and it forgets it as soon as you progress to the next photo. From a UX and product perspective, it is either very bad design or a bug they missed (assuming it is like this on all models)

[eager pier]

Ah yes, that thing. Can't the battery dying also possibly prevent your system from working because then SRAM will just be full of uninitialized data?

I've seen people have to literally drill into that chip to replace the battery

[terse flare]

According to the CD-i hardware specialist rosewood, that is possible. But my Mono I PCB that I use to reverse engineer also no longer has charge. And so far... it is fine? He offered me a repair since we are both German and a delivery via mail would be possible. But I've declined until this core is finished.
Concerning the drill, I've used a dremel on my first model 450 that I bought shortly after the release of the AVGN episodes. The operation is dangerous because of the dust. Eventually it worked, but the laser was weaker afterwards for some reason. 🙁

[eager pier]

Some of the modifications appear to also add a capacitor?

[terse flare]

That is NOT a capacitor. That is a 32kHz crystal for RTCs

[eager pier]

Maybe that has something to do with the laser

Ah

[young grotto]

stuff like this is exactly why i'm so glad we have a core now

[eager pier]

Yup, those appear to be the entire chip removed from the package, it's normally a lot bigger

[tranquil leaf]

From where i come, we call it a DALLAS Real Time Clock

[tranquil leaf]

That's the same kind of sh...

[eager pier]

Most people appear to call it the "Timekeeper", at least in terms of the CD-I

[tranquil leaf]

Yeah, Dallas RTC is for old PC... It's also a battery and an oscillator soldered to a board and is the cause of problem with old motherboards

[eager pier]

Most of the DIY attempts I see to replace the battery have drilled off just a half of the chip that has the battery instead of the entire casing

[tranquil leaf]

https://www.ebay.fr/itm/152608537289

[eager pier]

SegaCD is slightly unique too. While it is a button cell that is soldered in instead of being embedded in a chip, it's rechargeable unlike nearly every other soldered button cell in a console. Would probably not be a good idea to replace it with a standard lithium battery

[terse flare]

Maybe a core specific question. There are 3 issues that I currently have difficulties to debug:

• Missing audio after continue from pause with Addams Family (both discs), I really have difficulties to consistently reproduce this to even analyze it. One reason for that is...
• Playback control module crash with Addams Family. I'm able to reproduce it, but I fail to understand the problem because it seems to be not a single process but a multi tasking one. It also seems this is not a regression. Since November when this title started working, there was always a chance for it to happen.
• Missing audio with Mad Dog McCree? I don't even know how to reproduce this.

So, the question is: Are there existing issues in the same direction that I'm not aware of that might help? Sometimes a fix of one issue, fixes another.

[nimble epoch]

I didn't seem to have missing audio during my test with Mad Dog McCree?
Is there a certain section where it happens?
Maybe I missed it.
Did have an issue during callibration mode, invisible crosshair.

[terse flare]

@valid onyx has mentioned it once. But I'm not sure I've yet read about a way for consistent reproduction. There is a randomness to it. Those are the worst problems to debug. Invisible crosshair also should not occur. If you know how to reproduce this, please tell me.

[nimble epoch]

Just tested with the latest version, the crosshair is visible during practice mode.
There was no crosshair during calibration for the gun on real hw if I remember well, just had to aim with my Mad Dog blue gun attached with 2 cables and just shoot the targets.
Seems to be perfect to me.
I will record a full playthrough to determine if I have any audio issues.

[terse flare]

The hangup of playback controls during pause is also reproducible with "Imagination in Motion". It is interesting though that it fixes itself after 4 minutes. That is curious.... I do wonder if there is a correlation with Lost Ride sometimes not being able to pause itself during the map screen. Hmm... maybe I should construct something with SignalTap to catch this exact case. I have something in the oven that should fix some of the timings. But maybe it is not yet enough.

Sometimes I have difficulties to decide when to push out another release. It might make things worse after all.

[valid onyx]

Yeah I had it once. One video was without audio. The next had it again, and I could not reproduce it either

[iron blaze]

@terse flare i heard the CD-i core was having trouble streaming from CD when screenshots and stuff happen. Give this main a try.

https://github.com/tonytoon/Main_MiSTer/tree/simd_memcpy

[terse flare]

Since you've put quite some effort into this, I've analyzed the effects of the changes with SignalTap. I'm not sure whether I've ever shared shots like these, so here we go. With current Main ae8e243, the cache levels look like this when playing some CDDA.
From left to right: Raw Screenshot, PS3 Replug, Scaled Screenshot. As one can see, the scaled screenshot is draining the cache to zero. I've also noticed here again how dangerously close we are from underflowing just because of some USB activity.

With your changes applied, it is not even possible to see a dent in the cache level. So it works! I assume you have done what was suggested a while ago and solved this with multi threading?

Good job! With your work, it should be finally possible to make non-destructive screenshots from the CD-i core.

[iron blaze]

Basically. Rewrote the memcopy to use simd which sped it up way more than I expected, and it handles image processing / saving using an asynchronous worker.

[eager pier]

By image processing do you mean the disk image, or something like taking a screenshot?

[iron blaze]

It copies the raw RGB image data for the screenshot to a memory buffer, then rescales that buffer (if needed) and converts it to a PNG and writes to disk. Everything after the copy is done asynchronously now so it doesn’t make anything wait, and the copy is drastically faster now too.

[eager pier]

Nice!

[bold dragon]

So now you have a stronger reason for the commit; it solves a real-world problem and isn't just a technical demo

[iron blaze]

Exactamundo

[young grotto]

thats some good old fashioned optimization work i love it

[bold dragon]

True dat. Good work.

[terse flare]

I may need to get something off my chest, since the last unstable build of this core was about one month ago.
This fact alone seems to be generating pressure inside myself, which - I know - is not very healthy thinking. I'm noticing that my mental health is degrading over the course of the last weeks and my cognitive capabilities are also slightly decreased. This is especially bad at work, where I’m expected to perform for the money I’m given.
I'm not 100% sure what is causing this or if this is even related to this FPGA project or to other parts of my life (e.g. work, the German weather).
Since I've noticed discrepancies with some of the VMPEG registers, as I've mentioned earlier, I've tested and tested but so far those efforts are not giving fruit to an improved core. Only more visible bugs in the already released ones. And that saddens me slightly because I’m constantly afraid of failing.
I think, I should take a break for a few weeks. I might be slightly burned out. But communication is key, I don’t want to ghost the fans of interactive entertainment and I would you to be aware of the current state.
I might advertise myself as a dude in a Turrican suit, but the human inside is actually rather fragile.

[iron blaze]

Take a break - they're invaluable and you've been busting your butt.

[young grotto]

as someone who enjoys watching the process as much as the end result take all the time you need 🙂 we'll be here to cheer you on whenever you feel like coming back to it

and yeah the progress already is ridiculous so even if you stopped now you've achieved something incredible

[iron blaze]

@terse flare could make sure my latest changes didn't introduce any regressions to my fix for cd access stalling for screenshots? I'll submit my PR if they're still good.

https://github.com/tonytoon/Main_MiSTer

The draft PR is at: https://github.com/MiSTer-devel/Main_MiSTer/pull/1126 -- if you can note that it fixed the issues you were seeing that'll help get it approved.

[eager pier]

Please take a break, it's fine, nobody here will have an issue. You've been working insanely hard on this

[clever storm]

@terse flare look after yourself, we all know only too well nowadays how important our mental health is. ❤️

[stiff flare]

”Mah @terse flare, taking a break is what all true warriors strive for!”
|/

[celest hollow]

Relax my friend. Take a well deserved break

[tribal axle]

The world is crazy right now. I think we all have a constant impending feeling of doom. Definitely take a break. Take care of yourself.

[celest hollow]

I'll consider myself fortunate then, I don't feel that way at all

But if Slamy is feeling undeeded pressure, the dude should take all the time he needs. happiness and life come first

[unborn grove]

Alles Gute, Slamy. Ich weiß deine Arbeit wirklich sehr zu schätzen! Liebe Grüße aus Berlin

[spring gate]

Look after yourself Slamy, and hopefully Spring has finally arrived in the Ruhr Valley. Today feels like the first dry, sunny and not windy day in Edinburgh in about five months. Spirits lifted. 🙂

[iron blaze]

Even if nobody seems to be saying anything about it, know your work is appreciated. For every person who pipes up, there's another ten staying silent.

[nocturne vortex]

but deadly

[strange crater]

Slamy, you’ve taken a platform which was previously tied to a dying piece of hardware and single-handedly made it more accessible than it’s ever been, allowing people to play its games that would have never gotten the chance before. The fact you’ve gone on top of that and implemented a black-box accessory that few people before you even really understood the inner workings of is nothing short of incredible.

Take a breather, man. You’ve earned it.

[sterile cedar]

I had never tried the CDI before your core Slamy, as many others here. Pretty sure no one has gotten so many new people to try CDI games as your project did. Whatever you do going forward, thanks for all the hours you have put into this.

[restive jay]

To add to what Elphive said, the CDi is basically a whole lost technology. It's crazy that the CDi is technically a video playback standard and currently this is the most complete implementation

Honestly this is what cinched the MiSTer for me. "Sure I could just fiddle around with software emulators, but can software emulators do Flowers of Robert Mapplethorp? I rest my case"

[lavish crow]

Yeah it’s pretty incredible. I think the CD-i is the first/only console core on MiSTer or any FPGA device that you can say is clearly head and shoulders above what software emulators are available

I don’t think there is any implementation of the DVC outside of this core. Thanks for making me a fan of the CD-i Slamy! It’s a fascinating piece of hardware with such a unique and vast software library. It’s better viewed as an experience than a strictly a game console. I can’t think of a better time capsule of 1990s culture

[spring gate]

You also have also gifted us all a way to experience the majesty of the Photo CD format, for which I am eternally grateful. 🙂

[bold dragon]

PC Engine was, and then I shared the learnings back to Mesen2, so they're about even now

And I would say that Saturn is now too

[spring gate]

Super Cassette Vision may have an edge on soft emulation still

[restive jay]

Honestly the "head and shoulders above" part about MiSTer is that the UI works well and with very few exceptions you don't have to decide between multiple cores

[spring gate]

I don't think any are quite as profound as CDi though, which has eclipsed the software emulators

[restive jay]

Probably a lot of small things that add up

[lavish crow]

I do agree that MiSTer’s Saturn core is probably better than Saturn software emulators, and N64 is about even with Ares (I think they both play every N64 game w/out issue), but CD-i on MiSTer has a significantly higher compatibility than any software emulator afaik

For the really obscure stuff like Super Cassette Vision, Casio PV-1000, Bandai Super Vision 8000, Mega Duck, etc. I think MiSTer is king too

[spring gate]

Yeah and CDi software emulation development has slowed to a snails pace as well

[iron blaze]

I will occasionally joke and shitpost about the usefulness of a CD-i core but it's a historically fascinating and important system to be able to replicate - that makes me support it all the more. The games aren't good, but the system itself is the centerpoint of a very interesting time in gaming history which makes it a more than worthy endeavor.

[zealous wagon]

hi all, i'm struggling to get my head around generating a split rom please? i can't get dvc to work

[spring gate]

Everything you need is in the BIOS DB, just select from update all menu

[zealous wagon]

ok great thank you will try now

[terse flare]

Both should work. I've tested it once, I think. The split ROM makes sense, but is not part of the MAME set. The ROM loader will always load the full ROM, but the memory mapping ensures that the second half is always a mirror of the first.

[terse flare]

You should consider some of the not so popular games as well. There are AAA games on this platform. Lucky Luke, The Apprentice and Burn: Cycle are examples for highly polished titles.

[zealous wagon]

Thank you, I have managed to get the boot roms. Is only the stable version unable not to support dvc. The core is excellent, some magic work you've done.

[strange crater]

I’d also highlight Zenith - definitely worth checking out, as well.

[harsh ember]

Burn out sucks. Legit take the time off working on the core. A few weeks or a few months.

It's not worth ruining yourself over.

[rough sable]

take care of yourself @terse flare. The community appreciates you a ton and even if you walked away for half a year or just let it be where the core stands now, you have given us a lot and we are thankful. Don't impose self-inflicted pressure, I get wanting things to be perfect because I do it as well, but small wins equal large wins in the grand scheme of things.

[candid adder]

Just tested running an audio CD tonight and it hangs at the bios screen. Figured to put it here as I was in the Jaguar section testing all the CD cores on here while I was able to and wanted to see which cores audio CD’s worked in and which didn’t.

CD hangs on the CD-i. It’ll loop where you see the cursor, but the orange light will still blink.

It works on the NeoGeoCD, MegaCD, PSX, and Saturn cores without issue and on the Jaguar core converted to CDI. It doesn’t work on the 3DO or CD-i cores.

So figured I’d let you know so you can look at it when you’re ready.

[terse flare]

When I first read your report here, I was confused, because I've played an Audio CD only a week ago on the core. It turns out that - for some reason - the Audio CD is only played back when the CD is loaded into an already waiting player shell and no reset is induced by putting in that disc. If a reset was performed, the Audio CD is not correctly played.
Concerning the orange LEDUser LED, this means, the NvRAM was changed and a write to SD card is pending to be executed when the OSD is opened.
This is quite a finding. Thx. These are the symptoms of the CD player application not being able to read the Table of Contents. I'm not sure why this happens and put that onto the TODO list. Nonetheless, a freshly booted core with Debug Options -> CD image live update and Hardware Config -> NvRAM live update should be able to play an Audio CD on your setup too.

[candid adder]

I’m at my first doc of the day so I’ll let you know how it goes when I get home. Got my three month checkup today and then my gel shot injections for my knees in the afternoon so I’ll be busy until then.

[candid adder]

I didn’t see that under debug options but I turned the second thing on. Also turned on SERVO Audio CD under debug options and same issue as before

Went right to this core from power on. Didn’t load any other cores before it

Using the core from Feb 22

[terse flare]

Which MiSTer Main are you running?

[candid adder]

The latest unstable.

The one with the 3DO fixes in it as well for the OSD menu

[terse flare]

I've now checked with said MiSTer unstable from the nightlies and against the CD-i core 260222. It is working on my machine... huh...
We also had an in-between release designed to fix mostly issues with SAM. #1103404843512631357 message
I don't recall why I haven't made an unstable build from this. I assume I've thought to do that after I've expected to fix some bugs in the video-card-not-to-be-named.
I'm not sure how I can help you. I would expect to work at least once. But the core must not make a reset.
There is one thing, you can do. You can upload the disc image somewhere, so I can analyze it. So far, I've only tested my core against the CDs of my childhood. :3
And those were both working.

[candid adder]

I can put it on my Google Drive

Listen, I get the burnout man. I’m dealing with it myself currently (and about to make some major life changes due to it). I’m not expecting immediate results at all. This is one of those me letting you know and when you can get to it, then you get to it.

I was just told to report it here and that’s what I did. I apologize if it seems like a little much right now

[terse flare]

One of the biggest issues with this project is the intense complexity of the VMPEG hardware. It is difficult and I don't have any help. At some point I had hope that I could motivate the software emulator devs. One seems to have given up. I can understand that after 10 years of development. I'm not angry, because that person has kickstarted CD-i emulation from zero. One had an emotional breakdown and gave up CD-i entirely because the first one was too mysterious about secrets. And another one is too busy with fixing bugs he doesn't understand, because he is not even willing to invest into CD-i hardware. I think that sums it pretty much up. Maybe I'm too harsh.
When it comes to fixing Audio CDs, I expect this to be a problem one could solve - a german saying - on a single butt cheek.

When I have the disc image here, It either runs without hassle or it is indeed an issue that needs to be checked. But then I already have it an can analyze it when I feel better.

[candid adder]

Here’s the image. It works fine in the CD cores other than this and 3DO.

https://drive.google.com/drive/folders/1RacKEviJfDJcsEchfVx1FiTb0Qw3I4hB?usp=sharing

Even had it working in the Jaguar core converted to CDI

[wintry dust]

One had an emotional breakdown and gave up CD-i entirely because the first one was too mysterious about secrets. lol/wtf/sadface

[terse flare]

Thank you. I've downloaded it and I'm able to reproduce it locally. It is weird. It feels like the media player is trying to read the TOC but somehow is not happy with what it gets. I've made one try on the PSX core and it plays normally. That is ... surprising... Thank you for this finding.

[candid adder]

No problem. Thank YOU for the core!

Though i should thank that one user last night that asked if audio CD worked on the 3DO core which led me to test all the other CD cores out of curiosity.

@spring gate since you asked and I tried, you deserve credit for this too.

[stiff flare]

Look at it this way dude. You’re king of CDi now. This core is the foremost and best CDi emulator in existence.

https://tenor.com/view/the-king-hyrule-zelda-ytp-eyes-looking-gif-22475607

[wintry dust]

no pressure or anything

[gray badge]

On the bright side, mentioning DVC is no longer taboo

[iron blaze]

I wouldn’t even know where to get started doing fpga dev. I feel like a visiting wizard who’s way out of his element at the Miskatonic Institute of State Machines

[quasi viper]

https://news.ycombinator.com/item?id=46150173

meifun

Ask HN: How does one get involved in FPGA development?

How does one get started with FPGA development in 2025? I have code that runs on my MacBook and I met a trader who suggested I get my code running on FPGA hardware. He didn't know how exactly to get started.What is a viable path in very late 2025 for a hobbyist to get started?

Hacker News · Dec 4, 2025, 5:24 PM

[tame juniper]

thank you slamy. the community all very appreciative of everything you’ve done here, and obviously also want what’s best for you. enjoy the break and time away from it, and ty again 👑❤️

[nocturne vortex]

First you deliver a CD-i core and now this saying, amazing. I assume the etymology of one butt cheek is gas from all the wurst

[terse flare]

I think, you got the metaphor wrong. Or maybe I did for all those years? xD You see, when doing a job in the office, humans are sitting. And usually they sit on their butt. So if you only need half of yourself for a task, you can sit on the edge of the chair. I hope this is where this saying comes from.

[nocturne vortex]

When building a CD-i core you don’t need the whole chair because you’re always on the edge of your seat in announcer guy voice

[west pagoda]

“You paid for the whole core, but you’ll only need…….the EDGE!”

[iron blaze]

@terse flare I'm taking a stab at the USB devices causing stalls next: https://github.com/MiSTer-devel/Main_MiSTer/issues/1140 do you have any more information on what happened when you saw USB stuff making things go crazy?

(the screenshot stuff got accepted into main so screenshots shouldn't cause stalls anymore)

[young grotto]

i think zakk knows a bit about it, something to do with the entire like, controller statemachine getting rebuilt on every new device

worst case is diconnection of a bluetooth device iirc

which is the same issue as usb with some additional bt specific overhead

[iron blaze]

I've actually got most of a fix in place - I'm looking for known scenarios or other possible info beyond what I can replicate on my end just repeatedly wearing out USB ports repeatedly plugging things into and out of them in assorted orders.

Hadn't heard of the BT part.

[young grotto]

yeah if you have the like, bt sleep thing turned on in the ini, it sends idle bt devices to sleep

and you get a big old stutter

happens to me a fair bit since i use a bt pad to get into a game then switch to snac a lot

bt_auto_disconnect - Amount of minutes of inactivity before Bluetooth disconnects automatically. Default is 0 (off).

[iron blaze]

I'll look at that part as well and see what I can figure out.

I'm not sure how much I can improve it all, but taking a stab at it should be a learning experience at least.

[terse flare]

I gave up very early on during analysis to avoid creating yet another construction site. I did increase the threshold for the spike profiler from 1ms to 20ms - to avoid noise - and noticed this when plugging in a PS3 controller.

+----- Name -----------------------------------------+ Inc(us) + Exc(us) +
| co_poll                                            |  179568 |       9 |
|   user_io_poll                                     |      12 |      12 |
|   input_poll                                       |  179546 |      18 |
|     input_test                                     |  179528 |  179528 |
+----------------------------------------------------+---------+---------+

I actually didn't knew about bt_auto_disconnect, so I gave that a try with a value of 1 to see how bad it actually is, since users have reported problems when watching a movie

+----- Name -----------------------------------------+ Inc(us) + Exc(us) +
| co_poll                                            | 3719015 |       5 |
|   user_io_poll                                     |      11 |      11 |
|   input_poll                                       | 3718998 |       9 |
|     input_test                                     | 3718988 | 3718988 |
+----------------------------------------------------+---------+---------+

It's awful. Even worse. >.< No wonder, it is an issue.

[iron blaze]

That helps out with stuff to look at, thanks!

[terse flare]

I've just noticed, it is not 371 ms. It is 3.7 seconds!!!!

[young grotto]

that sounds about right

thanks for looking into this stuff deepthaw, its the unglamarous side of mister development but its also where you can get wins across all cores

i actually dont use a lot of wired usb controllers so i didn't realize that case was so much better

3.7 seconds is absolutely crash-town right

[bold dragon]

Yeah BT is a mess of complications and slowdowns

[young grotto]

idk how viable it is to async but that seems the obvious direction

[iron blaze]

That's what I'm going to have to do I suspect. Even if it's just one device, that single device can stall for ever.

[terse flare]

Your Snatcher image is different to the CUE files I've seen before. There are no entries for INDEX 0, which causes problems with the CUE loader.
Take this. It is the latest MiSTer Main, released yesterday by Sorgelig + these changes for you to test:

CD-i: Fixed Q subcode of every other sector during Lead-In
CD-i: Fixed single bin images without Index 0

It also turns out that I ... broke the same code for subcode data like 5 times by now :-(. The TOC reading was malfunctioning, causing every other Lead-In sector (the area of the table of contents) to be simply zero. This hasn't bothered the CDIC driver for reasons I don't understand. This is also something that I could have never seen in the Verilator simulation, because the Linux side on the MiSTer was the problem. Please combine this with #1103404843512631357 message and I think, It should be fine.

To even understand this, I've written a small assembler program as ROM replacement to dump the Lead-In area to the UART and noticed the issue.
Here for technically affine CD experts to compare, to understand what I mean:

Broken:

 01 00 01 00 00 00  00 00 02 00 B1 0B  
 00 00 00 00 00 00  00 00 00 00 00 00
 01 00 01 00 00 01  00 00 02 00 1B 5A
 00 00 00 00 00 00  00 00 00 00 00 00
 01 00 01 00 00 02  00 00 02 00 F5 88
 00 00 00 00 00 00  00 00 00 00 00 00
 01 00 03 00 00 06  00 07 03 60 29 2F
 00 00 00 00 00 00  00 00 00 00 00 00
 01 00 03 00 00 07  00 07 03 60 83 7E
 00 00 00 00 00 00  00 00 00 00 00 00
 01 00 03 00 00 08  00 07 03 60 E6 87
 00 00 00 00 00 00  00 00 00 00 00 00
 01 00 A0 00 00 12  00 01 00 00 61 BA
 00 00 00 00 00 00  00 00 00 00 00 00
 01 00 A0 00 00 13  00 01 00 00 CB EB

Fixed:

 01 00 01 00 00 00  00 00 02 00 B1 0B
 01 00 01 00 00 01  00 00 02 00 1B 5A
 01 00 01 00 00 02  00 00 02 00 F5 88
 01 00 02 00 00 03  00 03 28 45 3F B0
 01 00 02 00 00 04  00 03 28 45 58 64
 01 00 02 00 00 05  00 03 28 45 F2 35
 01 00 03 00 00 06  00 07 03 60 29 2F
 01 00 03 00 00 07  00 07 03 60 83 7E
 01 00 03 00 00 08  00 07 03 60 E6 87
 01 00 04 00 00 09  00 12 12 47 47 9A
 01 00 04 00 00 10  00 12 12 47 EB BC
 01 00 04 00 00 11  00 12 12 47 41 ED
 01 00 A0 00 00 12  00 01 00 00 61 BA
 01 00 A0 00 00 13  00 01 00 00 CB EB

The Table of Contents of the CD has every entry listed 3 times in a row for redundancy. This means, if one entry is broken, it can be reconstructed. The last 2 bytes are a CRC to allow checking for validity.

Seeing these prints, I can also see why it sometimes worked. The values are stretched out and some entries are missing.
That cache is actually not helping with stalls. It's just for reading larger chunks from main storage.

[young grotto]

i kinda love the idea of CRC fixing an error like that

no possible way they could of forseen someone running a cd image on a fpga with a bug in it, but it still got the job done

[terse flare]

I'm also amazed that a totally broken TOC sometimes worked. I also have some hope that this small fix also repairs the random occurrence of broken music in The Apprentice, when CDDA is played. But I can't consistently replicate that to confirm.

[young grotto]

you can't somehow disable crc for testing purposes can you?

[terse flare]

I can, but how would that help? The CD driver of the CD-i will just throw everything away.

[young grotto]

i was thinking it would kind of give you "no guardrails" incase there's anything else subtly off getting corrected by crc

[terse flare]

Oh no, the CRC is not "correcting". This is no Forward Error Correction like Reed-Solomon. The CRC is simply there to validate 1 entry of 3. So if 1 of them has a valid CRC, it can be assumed that the entry is good. And even if the CRC itself is broken by comparing all 3 and 2 are equal, things might still turn out alright.

[young grotto]

ah, so no like, parity/healing, just detection

[solid quest]

Really great that you’re taking a look at the BT sleep, it’s a nasty experience!
I’ve been testing the nightly main with the screenshot changes and it’s a massive improvement, especially on the cd cores, many thanks

[candid adder]

Just seen this. I’ll test it out later today. Thank you so much for looking into it!

[candid adder]

Just tested with that file you uploaded and it worked! Volume is low, had to move my soundbar to 60, but it works! Didn’t try with the latest MiSTer main as I’m not sure if that fix was in the latest main or not.

Never mind about the volume. Much like the Jaguar, you have to raise the volume manually. I think those are the only two CD systems that do that.

[terse flare]

Thanks for testing. Before making a PR to mainline, I wanted at least one user to test that binary. You did, so I guess I can create the PR now.
Concerning volume, I'm not sure how to tackle that. A fresh/corrupted NvRAM starts with the lowest volume. I don't know what Philips was thinking back then. It might be possible do provide a default NvRAM with the volume maxed out, when none was found on the storage. But I'm not sure if this overcomplicates things. Also, it seems, the Volume setting is only affecting the CD player. Applications just overwrite that.

[stiff flare]

I wouldn’t worry about that, the likelihood of someone using the CDi core to play music CDs is incredibly small. This Discord may actually have the only person who would.

[candid adder]

Exactly. I figured that, especially with having to actually turn the volume up, that’s accurate to how it is. Just wished it save the settings upon power down but I don’t know if the legit hardware does that either. So I’d say for now it’s working.

[stiff flare]

I think saving the volume is a nice idea imo

@terse flare is that feasible to support?

[candid adder]

It’d be nice, but maybe a little too niche and Slamy’s overworked so if he can, great. But if not, it’s no biggie.

[eager pier]

But never zero, just like the odd of being attacked by a shark in your own home.

[terse flare]

@stiff flare @candid adder Hmm.. actually it is totally working right now even on the current core. I've dumped some of my childhood CDs, converted them to CHD and collected them in /media/fat/games/CD-i/AudioCDs. The save file /media/fat/saves/CD-i/AudioCDs.sav is now the NvRAM for all images in that folder. I set the volume once in the system shell and all of them are getting the same treatment.
This experiment was performed with CDi_cdmountreset.rbf, which is the latest core I've uploaded manually here. Maybe I should just push that to main, to cause another unstable build. Without that change, the CD swapping mechanism is slightly broken.

There is one behavior that is currently not managed by the core: When an Audio CD is ejected from a real CD-i 210/05 by pressing the Eject button on the front, the machine is not reset as it is performed when running an interactive application. Even that latest rbf is simply doing a core reset. This can be avoided by manually opening the tray using the button on the bottom right. This way one can listen to multiple CDs without a reset.
I recall, I've experimented with properly opening the tray once. But since applications usually totally hate that, I've abandoned the idea. Though it seems to be useful for Audio CDs.... but really not many will swap Audio CDs on the MiSTer... I guess. I think the PSX core also doesn't support that.

I've also noticed for the first time that the MiSTer OSD doesn't support UTF8 in the file names. So far all my images had only ASCII characters in them. That might be a problem for international users. But that is very nitpicky.

[terse flare]

Concerning "Volume up", this core is not ideal for audiophiles, since the lowest bit of the PCM samples is pretty much ignored. You can test this by setting Debug Options -> Disable Audio Att. to Yes. This will directly connect the CDIC DSP to the MiSTer framework without the connection to the AD7528 mixing circuit and your Audio CD will be louder. You won't be able to hear the DVC output and games that perform panning or Stereo to Mono mixing will hurt your ears in case you use headphones.

[young grotto]

psx lets you put a "noreset" file in a games folder that if present suppresses the reset on disc swap

this lets people play stuff like vib ribbon where the game expects you to open the lid and swap in audio cds

[eager pier]

Yeah, I remember talking about how that wasn't enough for a game long Monster Rancher and eventually the noreset behavior was changed so it persists until a core reboot rather than is only honored for that one time if the file is in the folder. Not sure which method would be best for CD-I

[candid adder]

Could it be because I have mine in a separate folder from the CDi one called AudioCD?

[terse flare]

Hmm... no... I've just tested that. I've created the folder /media/fat/games/AudioCDs and selecting a CHD inside also resulted into the usage of saves/CD-i/AudioCDs.sav. So this should be totally working.
Depending on your knowledge of Linux, I could offer you a MiSTer binary with debugging prints to check what is really happening. But until then, we have other options. When you close SETTINGS, ignoring whether you have changed anything at all, the system shell will write to NvRAM. The User LED on your MiSTer should light up, signaling a pending save write. When you open the OSD, it should say Saving.... Does this occur?

[candid adder]

I’ll have to check that

I don’t remember it doing so.

I barely know anything about Linux other than using a Steam Deck a while back

So that’s kinda hampered me wanting to get away from Windows 11

[candid adder]

Okay, so good news bad news. Good news is that it did save the settings upon power off. Bad news is that I think something changed in the MiSTer main used to fix the Jaguar stuff that broke audio CD loading. Because I’m getting the same issues as before.

This is the MiSTer file I was using. This was uploaded yesterday.

Not sure what broke in there but something had to. Even did a cold boot because I wanted to make sure the settings stuck upon power off

So one foot forward but one back, which is par for the course with beta files.

[terse flare]

The file you've uploaded here still has the broken TOC problem. I don't know who has uploaded that. I assume some person on the Jaguar side of things? Please don't mix these binaries up. The core devs have no control over the Linux side of the MiSTer. We need to perform a change request on Github in hope that Sorgelig or one of the Maintainers is accepting that change. That takes time. In the past, there was a solution for this that is hated by some here and now by me too. You can do this in the MiSTer.ini file to have a custom Linux binary for a core.

[CD-i]
main=MiSTer_CDi

But pleeeeaaase DON'T. When this core was in its infancy, it was encouraged to do that and users have forgotten they did it. So we constantly had to debug problems resulting from loading outdated files.

[remote rapids]

woah, new build!

[candid adder]

Do you know how I can fix the broken TOC with that disc? Especially with me planning to rip some more discs to use?

[terse flare]

There is a misunderstanding. Your disc image is fine and valid. There was just a bug in the MiSTer binary. The unstable one here #unstable-nightlies message is the officially new unstable build that also contains a fix for the broken TOC. Though it probably is not the one you maybe want for the Jaguar, if there are custom fixes in there.

Maybe it is not 100% clear. A core consists of 2 components. There is the FPGA side that is 100% in control of the core dev and is isolated for every core. But the Linux side is shared between all cores. And updating that can be a pain and takes time.

[gray badge]

It should be both jag and cdi now. The one little miss was using was just the jaguar stuff I think

[candid adder]

New MiSTer nightly fixed it. So audio CD is working on CD-i again. About to check the Jaguar stuff to make sure nothing’s broken there. Gonna be mainly off my feet for the next couple days as every gel shot I get in my knee, I have to limit my movement for the first 48 hours after.

[quasi viper]

The nightlies are always a bad case to test cores against.

You can use these only to test the nightlies themselves., not to test other stuff against them

[golden axle]

Basically yea, they're not cross compatible

[quasi viper]

It's bad habit in general.

[spring gate]

We are due a new unstable main soon as 3DO has a custom one for loading the kanji font needed for some games, so if there is an update you have done for CDi then hopefully it can be merged in fairly quickly when 3DO update is also merged

[terse flare]

No worry, it has been merged this morning. The changes of the custom MiSTer binary from #1103404843512631357 message are now part of mainline

[candid adder]

Jaguar’s working so far

[spring gate]

Good stuff. I wonder if 3DO was merged as well...

[wintry dust]

yes

[candid adder]

And the Jaguar’s broke again 🤦🏽‍♀️

The pain of being a tester, I can only imagine the pain of being a developer.

Swapped between CD and cart a few times and boom.

But yet I was able to “unstick” it by loading a CD game again, then a cart again. So it’s happening at random it seems. On the bright side, you can switch between cores and STILL have the Audio CD work on the CD-i core now. So there’s another plus

[gray badge]

Post your steps in the jaguar channel and I will see if I can reproduce

oh nm, I see it's there

[stiff flare]

With your changes now merged, which is the latest CDi core you’d like pinned?

[terse flare]

Please this one. #1103404843512631357 message They fit good together. I hope it is fine that the latest pinned core is not one from the unstable builds thread.

[still kraken]

Hi.
Does anyone know how I can make a VideoCD that I can play using the CD-i core?
I have tried almost everything and it just wont play. I see the PLAY CD-I icon in the menu but when I press it the screen flashes a few times for so to go back to the same menu again. I am using this CDi_cdmountreset.rbf core I found here.
I have also executed the update_all script and I see I have these 3 ROM files.
boot0.rom : 2969341396aa61e0143dc2351aaa6ef6
boot1.rom : 3d20cf7550f1b723158b42a1fd5bac62
boot2.rom : 9694c466f9b65c1990a81b7a6280546b

I assume that I am not able to create a valid VCD bin/cue file?

[terse flare]

For the Philips CD-i community this is like Groundhog Day. Every now and then a user is asking about VideoCD generation in a way to make it compatible with the White Book standard, so we come together to observe whether the DVC is sticking its head (or is it the butt?) out of the player to check for the feasibility of VideoCD generation.
Since it is already spring, most DVCs are already out, building their nest and looking for a mate. But not all hope is lost.
Some users have tried to make the process slightly easier by automating it.
For example this project https://github.com/mrchrisster/mister_cdi_vcd_creator/
At some point before that I've also tried to collect some info on this here: https://github.com/Slamy/MPEG1_Handbook/tree/main
Please keep in mind that you might enter a world of pain with this topic and there are multiple approaches to this. We also have one user on the Philips CD-i discord that tries to use the original VideoCD Toolkit back from the 90s.

[terse flare]

Also, please start with already existing VCDs to check if your setup is working. Do they play?

[stiff flare]

What if we used the CDi core to play CDi games?

[gray badge]

But I have all this multimedia

[eager pier]

Now that's just crazy talk, it's like using the Jaguar core for non CD+G things

[still kraken]

I don't have any VCD's any more. They all went to the bin many many years ago. But now (obviously) I want them back and play them on my 1084S PAL monitor. Somehow (I don't know why) I want to watch some old moves in 4:3 on a CRT screen.

The reason I thought it was possible to play VCD's using the CD-i core was that I found this YT link: "MiSTer FPGA Philips CDi RC3 VCD Test - Terminator 2"
And this directed me here: https://kinworx.com/video/yt/
It looks like the CD-i core is used (unless a real Philips CD-i drive is used) and that VCD's are playing just fine.

[still kraken]

The script "mister_cdi_vcd_creator" worked 🙂
Thank you

[terse flare]

Playing just fine... I wouldn't call it that. It is the best we have right now, would be more accurate. I'm also not 100% sure about the state of the mister_cdi_vcd_creator. At some point there were still issues with scanning through the movie. This is more or less caused by the CD-i bridge application. Usually, when authoring a disc you want to have chapters and the ability to fast forward.

[fervent urchin]

I did some updates to the script after some testing from @river idol . I think it can still have issues with buffer starvation but if it doesn't report anything during transcoding it should be pretty solid

I don't use seeking and right now it uses your custom bridge as a default. So not sure if FF would work...

[tranquil leaf]

Honnestly, the mrchrister script is far above everything we could have hoped in such short time... Thanks to both of you for making this happen.

[fair obsidian]

i just tested the core again with a movie,it plays it well.

think the movie quality is between vhs and a svcd

[still kraken]

Did you convert to VCD yourself or did you test with an original VCD?

[fair obsidian]

no,dl a vcd,but probably you just can convert a xvid to cdi vcd format

and most aren't preserved,searched enough

[restive jay]

Besides the well-known MPEG module incompatibility, are there any known examples of the CDi having hardware variance that actually impacts the experience of using the system?

[fair obsidian]

I never thought i will looking to vcd's in 2026 😄 ,now you realise what way we made with everything,its unbelieveble.

[terse flare]

Your question is important and valid. But I think, the majority of users here - me included -, don't have the experience to answer it. This is best asked in the CD-i community directly where you fill find the veterans that hoard 20 different players in their closet.
If there is one thing, that is broken with the Mono I based CD-i player (this core is based on it), that would be a design flaw, making it impossible to alter the volume of the MPEG audio via Ir Remote. Philips has fixed that with later revisions.

Ok, I've just glazed at my 3 players. The CD-i 450 is garbage. It has no RGB out and no front panel. And the DC adapter uses an RJ45 plug. You know? The one used for Ethernet. The one plug, where the nose can break off. I don't know why they did that...

[restive jay]

I did a bit of research, and the answer is sort of:

• Different amounts of RAM
• Different amounts of NVRAM
• Certain ports made available
• Some supported certain accessories like a keyboard and mouse

[tranquil leaf]

https://x.com/derekpascarella/status/2040756813553885195

VCD is the new hype thing

Derek Pascarella (ateam) (@DerekPascarella)

(1/4) It has risen...

After 25 years of being locked away behind some of the most insane software/hardware protection I've ever seen for a $25 unlicensed console accessory, SEGA Dreamcast's DreamMovie has been unlocked, and is available for all!

DOWNLOAD
https://t.co/bNMs0XzWC0

X · Apr 5, 2026, 11:41 AM

[fair obsidian]

i'm testing days with vcd on the cdi MISTer core,works nice,hope this also works nice,never heard of this to be honest

[rugged panther]

https://tenor.com/view/star-wars-revenge-of-sith-anakin-vader-darth-vader-gif-19644107

[fervent urchin]

Just wanted to mention thanks to the CD-i core my son is improving his math and German skills with Adi Spielerisch lernen software

[restive jay]

The CDi's library is horribly undocumented. Maybe there should be a group effort to list what everything actually is

[candid adder]

There’s a lot of gems. Maybe not so much in games but it wasn’t meant for that. It was meant to be multimedia. There’s plenty of movies, photo CD, interactive discs, especially training and promotion. That alone should be worthy of it being better documented.

[restive jay]

Yeah. I see the CDi as a vision into a world that you don't normally see. It's a vertical slice of the world's economy on the local scale

[lavish crow]

The biggest question mark to me is Urban Murderfiles #1. It’s supposedly a game that officially released but it’s undumped and no one has any screenshots of it or physical evidence it exists afaik

[candid adder]

It is a 90’s goldmine as many industries used it for interactive purposes. It pretty much replaced laserdiscs in those sectors.

Some of the first “video racks” I helped make had CD-i units for boardrooms along with DVD and VHS. Basically a way to be backwards compatible and forward compatible with various media that may need to be presented

[restive jay]

And it had to have been a killer format for Karaoke

[lavish crow]

Has anyone tried the Karaoke discs on MiSTer?

[candid adder]

If I wasn’t under limited mobility right now, I totally would

[restive jay]

I have tested CDI Karaoke 1

[golden axle]

I'm sure I at least loaded up a couple

[restive jay]

It would be neat if someone made a list of songs in each one. I'd be very curious to hear what the popular Karaoke songs were

[fervent urchin]

I tried some random German corporate training CD yesterday. A company wanted to hire a product design company for their new office furniture and it showed situations and how to react correctly.

It was pretty hilarious actually, mostly because of the 90s outfits and people smoking in offices

[tired beacon]

As a former Burger King employee from 30 years ago I was disappointed by that training CD.

Very boring. I was hoping it would have some terrible videos.

[gray badge]

A slamy patreon post?!

very cool stuff, Slamy. I thought you were taking a break, and yet, here you are dropping more cool shit

[terse flare]

I'm probably the worst Patreon creator out there. Users are already leaving. But I guess it is fine. No one is forced. 🥴 There wasn't an update in a while, so I can understand that.
Compared to developing FPGA stuff, this indeed was a break. For 2 weeks I did nothing at all. But then I decided to give it a try and it didn't take too long to build this. I'm slightly worried to even talk about this because some users might expect me to go full time on the CD-i core but I just wanted to do something different. Something refreshing.

[gray badge]

i'm here for your cd32 core. no pressure :p

[terse flare]

As you might know, I'm a full time embedded software engineer and no professional FPGA designer. This is why this is so much easier for me and not exhausting.

Since some people reading this might be confused now, here is the video. So you know what this is about.
https://youtu.be/vv5txHDCZd8

[young grotto]

if a core doesn't run bad apple is it even really a core? this is vital validation imo

i wouldn't read toooo much into what patreon users do, there's a few different ways people tend to use it

and there's a lot going on in the world that might make someone reconsider any luxury expenses

[maiden wigeon]

patreon is the onlyfans of FPGA developers, maybe add feet pics

[terse flare]

You have the absolute best profile picture for making that statement.

[lavish crow]

Best comment of the day

[young grotto]

i personally think of my patreon as supporting the creator, not the project neccesarily

[stray plank]

Awesome! This could have been a competition entry at the Revision demoparty last weekend.

[gray badge]

Exactly. As far as I’m concerned, we owe Slamy backpay

[terse flare]

Naaaaa, everything is alright. I'm not starving.

[terse flare]

I think the CD-i actually has no demo scene yet. If I knew, I would have submitted it. Maybe next year.

[stray plank]

That's what the wild competition is for.

I won the wild competition at the Outline demoparty two times with demos for my holographic display (Looking Glass Portrait)

Well, one was a game, a holographic version of Vectrex Mine Storm and the other was a real-time holographic recreation of the Amiga Juggler demo.

[valid onyx]

This is very cool @terse flare ! It also helped me fix an issue I've had for forever; on some cd-i content, on my 1600x1200 hdmi display, I kept seeing little squares in the background of large similar colored areas. Turned out this was because I had a Triniton filter on it, when switching to PVM CD-i just looked so much better 🙂 This great demo suddenly made me realize this might be the thing to change 🙂

[still kraken]

Rambo2 VCD playing using the CD-i core 🙂
Quality when displaying on the 1084S PAL monitor is just superb. I love it.
Thank you for the great work making this possible.

[terse flare]

I also love my 1084. Considering the age, it is still super bright. Unlike PC CRTs that are difficult to get in good condition. At some point I gave up in hope that TFTs or OLEDs will solve the perception blur issue in a helpful way.

[terse flare]

Also
User: Go Rambo VCD Brrrr! Peng Peng
Dev: Oh no! Hopefully he will not activate Slow Motion or pause and resume the VCD too often. Hopefully he will not pause at all. Hopefully the mysterious macro block issue that I've seen lately will not interfere with their viewing experience.

[tawny kestrel]

VCD = Lego Vision movies. The macroblocking is hillariously bad on some movies. But 1150 kbit/s wasn't good enough for any useful compression back then.

[unborn grove]

Which Rambo2 VCD is this? Which release, which country?

[unborn grove]

They say it was just as good as VHS back then but that’s simply not true It’s worse I have seen some decent VHS tapes on good players

[stray plank]

Where I was an intern, we were not impressed seeing CD-i 'full motion video' for the first time ever. They had encoded it with their Silicon Graphics Onyx computer. We didn't care for the macro blocks, how someone's head suddenly snapped into motion and the weird effects going on with the leaves of a bush in the background. We also preferred VHS at the time.

[still kraken]

It is a DVD that contains both 16:9 and 4:3 version. I ripped the DVD to a nkv container and then used the script to convert to VCD. (There is a script on git that creates a valid VCD compatible with the CD-i)

[candid adder]

What’s it called? Might think of taking a few things I have and making VCD discs out of them.

[gleaming nimbus]

the Update All CD-i core is from 20250626. i need more recent unstable for DVC, i take it?

[gray badge]

yeah, you need the unstable core. should be in the pins

[jolly badge]

Link please ^^

[gray badge]

not here

[jolly badge]

May I ask why not? Is there something illegal in that script or because it's not really CDi discussion even though the VCD playback is a great feature of the core. If I may not ask, just ignore this post.

[karmic bramble]

Movies also have copyrighted content.

[wintry dust]

I think they wanted the script, not the rip

[gray badge]

oh sorry, I thought you were asking for the movie. my mistake

[wintry dust]

mods, always screwing it up

[gray badge]

with great power comes poor reading comprehension

[jolly badge]

Naah. That would be stupid. Just the script ^^

[gray badge]

I should drink coffee before I start typing. But here I am, typing without my coffee

Friday funday

[wintry dust]

casual friday, hope you're all wearing jeans and your polo (tie optional)

[gray badge]

cargo shorts and crocs

[stiff osprey]

I always wear a tie with my polo shirt. It's casual friday, not slob friday

[gray badge]

you are such a company man

dress for the job you want

[terse flare]

I'm assuming it's this one https://github.com/mrchrisster/mister_cdi_vcd_creator/

[remote rapids]

https://x.com/dril/status/473706394009759744

wint (@dril)

blocked. blocked. blocked. youre all blocked. none of you are free of sin

X · Jun 3, 2014, 6:03 AM

[gray badge]

there is always a dril tweet

[nimble epoch]

It was a pleasant and nice surprise to see a Patreon post 🙂

[eager pier]

[odd trout]

I made Slamys Bad Apple demo my CD-i autoboot rom, plays everytime I load the core now

[deep roost]

I didn't even know that CD-i didn't have Bad Apple.

https://x.com/yoshinokentarou/status/2042722734946541861

吉野@連邦(renpou.com) (@yoshinokentarou)

幻のゲーム機「Philips CD-i」で、遂に「Bad Apple!!」がリリース!!
実は今日までCD-i版 Bad Apple!は存在していなかった。

また本作は、ありがちなMPEGの動画再生ではなく、RLE画像＋ADPCM音声のストリーミングで30fps再生、しかも768×280の高解像度対応と技術の塊だッッ...

▶ Play video

X · Apr 10, 2026, 9:53 PM

https://youtu.be/vv5txHDCZd8

Slamy continues to be based.

[terse flare]

Before starting, I've actually searched for this, in hope to find an existing implementation. Bad Apple existed for so long and many creative ports have been made. Though I guess, the CD-i is just too obscure. So the only users that could have made a port are the ones that already know the development process in and out. This Japanese post calls it a "marvel of technology". But it really isn't, I think, because the idea is derived from Burn:Cycle. I've just reinvented the wheel, because I knew the hardware could pull it off. The remaining question was "how?".

The most complex part was probably the video encoder, I've written for it. The video file was first converted to a collection of PNGs using FFMPEG and then a tool I've written in Rust, converted those PNGs into the native framebuffer format of the RL3 video mode. Then it writes a Green Book CD track while simulating the "consumption of frames" during playback to derive the buffer fullness during execution. After all, the CD is always the same data rate, so the video encoder had to insert empty sectors now and then to match the playback speed when actually running. That was really the most complex part during video encoding. On the other hand, the encoder needs to make sure that, during playback, there is never an underrun, so the analysis goes both ways.

On the CD-i itself I first went with the route of having framebuffers and copied the video data into them. It turns out, the CD-i is too slow for this, so I've consulted the experts on the CD-i discord and the final solution has no framebuffer at all. Instead the CD sectors are - as usual - written to system memory via DMA on their arrival and the video encoder was altered to always have atomic lines. Then, the DCA/LCT (SNES users know it as HDMA, Amiga users as Copper) was used to reposition the video pointer during raster scan. This way, the CPU is only busy with reading out the meta data to know when a new frame has arrived and converts this info into a display list.

The audio part is actually fully automated by the CD-i SDK. The ACUShell ADPCM encoder was used here to create the stream from an AIFF file and the playback on the CD-i doesn't consume any CPU power at all, because the DSP inside the CDIC does it all.
The biggest surprise for me was the high quality even with just 4 Bit samples.

[spring gate]

Well done making that Slamy, that is a real achievement, and useful to the wider CDi community. It is funny though that nobody did Bad Apple CDi before and even CD+G got it before CDi 🙂

[rough sable]

You are an amazing developer and many of us appreciate your hard efforts for such a niche system to preserve it. I had an original CD-i and it simply barely functions at this point.

[terse flare]

This might be the fate of all CD based machines. Maybe you can still fix it somehow in case you are motivated enough to investigate the cause of the issue. On the CD-i discord there are still purist users that frown upon emulation and are educated in fixing these machines. Philips also did its part with adding service menus to the players. But I'm not sure how they work. xD I probably would replace the drive with an ODE as soon as it exists.

[vernal crater]

quick question on the core hardware. im trying out a few apps and games. I loaded Dragons Lair as i heard its a good port and i got this error. Is there a setting i can change or do games that require hardware expansions not yet supported?

[wintry dust]

you need unstable/wip core, see channel pins

[vernal crater]

ah good to know. Ill check it out, thank you

[vernal crater]

worked immediately, didnt realize i was on an earlier version. Thank you

[kindred cosmos]

Will this work under the Linux subsystem in Windows 10/11?

[terse flare]

Öööh... isn't that just a Linux machine with its own Kernel running embedded in Windows? I would totally say, yes. I would even assume it would be working on MSYS2, since no Posix specific features are required. Bash is running on Windows too after all. But yeah, WSL will work.

[kindred cosmos]

Sweet, thanks. I just don't have access to a Linux machine at the moment.

[stiff flare]

If the CD-i was so good then where’s the DVD-i?

[gray badge]

They skipped all the way to DVD-R

[pulsar moth]

It was called Nuon.

[tranquil leaf]

Yes

[eager pier]

What about MO-I?

[iron blaze]

DivX-i

[candid adder]

Technically, the Nuon was Jaguar 2

It had VLM and sequels to Tempest and Iron Soldier

[unborn grove]

Really???

[stiff flare]

And also a spiritual sequel to Atari Karts

sort of

[eager pier]

To build a Nuon emulator thjough, you must first build a DVD and CD player

[deep roost]

But, there already exists a Nuon emulator called "Nuance".

[quartz crest]

Technically mame as well

Wow, sad

Cool that people picked it up and continued. Awesome.

[restive jay]

Fun CD trivia:
The Redbook CD format doesn't support mono

[gray badge]

Dream Odyssey fans in shambles

[strange crater]

I did see someone once put a 2-hour classical performance on a single CD by just putting half the album on the left channel & the other half on the right one

[restive jay]

Yup. And they'd bundle a switcher with the CD releases

Really weird to think that CDs were ever expensive enough that it would be easier to bundle a switcher with every CD than do 2 CDs

[strange crater]

It took them a while to figure out how to get the glue layers right. From what I understand, early on there were a lot of rejects, which drove up the cost (that and new expenses related to digital encoding/mastering/etc.)

Once they figured all that out reliably by the early 1990s, CDs quickly dropped dramatically in price for manufacturing.

[valid cairn]

I'm worried
About the sheer density of 90s content packed into such a tiny space
Which is enabled by this core on the Mister

It could create a tear in the very fabric of time and space

[strange crater]

Agreed. We should stockpile JNCOs & Beanie Babies just to be safe

[valid cairn]

Played some cd-i Tetris last night, came to a conclusion. Everybody on earth just needs to cool out a bit.

https://tenor.com/view/spin-world-globe-mapping-details-gif-3270140137824791678

[stiff flare]

very true

being mad is easier than being happy

[iron blaze]

what's that supposed to mean jerkhole

[stiff flare]

what’s a jerk hole

like where you put your beef jerky?

[iron blaze]

i don't even know. i assume beavis and butt-head said it once

Beavis & Butt-Head are great because you like them when you're stupid, and then when you're smart, you like them even more.

[kindred cosmos]

Lets all chill out to some Star Trek VCDs.

[gray badge]

https://tenor.com/uhlnXniOIIf.gif

[valid onyx]

I just realized in English 'mad' can both mean 'angry' and 'crazy'. That's mad.

I mean, I knew it, but I never realized it.

[west pagoda]

pretty sure it can also mean very

as in “that’s mad crazy, yo!”

[valid onyx]

https://tenor.com/pTJ771i27Ue.gif

[west pagoda]

so, translating, “being very is easier than being happy”

[valid onyx]

makes total sense.

[west pagoda]

I guess this is another situation where I just can’t understand what @stiff flare is talking about

[valid onyx]

which explains why he is a mod and we aren't. He's a genius, and we can only be happy to be in His Presence

[west pagoda]

oh, I thought he was a mod because he was the preeminent train scholar on the server

[valid onyx]

That probably helped too

[eager pier]

Well all I have to say about that is "Buffalo buffalo Buffalo buffalo buffalo buffalo Buffalo buffalo"

[stiff flare]

Same tbh

[strange crater]

Buffalo’s buffalo can’t buffalo Buffalo buffalo Buffalo buffalo buffalo, because Buffalo lacks Buffalo buffalo Buffalo buffalo are capable of buffaloing. Trust me on that, as a Buffalonian.

[strange zephyr]

You try being happy when affected by your toddler in command, while also being thousands of miles away.
But this is a bit spicy, so I'll stop for now.

[valid onyx]

So, what'yall playing on cd-i?

[gray badge]

I still need to finish burn cycle. make good on my childhood

[stiff flare]

That’s a cool game and the video is really high quality and smooth

No major mpeg artifacting

[glacial herald]

Name That Tune with Bob Goen to fill out my lexicon of obscure 1-hit Wonders

[valid onyx]

It's not using MPEG, it is using the same 🪄 that slamy's bad apple uses

I myself want to give Chaos Control a go; remember it from.......well, the fact that I remember it at all should betray my age, let's not make it even more compromising. Wondering what it'll look like now 🙂

[little light]

for the DVC-flavoured core, I need the linked unstable MiSTer main, but would i rename it to MiSTer_CDi and enter this line in the Mister.ini?

[CD-i]
main=MiSTer_CDi

[spring gate]

No, DVC was merged in merged in months ago and no unstable Main needed for any core (unless I have missed something)

[little light]

okay, I tested it all out, just to be sure:

update_all core with update_all Mister_main = no DVC
update_all core with unstable Mister_main = no DVC
pinned core with update_all Mister_Main = DVC!
pinned core with unstable Mister_Main = DVC!

I don't see any flashing indicators when I play Dragon's Lair, so I've enjoyed the last 10 minutes of the knight dying immediately after the bridge, lol

[golden axle]

Try brain dead 13 next

[stiff flare]

Yeah my bad, I don’t know what video compression method it does so I said mpeg lol.

[strange crater]

Mostly Zenith - still trying to beat it. Occasionally also Pac-Panic, Micro Machines, The Apprentice, and Master Labryinth.

[valid onyx]

Oh, puzzle wise, I also love mega maze every now and again

[valid cairn]

I adore cd-i tetris. The last time I played it, before slamy's wonder core, was on a pre-release dev kit in the penthouse of a skyscraper, surrounded by supermodels and coked up executives.

(Disclaimer: this comment is made up)

[valid onyx]

I wanted it to be true, though. Dream shattered

[valid onyx]

I just played it for the first time, and I have to say: years of GB tetris has engrained that control scheme in my mind so hard, that I KEEP DROPPING BLOCKS when I want to rotate them the other way. So, whilst I like the music-of-the-era and the backdrops, I find it very hard to play it well

I returned to a game of Tetris (Rosy Retrospection) on the GB, and I fared much better there. 🙂

[valid cairn]

You're right of course. It needs to be modded or patched with modern controls like rosy retrospection.

[terse flare]

Are you playing with a mouse. It makes the game much easier to control, especially in the vertical direction. Horizontally it is weird. It feels like the horizontal position is managed on a pixel basis internally but quantized to the tilemap when used for the gameplay.

[terse flare]

This is one of the many issues, the CD-i version has. Dragon's Lair really needs a fan based version to fix this.
Considering the many versions of the core, it is true that DVC support is merged into Main for quite some time now, but the last official stable release for download by update_all is from June 2025. I don't know when the right time for a new stable release would be. There are too many known problems with the current VMPEG approach to call it stable.

On the other hand I'm against the usage of the current release from update_all, because it is lacking some smaller fixes. In general one cannot go wrong with the core pinned in this channel as long as expectations for VMPEG accuracy are not high.

[nocturne vortex]

That’s the setup listed under system requirements

[stray plank]

I could tell it wasn't real because the supermodels weren't coked up as well

[sterile cedar]

Are there any versions of dragon's lair that have flashing indicators? I think they were all the same on 5th gen consoles. Without them it is just a guess your input or die game. Adding the indicators would make the game end like in 15 minutes though.

[spring gate]

There is no rule, or really right or wrong way to do releases, but if there is no downside to being on the latest version then I would say it would be better for users if the one in Main was updated. You don't need every bug fixed to update in Main, otherwise we would have less cores than I have fingers. 🙂

-one caveat being it isn't good to have a core in Main that requires an unstable Main to run. Fine if it is needed for some features, but not to do things the previous version does fine on the current stable

Although I don't think this is the case currently with the CDi core

[eager pier]

So like Reese's?

[strange crater]

I’ve been using a trackpad. I really want to try this one with a trackball, though - I believe the game was made with this in mind:

[terse flare]

I think you are right. Since the pinned core CDi_cdmountreset.rbf didn't introduce any troubles (I guess) and I've counted at least 2 users that accidentally have used the update_all core in expectation that everything is there, we should declare a new stable release. The latest pinned core should have no negative down sides to base case CD-i software. It does however still have a regression since unstable 260222 that causes Addams Family to fail when opening the controls on Disc 2. I will extract the fix from my current messy work in progress and add it to the latest pinned state. It is not the best DVC core maybe but it should at least play everything. And who wants to have a core without "MAMUSHKA!!" ?

[terse flare]

The original arcade should have, doesn't it?

[quasi viper]

o.. I have that thing. never figured out what device it goes with. Of course. CD-i !

Anyone wants to buy a CD-i trackball?

[strange crater]

I mean, if you're offering...

[quasi viper]

I'll try and dig it up this weekend

[strange crater]

The original arcade version of Dragon's Lair actually doesn't do any video mixing at all. The score handling is entirely done with a set of 7-segment displays above the video monitor, and all the program outputs on-screen is the LaserDisc player output.

There are audio indicators when you make a valid or invalid move, though (which don't exist on many ports)

(This latter point is why I find a lot of the ports quite difficult to play, especially since the move timing on arcade is completely different to CD-i & 3DO. I've memorized the arcade version to the point I can beat it 1cc, and I don't think I've gotten past the second screen on many of these ports.)

[river idol]

The CD-i (and Sega CD and 3DO) match the arcade in that there are a limited number of flashes baked into the animation to indicate player moves. Some laserdisc games of the era would literally tell you the exact move if you failed (Cliffhanger would put up text like "You should have gone left") but Dragon's Lair (until the DVD era) didn't have overlays telling you explicitly every move.
One thing to understand is Dragon's Lair was the first game of this kind. Also, the QTE (as it would be known as much later) style wasn't its original design. The original version was a turn-based RPG where you could explore the castle somewhat freely. The change to quick reactions was after feedback from a trade show to arcade operators who thought it was too aimless and they also wanted a game people couldn't just wander from room-to-room on a single quarter. Bluth and Dyer didn't have time to reanimate the whole game, so the hint of flashes were added to the scenes where they felt what the player should do was the least obvious. The first sequel, Space Ace, has many more flashes and also more branching options. The power-up mechanic let's the player pick an easy route or a "Space Ace" difficulty path on a scene- by-scene basis. If they're aren't any flashes for a move, it's because they thought it was "obvious" (e.g. Dexter is surrounded in three directions by hazards and only one direction is clear, go that way). When you get to Dragon's Lair 2, almost every move has a flash indicator because they realized people want to see as much of the animation as possible, but to compensate they also upped the number of moves significantly. Remember, these were still arcade games that they want to churn an average play through in a few minutes at most.

An interesting bit of history, Rick Dyer didn't give up on the idea of a turn-based RPG laserdisc game. The game Thayer's Quest (made without Don Bluth) was basically the original Dragon's Lair design and it flopped in the arcades. Dyer also tried to use it to launch his Halcyon home laserdisc game console but that didn't work out. So 10 years later, he took the animation from Thayer's Quest and re-dubbed it and changed the plot and released it on CD-i (and 3DO and PC) as Kingdom: The Far Reaches. That did well enough to have a sequel, Shadoan.

[eager pier]

Huh, interesting. I remember seeing Thayer's Quest demoed with the Halycon. Didn't know it was re-released as a different game

[terse flare]

Well that was certainly faster than I've expected. It might be true that a new unstable build is available. But you don't need to download that. It is very similar to the previously pinned core + a fix for Addams Family Disc 2. More important is that by this minute CDi_20260502.rbf is now part of update_all. Here the changelog when compared against the previously stable release 20250626:

Release 20260502 (changes since 20250626)
- Unstable DVC now integrated into a stable release
  - DVC implementation equals CDi_cdmountreset.rbf + GOP fix for Addams Family Disc 2
- CDIC: Added CD sector cache
  - Catches problems caused by short HPS stalls
- CDIC: Fixed reading offset of ADPCM sound parameters
  - Fixes menu sound effect of "Golf Tips"
- MCD212: Fixed switch from CLUT to DYUV without VSR reload
  - Fixes graphical corruption during fade transition in "Lost Eden"
- MCD212: Fixes OFF coding
  - Fixes graphical corruption in "David and Goliath"
- SERVO: Fixed disc type according to actual hardware (210/05)
  - Required for the vcd module to even attempt detecting a VCD medium
- MCD212: Increased ROM bus bandwidth
  - Was slightly slower than real hardware. Now slightly faster than real hardware.
  - Fixes black flicker during the intro of "The Ultimate Noah’s Ark", when running in 60 Hz mode

While an unstable Main is not required, I highly recommend one equal or newer than 20260330 for the TOC fixes, which are required for Red Book Audio to work properly.

[gray badge]

will you be pushing an official build as well?

nevermind

I see you did 🙂

[rugged panther]

Holy shit this explains why I hated it so much

absolutely bonkers you'd put in a $1 bill the screen would flash and you'd die

Literally worst arcade game I'd ever played

[golden axle]

Brain dead 13doesn't have any indicators but it at least makes a sound for when you press an input and another for when its the correct input but after a while it stops making sounds all together, not sure if that's accurate or an issue with the core. If you save, quit and reload at least the sounds come back

[restive jay]

Imagine if Retroachievements started supporting the CD-i

"Watch the entirety of Flowers of Robert Mapplethorp"

[radiant minnow]

@terse flare I am VERY glad you're back to this! The core is almost there... and this core gives me IMMENSE amounts of fun!

[terse flare]

This was really just grabbing fruits on the ground. It doesn't mean I'm able to fix the remaining bugs. I haven't even really coded, to create this, since it is a mix of the changes I had around that I certainly know of, not doing anything worse. Please no pressure.

[radiant minnow]

Ah no no, of course! ZERO pressure! The core is in a very good state, that's what I was trying to say! 😄

In fact it's WAY better than a real CD-i in many ways

(Controls, battery, display, disc access, and others!)

[river idol]

That said the arcade version is way easier and more forgiving than the CD-i (and 3DO) versions for a few reasons. 1) In the arcade you get an audible cue for when input is allowed and accepted. Non-accepted moves give a buzz sound, so you know it wasn't yet time to input a move. Registered moves give a positive "boop" sound so you know a move was recorded, and more importantly you can get ready to enter the next move. 2) The arcade game skipped the tricky drawbridge intro which requires multiple moves with no flashes to hint the player. The arcade version instead went to the room with the wall trap that requires a only single move and a helpful flash. The drawbridge scene was discouraging players to try again during play-testing, so they skipped it before the wide release. 3) The home versions changed many moves, added moves, and narrowed the timing windows for...REASONS. Add in the lack of audio feedback for when it's okay to move and which moves counted and you have an unplayable mess compared to the arcade (which I have finished many time on a single credit). It makes me wonder if the company doing the port merely got the video and a description of how the game played rather than access to the arcade machine let alone the source code. If you want an authenticate Dragon's Lair (and its brethren) experience, check out the Daphne emulator. With that you can even play the prototype version which allowed more exploration of the castle and problem solving before they switched to the pure QTE gameplay.

[rugged panther]

I never, EVER, ever want to play that stupid game again lol

[dapper kettle]

How is the switch port of dragon's lair? More like the arcade or the CDI/3do version ?

[gray badge]

ignore me. need to do some more looking

[quasi viper]

noted

[valid onyx]

Super interesting info! And I just watched ben heck playing around with one for 40 mins 🙂 Will play Kingdom now at some point, too!

[chilly flax]

At least space ace on CD-i got the flashing and audio cues

[strange crater]

Kingdom, itself, is worth going into in a bit more depth - it’s sort of a complete reimagining of Thayer’s Quest. It uses all the original animation created for the RDI project, alongside a newly animated intro, a new ending, and two new additional areas.

However, every animation in the game has been redubbed and recontextualized, changing most of the game’s puzzles to give them different solutions. Dyer also changed most character and item names, and added more hints throughout to generally give the player more clear directions on where to go next. (There’s also an in-game map system, which the original game lacked - you’d actually have to navigate from a tiny map on a pamphlet they had in a holder on the side of the machine.)

As such, while I do think Kingdom is the definitive version, I’d also recommend trying the direct remaster of Thayer’s Quest done by Digital Leisure as an Interactive DVD. This will give you something a lot closer to the 1984 arcade experience, with a lot of Kingdom’s quality-of-life features.

I will add that I wouldn’t recommend emulating the arcade version of this one on DAPHNE - while it does work, the arcade version relies on an absurdly complex bespoke control panel, and as such it’s just not worth rebinding to a keyboard.

[rugged panther]

Jokes on you I won't play any version!

[valid onyx]

Thx!! So cool to know!

[eager pier]

Huh, the sequel was also released on CD-I, but only in Europe, but not on the 3DO, only the orioginal was on that

[strange crater]

Quite a lot of CD-i software only released in Europe - same goes for Myst, Micro Machines, Pac-Panic, and Lost Eden, among others.

[eager pier]

Yeah but the first game did release the us, and the sequel released in the US on DOS

[strange crater]

Kingdom II got an Interactive DVD release in the U.S., as well.

It’s unconfirmed if a DVD version of Kingdom: The Far Reaches was ever available. Instead, most people got the DVD release of Thayer’s Quest, which used the original Halcyon/arcade audio tracks & sequencing but kept most of the quality of life features from Kingdom.

[radiant minnow]

I always loved Kingdom / Thayer's because of the aesthetics: it looks SO much like a typical 80's animated series one would watch on TV!
I don't know why exactly: the colors, the animation, character faces...

[terse flare]

It is CD-Friday, so I guess an update might be interesting. As you might know, the DVC still has certain instabilities. Sometimes Audio is missing after resuming a VCD. And the Slow Motion function causes a permanent audio video desync until the MPEG decoder is fully restarted by fast forwarding or a chapter change. I wanted to find the root cause of both issues. So I've started with Slow Motion.
Whenever a "audio & video synchronized MPEG playback" occurs (keep in mind that the DVC does decode audio and video separately and the ROM driver software needs to sync them) and slow motion mode is entered, the audio is stopped and the video continues with a slow playback. But the CD keeps spinning, resulting into a build up of MPEG video data in the base case system memory. The movie player application is responsible for pausing disc reading to avoid this. When slow motion is changed again to normal speed, audio is restarted at the current position that is freshly read from disc. So audio is in sync. But video has multiple frames of superfluous material to discard. A CD-i seems to be not capable of just throwing away MPEG data and that makes sense, because a video frame can depend on the data before that. After all, that is how MPEG compression works.
So, a real VMPEG based DVC seems to be going into panic mode and decoding as fast as it can. It almost looks like a soap opera effect.
The issue is, that I still don't understand how the real hardware is instructed to do this. I'm unable to find it in the driver code. I've asked CD-i Fan for guidance and he also doesn't know.
According to the Technical Summary of the MCD251, there is some hardware circuit mentioned and I have to assume now that this is how it is done. If you don't know already, the MPEG Audio is the time keeper of the whole DVC software side and the system clock reference is frequently sent to the video side to keep it in sync. So far, this core hasn't made use of that.

If all goes well, starting with the next release, my implementation of VMPEG will detect that case and goes into overdrive as well to fix the Slow Motion audio video desync. Paired to that, the next release will also finally use this info to slightly nudge the frame rate to fit the MPEG timestamp. So far, only the frame rate of the MPEG stream was utilized for this and I think that might be an issue. I've tried to find a place in the driver code where the software does speed up or slow down the video playback but I was unable to. Therefore I assume that MCD251 is also responsible for this.

The next issue is the pausing and resuming. Sometimes after resuming a video playback from VCD, the audio is missing. And I think I do know why since this morning.
This is an analysis of the MPEG Audio input stream FIFO fullness attached to the input of the MP2 decoder. It also lists the occurence of FMA and FMV data as small spikes. It is noticable that Audio data is much less frequent than video data, because the bit rate is much lower.
The FIFO quickly ramping up when sectors are DMAed in and slowly consumed because the decoder is reading from it. When performing a though experiment, one might notice that an underflow can be reproduced by starting at the wrong time.

This is an error case after resuming from a pause. There is a clear distance between FMA data insert and the start of the playback. Still an underflow occurs. I'm not yet sure what this means. The next FMA data occurs at quite a distance in the future, so I would have expected that the playback also would have started later to avoid underflows. There is a fast fix for this by just allowing the decoder to simply continue after such an event. But that would cause a small plop in the audio output.

We hadn't a story like this for quite some time and I had more energy this week and a little bit more time. So thought it made sense to give you an update that something is happening. Still not sure whether this fixes all remaining issues bit I can at least try.

[valid cairn]

Absolutely awesome!!!

[gray badge]

Slamy just casually conquering MPEG decoding. Nice work, man

[terse flare]

This is not feeling casually for me. I had a table flip moment with the DVC a few months ago. That's why I've stopped in the first place and continued with Factorio again. But now I've entered Gleba and thought well, VMPEG is not as awful as the Pentapods.

[gray badge]

Well you make it look easy, duder

[quasi viper]

are you developing for Factorio?

[terse flare]

Noooooo! Woah! I was playing Factorio. It is incredibly addictive and dangerous.

[quasi viper]

o. hahaha 🙂

well. could be!

Never could get past the demo, it seemed like work to me

[terse flare]

That's because it is. It's like PCB design but as a game. I find it incredibly relaxing and could recommend it to anyone that is mentally stable enough to stop playing it.

[quasi viper]

I am mentally unstable enough to never startnplaying it I suppose 🙂

[eager pier]

Never played Factorio, I did play TIS-100 though

[quartz crest]

this is like my wife. she is an engineer, engineering processes, and then she plays a process engineering game

she plays factorio and others like it too

[young grotto]

about pops, something interesting we found out about n64 is that the real hardware actually has a pop suppresor

[stray plank]

Same. TIS-100 is great, especially if you have experience in programming assembly code

[young grotto]

factorio is an unparllelled timehole, i think i could play it forever

[eager pier]

I don't, but it was still fun to play, I didn't get far though. I need to print out that manual, kept forgetting what does what.

[stray plank]

It was fun to compete with my programmer colleagues for a bit at the time. I also didn't get far, but mostly because I have way too many different interests at the same time. I can imagine forgetting the details, I have that with lots of games that I haven't played in a while.

[upbeat egret]

How

[stiff flare]

I don’t know

[gray badge]

some help you are

[terse flare]

A small update: I've finally decided to get some physical CD+G mediums from eBay, since there is not really a chance to burn one from image (or is there?). Not really my choice of music. However, 6€ for both discs is fine. They were brand new and still shrink-wrapped since 2005. I plan to use these to analyse how a real CDIC provides the subchannel data. The core actually seems to have a bug somewhere in that regard. Sometimes, the CD player is forced into CD+G mode. I'm not sure why. But I think when correctly implementing CD+G support, that problem should go away too.
I'm not yet sure if the DVD drive on my PC is capable of extracting the subchannel information as sbi file, but I guess I will have to figure it out. It would be nice to have an image of these exact CDs to compare against the real reference.

[gray badge]

Oh man, Moondandy stocks at an all time high

[terse flare]

I said CD+G. Not PhotoCD.

[gray badge]

lol

[terse flare]

Jesus... you were right. https://discord.com/channels/647909397477195803/1474951090771460138

I don't know how long it will take to implement this. This is again new territory which requires analysis at first, some extensions to MiSTer main to utilize the sbi file (I assume sbi is fine?) and then the required changes to the core.

The DVC stuff in the oven is not looking too bad but I'm still unsure about the Addams Family glitch... So, digging into CD+G is a nice distraction with a possible sense of achievement.

[young grotto]

psx core loads subchannel data might be worth looking at

[tawny haven]

I recently got cdg working on the jag core. I ripped a karaoke cdg with a blu ray drive without issue.

I have some scripts from translating a few different formats of the sub channel data if you need them (.sub, .cdi, raw both interleaved and deinterleaved)

There are also things like this available

[west pagoda]

Curse you for getting that song stuck in my head

[tawny haven]

I wanna know...what you're thinking.

[west pagoda]

Yeah, that and “PURE ENERGY!”

[bold dragon]

Tell me what's on your mind

[terse flare]

This evening I've implemented a first experimental version of CD+G support based on "CloneCD subchannel images" (.sub) to have the subchannels R..W together with the already existing subchannel Q that one usually derives from the CUE file. Might be copyright problematic to post this, but I guess, it's fine. This is ripped from a German Karaoke CD I had recently bought from eBay.
I'm not yet sure about synchronisation between Audio and Subchannel data. The CUE file derived Q is matching the Q channel from the subchannel image but of course that doesn't mean that the raw sector data is also matching.

[bold dragon]

I don't know of anything where the synchronization has to be perfect. Just close

Or rather, I've never seen synchronization bad enough to be noticeable in any way

[tawny haven]

Box successfully checked. Did it lead to any insight on the game that was accidentally switching to cdg display?

[valid cairn]

The Flowers and MiSTers of Robert Mapplebutt...

[terse flare]

I think I've not said that it was a game that was forced into CD+G. It was a random occurence of having a normal Audio CD switching to CD+G that left me confused. The CD-i software does that as soon as Subchannel RW data is detected when playing a track for the first time after inserting the disc. And, yes, I now know why. When requesting a read operation from the CDIC in data mode, those CD sectors are written to the shared 8kB SRAM into one of 2 data buffers. The CPU can DMA these to system memory after completion. The subcode Q data is presented as 12 words, following those data buffers in deinterleaved and already CRC checked form. When reading and playing CDDA, the data buffers are abused for storing subchannel data from RW in interleaved form. The format is quite weird because it is not stored as bytes. 96 words of 16 bit with the upper 10 bits always 1 and the lower 6 bits the actual data one can deinterleave using software. I assume that residual data from previous read operations, which is not reset to 0 of 0xff by my CDIC implementation, causes a wrong interpretation of the CD player that this must be valid subcode.
TLDR: Shared buffers. I'm typing too much when at the keykoard.

This also means that a CDIC is not capable of reading subchannel RW in normal Green Book operation. At least I wouldn't know how that would be possible. With this chip there might still be features around that I haven't yet implemented, because they were never used by a commercial CD-i application. So far I've mostly implemented features when I or a user has noticed an issue. Like that one time when someone has noticed that both Christmas games were not working and it turned out that the disc needs to spin down once for whatever reason.

[tawny haven]

Glad you figured it out

[restive jay]

It's fascinating to me how little documentation there is on CDs in general

Or at least the edge cases of CDs

[bold dragon]

It exists, but it was locked behind paid technical tomes that were under NDA. So really you're just saying that you're surprised that people in the 1980s didn't breach their contracts when their now-mature pensions were at stake, once the internet was sharing this kind of stuff (20+ years later)...

[radiant minnow]

Pensions depending on private investment is such a vomit-inducing idea.
I hope we don't get that in my country anytime soon.

[bold dragon]

Not sure what your country is, but most of the development of CDs was in the Netherlands (Philips) and Japan (Sony). The people who signed up for the documentation were also in large companies, from various countries around the world - but ones with quite a bit of money. Of course, at the time, all parties would have attacked a leaker, if one was found (and all had sufficient money to make life painful).

But people today may have difficulty envisioning the wolrd how it was.

So clearly you would not know that the "fixed-benefit" pensions of the past - which are in some cases, still paying out - were definitiely driven by companies. Only Old Age Security benefits (enough for food) were government.

Even in Europe, this is before the union

[terse flare]

With #test-builds message you will find a test core with a fitting MiSTer Main to support Subchannel RW data and by that extent CD+G. This build does not support .cdi or .cdg. Only .cue/.bin/.sub and .chd files are supported for now. The reason for that is because those were the image formats that worked before and I hope this is sufficient for a first test.
The MiSTer Main uploaded there is the latest unstable with the CD-i changes in question. So you could replace your current MiSTer Main with that if you want to.

The new MiSTer Main is compatible with older cores by sheer luck. On pre-Sub-RW cores, the CDIC will continue to write to the SRAM and miss the next buffer by a margin of 6 bytes. So, the data is still written but at a position where it should not hurt since those CDIC RAM adresses are not used by the driver.
The old MiSTer Main is compatible with the new core, because the CDIC will just data starve without being angry.
During the design phase of the CD sector cache, I made the well thought move of making it adaptive. The first delivered sector from Main calibrates the data size. It probably wastes some cells in the FPGA but makes this compatibility possible in the first place.

[stiff flare]

What games can I play on CD+G

[golden axle]

I thought was like some photo cd format or something

[wintry dust]

it's so moondandy can put in his information society CD into yet another core

[nocturne vortex]

CD+G has just as many playable games as 3DO

[spring gate]

Unironically I have had that Information Society album for about twenty years, but alas on vinyl and not on the superior CD+G format.

I actually forgot I had it until the other day

[terse flare]

You could get creative and build your own Mix-CD+Gs with fabulous custom graphics.

Something tells me that there is only one user on the northern hemisphere that will actually use the CD-i core to play CD+G.

[restive jay]

Now we just need an easier way to dump them (from what I heard, you can't dump using any kinda modern PC)

[tawny haven]

Nah. It is easy. At least from linux.

[restive jay]

How so? I thought modern cables couldn't even read it

[tawny haven]

I bought a cdg karaoke cd off ebay, put it in my external usb blu-ray drive and did this. #1474951090771460138 message It ripped it first try and I converted it to cdi to use on the Jaguar core and it works.

I would suspect most cd/dvd/blu-ray drives would work

[wintry dust]

it's not a matter of 'can't rip it' it's 'can't rip it correctly'. you'll get something that works but may or may not be correct. but this is conflating redump levels of data paranoia with "can't be done"

most normal drives are garbage for the paranoia level most dumping groups desire, but you as a random user probably don't care

[tawny haven]

Correct that there is no error detection/correction. It will play exactly the same as if played from a cdg supported device though, as it would have the same chance of errors occuring.

[wintry dust]

at least once you rip it your errors are static!

[tawny haven]

In my case I actually thought I might have a few errors in my rip as I was seeing glitches. But it turns out there is a bug in the official cd bios of the jaguar for processing one of the cdg commands. (We built our own version that fixes it)

(We also found what appears to be an actual read error in one bit of the copy of the Information Society cd found on the internet. Whether it is a read error or a mastering error would need to be checked. It affects the word grandmother if anyone wants to check) #1055574003810578503 message #1055574003810578503 message

[unkempt dragon]

@fervent urchin is there a SAM specific thread in here to ask questions? I just got it all set up to play commercials but I can't figure out why the videos only play in black and white on my CRT over composite cables (all games run and display normally). This is on a SuperStation One if that matters

[fervent urchin]

I suggest you try the test branch to get this fixed. In test branch you can use the CDi core instead of the hacky m player way we use in main

Besides the forum there is no specific SAM channel

https://misterfpga.org/viewtopic.php?p=104090#p104090

Here are the instructions to change to test

[unkempt dragon]

i don't see "unstable cores" in update_all. Am I missing it or is there a separate script I need?

[fervent urchin]

It should actually work with the stable CDi core since slamy has pushed it to main a couple weeks ago

Let me update this, no need for unstable cores

[somber kettle]

It's probably the rip. The same track on the Rock Paintings CD which came with the Sega CD doesn't have the glitch.

[unkempt dragon]

hmm, now it seems like it's loading the CDi core but just not playing anything, then a minute or two later loads up the game in a different core. Will have to look into it more tomorrow

[fervent urchin]

have you tried starting cdi core manually?

did it download the unstable release?

[unkempt dragon]

ha! I have no CDi core 😂 I was also messing with downloader filters tonight and apparently removed it before realizing SAM needed it lol

aaaand it's still not working 🤔 i was able to load the CDi core fine on it's own. Maybe i just need to update/configure SAM again since the core was missing before

[fervent urchin]

just updated test branch

it should now jusy use the regular cdi core

[fervent urchin]

Any luck?

[unkempt dragon]

Ran update from the SAM script and still wasn't working. I should be asleep though so I'll do more troubleshooting tomorrow lol

When SAM started it just went black and I couldn't open the OSD or close out of SAM at all

[fervent urchin]

Sounds like it has issues loading CDi. A reboot might help as well

[unkempt dragon]

The videos are in a new format right? Where does it save them to so I can check that they downloaded

[unkempt dragon]

i rebooted and checked all settings as well as changed to stream the videos instead of downloading them locally, and videos are playing now. But now it seems to have trouble finding my disc based games possibly because they are contained in sub folders? is this a known issue?

when it plays video for nintendo consoles it is usually able to load the appropriate game without issue

[fervent urchin]

Downloading them locally is always happening in the background, it's just a question if you want to save them when downloaded

You might need to reset game lists to get your disc based games working

You can do that in the menu under reset

Subfolder structure should be fine

I suggest just letting it run for a bit without interrupting so it can build the game lists

[unkempt dragon]

i generated the lists from reset menu and now its doing the same "stuck on black screen" as before. idk what's going on 🙃

[fervent urchin]

Sounds like you are missing the bios for CDi maybe?

Not sure why it would have worked before though

[stray plank]

The black and white signal could be a PAL/NTSC mismatch. I think the CD-i core defaults to PAL.

[unkempt dragon]

yeah i fixed that before. It's working now 🤷‍♂️ Maybe somehow loading he CD-i core manually first makes it work somehow? I'm just gonna let it run for a while and monitor it with SSH output

Is there a way to limit the commercials only to games found in the game lists? It's still trying to play a lot of games I don't have

[unkempt dragon]

still happening with discs, and i have this game in the folder

[unkempt dragon]

and after a couple videos the display just went to black again. Stuck with no menu. But the SSH output is acting like it's still running. I'm think I'm giving up on commercials lol

[fervent urchin]

something is wrong with cdi core here. I had my SAM running all night with latest test version and it's still runinng.

[unkempt dragon]

i got it from update_all, and I can open the core directly with no issue. Are you still using the unstable version of CD-i?

[fervent urchin]

yeah i deleted all other versions i had to make sure

Now playing: PSX Commercial - Final Fantasy IX
Starting now on the Sony Playstation: Final Fantasy IX (USA) (Disc 1) (v1.1)
Local file exists: /media/fat/video/PSX-WWF_SmackDown.chd
Now playing: PSX Commercial - WWF SmackDown
Starting now on the Sony Playstation: WWF SmackDown! (USA)
Local file exists: /media/fat/video/TurboDuo-Promotional_Video_-_Super_System_Card.chd
Now playing: TurboDuo Commercial - Promotional Video - Super System Card
Starting now on the NEC TurboGrafx-16 : Super Volleyball (USA)
Local file exists: /media/fat/video/PSX-Cool_Boarders_2___1.chd
Now playing: PSX Commercial - Cool Boarders 2   1
Starting now on the Sony Playstation: Cool Boarders 2 (USA)
Local file exists: /media/fat/video/arcade_JUNO_FIRST_starcade.chd
Now playing: arcade JUNO FIRST starcade.chd Commercial - arcade JUNO FIRST starcade
Starting now on the MiSTer Arcade: Juno First
Local file exists: /media/fat/video/PSX-FIFA_98_Road_to_World_Cup.chd
Now playing: PSX Commercial - FIFA 98 Road to World Cup
Starting now on the Sony Playstation: FIFA - Road to World Cup 98 (USA) (En,Fr,De,Es,Nl,Sv)
Local file exists: /media/fat/video/SNES-Looney_Tunes_Sunsoft_Collection.chd

I've had the issue with the black screen before as well though

let me think about what could be wrong with your setup

[unkempt dragon]

okay so if I wanted to completely start over, I'd uninstall SAM then get it from update_all, then disable it from updating again, then change it to the test branch and update from within SAM?

[fervent urchin]

what would be super helpful in squashing that bug is if you can activate debug and send me the monitor output

also, we should probably move this to the help channel or dm me if you like

[unkempt dragon]

i'll DM you once I get re-setup and have a debug log

[fervent urchin]

before you reset everything let's try one thing

i have an idea what could be causing it

[unkempt dragon]

ok

[fervent urchin]

echo "0000 0000 0000 0000 0000 0000 0000 0000" | xxd -r -p > /media/fat/config/cdi_sam.cfg

sorry just edited it

can you try this command and run sam again?

[unkempt dragon]

I'll have to get back to you later. I just fucked my setup cause I'm still tweking download filters and I accidently left out a "!" which means it deleted most everything off my card that wasn't explicitly a positive tag 🤦‍♂️

[fervent urchin]

no worries. that happens 😉

[unkempt dragon]

semi-related @fervent urchin : why not make a SAM dedicated post in https://discord.com/channels/647909397477195803/1049874847330738286 ? Like the "MiSTer Companion" or "RetroAchievments" posts where all of their feedback can be found

[fervent urchin]

good idea, but i don't think i can set up a channel myslef

[unkempt dragon]

It's a forum so I think anyone can make a post in there

Ope, nevermind

they got it set up different in this server apparently

I wonder who we can ask about that lol

[unkempt dragon]

@stiff flare could we get a dedicated MiSTer Super Attract Mode (SAM) post under the #1049874847330738286 forum?

[unkempt dragon]

@spring gate what's the protocol for getting permission to post on #1049874847330738286 ? Could we get a dedicated MiSTer Super Attract Mode (SAM) post in there?

[gray badge]

You could totally make one of you want.

[stiff flare]

Oops not ignoring you, I missed this

[unkempt dragon]

I would except it requires permissions apparently 😂

[stiff flare]

MiSTer Super Attract Mode (SAM)

[gray badge]

weird, I thought anyone could make one

[stiff flare]

IMMA DO IT

So I can add a nice photo

@unkempt dragon https://discord.com/channels/647909397477195803/1510271186687229982

[austere kernel]

I'm a little out of the loop here folks, what is the cdi_cdmountreset.rbf that's being pulled down by update_all? I assume it is anyway as I don't remember what the name refers to...

[terse flare]

That rbf is from February. Please use just the latest stable 20260502 or the latest build from the test-builds. Right now the last unstable is equal to the latest stable. The only addition of the newest test-build is CD+G support. But not many users will feel the difference.

[austere kernel]

Super. I'm going through all my Mister stuff here now as I haven't updated in a while. Many thanks Slamy 🧙‍♂️

[stiff flare]

@terse flare is a goddamn hero

legitimately turned me into a CD-I fan

[valid cairn]

The CD-I core is a beautiful thing.

This is me when I think about it

https://tenor.com/view/running-gif-11308139

[terse flare]

Negativity is a thing with German Engineers. This is why German engineering is the best in the world according to Jojo's Bizarre Adventure.

[nimble epoch]

There goes my hopes for Uncover Featuring Tatjana 😉

[terse flare]

Actually that game should work.😏 If you know the secret code of course to start it. There was some graphical corruption in the past but I think, the latest unstable shouldn't have them. Hopefully. I haven't tested it for quite some time.

[terse flare]

I'm actually totally unsure if I have ever talked about the issue with "The Apprentice" on this channel. I wasn't able to find evidence, so here is the issue explained to the point I know: Sometimes the title theme has crackles.
To your information, it is one of the music pieces that is always played as Red Book Audio. If no DVC is present, the ingame music is also Red Book. If a DVC is present, it uses Green Book Audio to be able to mix it with sound effects. This also means, when one leaves this game alone on the title screen it will play itself and constantly switch between Green Book and Red Book since it shows the Title Screen, the Highscores and some demo gameplay. At some day I forgot to turn the MiSTer off and there I had it.
I've decided to finally take a closer look and make a recording. Here some samples.

I'm not yet sure where it comes from. But I do know certain things. The crackles are spikes that repeat ever 589 samples. This number is not random. 589*2 is 1178. And 1178*2 is 2356. That is very close to the sector size of Red Book Audio. I assume a correlation. As the frequency of the issue is 2 sectors, there has to be a pattern. How a real CDIC does CDDA is unknown to me because the sample data is not viewable by the CPU as far as I know. You just tell it to play at a time code. So this issue should be very isolated and have nothing to do with CD-i stuff since it is just a data shovel. I'm using a double buffering concept with 2 buffers and my first idea was that the wrong buffer is played and the data is replaced during playback. But I was yet unable to prove this.

Here a screenshot for the interested reader, so you know how it looks like

[terse flare]

Ah, I'm so dumb and needed an AI to remind me. 🙁 588 samples is one sector. This is not mono after all, but stereo. So there is one crackle per CD sector. I've also now overlayed the raw CD image with my recordings and it turns out that the crackle is always on the boundary of a sector. Huh.. Hopefully I haven't spammed here. This helps me to reflect.

[tawny haven]

Is this while playing red or green?

[terse flare]

So far I was only able to reproduce this problem during the Title Theme. And that is red.

[tawny haven]

oh. was about to suggest if green I might guess there is an issue with not restarting at 0. If green uses adpcm there is a chance of overflow happening if the starting value has not been reset at the start of playback.

[terse flare]

Yes, how did you know? O.o Does the JaguarCD also use Green?

[tawny haven]

No. Red only. Google red vs green book audio says green does adpcm

I did do some adpcm stuff when fixing neo geo sound.

A message from back then found an issue that was made by the fan-hack for kof. The behavior for adpcm clipping vs wrapping was actually wrong so they encoded a sample incorrectly which would produce bad crackling on real hardware when it overflowed (emulators at the time used the wrong algorithm so it sounded correct there). A found a way to uncompress using the reverse of the bad algorithm, recompress with the correct one, then replace in the rom to fix the issue.

Just did a little experiment. After Rysha confirmed on real hardware that KOF2004SE has broken sound for Athena Down+Forward+B, I made a quick experiment. If you grab this: http://www.smspower.org/forums/11965-VGMLoggingWithOtherEmulators?start=200#67173 , you can convert broken samples to wav files. Then you can use this code: https://wiki.neogeodev.org/index.php?title=ADPCM_codecs#ADPCM-A , to convert it into correctly encoded adpcm files. The Athena sound effect is at 0x55f600-5613ff. So if you convert that to wav, then to correct adpcm and paste it back, it will play correctly without distortion. Just tried it, and it seemed to work. Obviously it would be better if the original hackers fixed it, but if you're desperate, you can do it yourself.

[nocturne vortex]

OH MY GOD yelled in a joseph joestar voice

[stiff flare]

weeb

[nocturne vortex]

I’m rubber and you’re glue what bounces off me sticks to you

[stiff flare]

NOOOOOO!!!

I don’t want to be a weeb!

[nocturne vortex]

Too late, dust off your miku body pillow and girl using the toilet resin statue

[strange crater]

I think I’m also able to recreate the Apprentice menu crackling on original hardware on my NTSC 220/37 (just a phone recording; excuse the CRT 60Hz hum):

[terse flare]

What? Am I chasing a white rabbit? Too be honest, I haven't tried to reproduce this on real hardware, because I never would have believed that this is normal.

[terse flare]

How high are the odds? Is it always happening? I don't get it. The CDDA data is usually not touched by the CPU.

[strange crater]

I’m not sure - it could just be that the CD-i outputs its audio signal hot, especially for The Apprentice. I’ve never noticed crackling with that setup, and was unable to replicate any crackling with an audio CD with the same configuration, but it is running under RF on my setup with the 220/37 and that could potentially be a factor.

I’m getting distortion instead of crackling during those sequences on the 910 - again, it could be error correction or just the audio level being high.

[terse flare]

Ah ok. I have to assume that this is a different problem. What this core is currently facing is rather rare but reproducible when tried often enough. I do have to say that manual testing this tedious. So it would be good if there was a scripting interface with the MiSTer to auto play controller inputs. This way I think I could fabricate a trigger on the bug. Continuously switching Highscore and title screen reproduces the problem at some point.

Is there a way using some bash scripts to control the gamepad?

[marble comet]

My chiming in uselessness: mister remote by wizzo, run from a PC, with any given auto-keyboard tool running on that PC? (Says the quietly spectating fan of your work)

[nimble epoch]

I've tried this one with the pinned core and Latest MiSTer main (CDi fixes) but it has the same issues.
The foto album doesn't show the pictures correctly and the picture on the playfield has also issues.

[terse flare]

Hmm.. I'm noticing now that I also have a regression for that title on my TODO list. Sorry that it seems to not work correctly right now. I will get in contact with you when I think it is fixed. It will take some time, because I've decided that the current style of replicating VMPEG is not ideal. It doesn't make sense to debug it how it is and I will look at this again with the current work in progress approach. But right now, I really want to understand what is wrong with the rather primitive function of playing Red Book Audio.

[terse flare]

It seems to be able to only drive the keyboard But maybe I'm missing something.

[marble comet]

Ah, you might be right. I have failed you! Unless you can remap controller buttons to keyboard in main? Notes im just guessing 🙂 sorry!

[eager pier]

Dammit Weeby

[eager pier]

Can Mister Remote let you send controller inputs though? I have used it and it seems like it's just for navigating menues, launching games, taking screenshots, etc. I have been able to use it to do keyboard input for computer cores but never was able to control a core that used controllers with it

[marble comet]

You seem to be right. Imma test and get creative right now!

[wintry dust]

just use some uinput based thing and create a virtual controller and a mechanism to send it commands. there's probably something out there

[eager pier]

One thing I never tried is configuring the controller to keyboard inputs then seeing if sending those over mister remote works. Though the problem is that mister remote makes you use an on-screne keyboard, even if using it from a pc browser

To be fair, it hasen't been updated in like 3 years, and I noticed several bugs with the way it displays files, which someone attempted to but ultimately failed to fix

[marble comet]

Ok, yeah. On main, I used mapped Joystick Buttons using the remote to input the keys to be mapped. Now, the mister remote is sending to the CDI for arrows and "A button". Defining withig the core worked too, for arrows and "B1". Its cumbersome and there's probably better ways, but its working technically. (Edit: Arrows were recieved, but not buttons)

[eager pier]

Honestly, I think a better way would be to just use some kind of controller emulator that you can scrpt, the kind used for TASes, you can likely make one out of somethign cheap like an ESP32, likely plenty of pre-made guides and software to do so, to simulate a usb controller

[marble comet]

But yeah, its probably pretty cumbersome

[eager pier]

Though AFAIK there is no real way to time it so you would just be running an input macro on loop hoping it dosen't go out of sync

[marble comet]

Eh, no. Arrows are working, but not buttons. Ima call my remote idea a failure 🙂

[eager pier]

I guess Slamy might be able to do a debug build that sends some sort of debug commands to help keep it in sync?

Dunno if that can have any impact on the issue though

[terse flare]

Please do not listen to this in public or with headphones. The noise is unpleasant. The problem is understood and I'm an idiot. This file was produced with a bare metal test ROM that delays the playback start command for CDDA for 30 video lines. That is also the reason, why the problem was not reproducible using the Audio CD player that comes with the machine. It requires a latency of handling the "Got Audio Sector" IRQ coming from the CDIC to reproduce the problem. After the CD Sector Cache was added, I should have implemented a certain rate to feed the data into the CDIC RAM. That never happened because I would have guessed it is not required. Sadly, it is required....
Here are the events

1. "CDIC, play from time code by reading raw data"
2. CDIC fills buffer A and causes IRQ when filled
3. The CDIC driver in the ROM will start playback with buffer A
4. Buffer B is getting filled while Buffer A is now starting to play.
   Since the playback of Buffer A is started during the reading of Buffer B, Buffer B is filled while Buffer A is still playing.
   On a real CD-i, the incoming data rate is equal to the sample rate, so both "pointers" have equal speed and don't play catchup.
   In case of the current CD-i core, a whole CDIC buffer is filled within 8 audio samples. So there is a chance of overwriting data that is currently played. It's a desaster with an easy fix.
   Hopefully I was able to explain it well. I've already thought about making a video to explain.