#programming
1 messages · Page 116 of 1
Nothing ever happens
Shout-out to badbutt21 and the guy who sent me this for keeping his name in there
rizin/cutter has always lagged like crazy for me for some reason
I had much better experience using IDA on Linux
is this a low-pass
also whilst I appreciate the rizin attempts at pseudo code output Ghidra's was much better (both of which are available in cutter)
(and both still pale in comparison to IDA's)

im actually un-nuking nuro
I found the big problem of AWA
it can't "read" what's on the stack
everything has to be hardcoded

I mean it's okay, I just need to do everything inside my brain
damn, nice work cpu
Well, what are you expecting lol
Call her a TV
ur welcom tv 


What?
This
But i still don't see what's being typed?
See the video again
The word that you type shows up on the touchbar 
Well that's just bloody stupid
As per the spec 
First i do not have a touch bar, second touch bars usually are located at the keyboard
You should not look at keyboard when typing
So does bime. Yet here we are, using bime on #programming 
But the idea is a keyboard that types for you by prompting LLM
And that keyboard does exactly that. It types into the system, and what you type on the keyboard are sent to LLM instead
Can't believe i scrolled up for 5 minutes straight just to get confused by stupid AI garbage
chat, how do I reverse an array, without accessing any elements besides the first one(as in index 0), without creating a temporary stack, without bitwise operators and complex math other than add and subtract


Smh my head lmao
you use the .reverse() method 
what language does not have a reverse()
AWA5.0 
Why not do temp stack tho?
AWA5.0
because I can't
Because the language only has 1 stack
this is all the functions, or instructions you have
Whut
So if you cant shove traverse am array backwards into another array and you don't have .reverse
How the frick yoy supposed to do anything at all
if you can only access the first element then I think you need scratch space
can't you blow every element and the merge them back. that should reorder them
Frick
I can group the top X elements though, and push them downward into the bottom of the stack at once

holy crap is that jerry hoshiumi
I'm trying to figure out an algorithm that works with that function
This language is even worse than bf
I can't store elements temporary
if you pop them, they are gone
then shrimply push the top element to the bottom of the stack N times 
that's what I initially thought of, but that would just lead to the same result as the input after N times 
"ABCD" ->(3 times) "BCDA"
etc
oh yeah right, it doesn't actually reorder them 
basically just turns the stack into a queue
what if you decrement each time? so ABCD -> (3) -> BCDA -> (2) -> CDBA -> (1) -> DCBA
i think it won't work... but sure
since this assumes you can move an element x amount of spaces. and not just to the bottom of the stack
you might need to do a bit more convoluted math to actually reorder them correctly
I can

that's the submerge instruction

although I need to group up some temporary variables, but that's already figured
pog
Could android stop deleting my client certificates without even asking? annoying as heck
oh nvm it wouldn't work
each loop is independent; the instructions couldn't know how many times the loop has been executed

that was very helpful to the clarity
you have the same issue I had where a single frame is high quality and the rest are low quality, i think I solved it with end-usage=vbr in the avifenc
ok... try this:
ABCD
move the first element one to the right
BACD
then take the first 2 and put them to the end
CDBA
then do the first one again
DCBA
this should be able to be procomputed as half the size of the array
Lmao, that is very cursed
not sure if this would would with uneven array sizes tho..
now what if my array has 12 elements
facts

so... my method might not work with longer than 4 characters...
i don't have a bachelor. so i suck at algorithm 
maybe try making a small python script that bruteforces the algorithm
i am here
yeah i am gonna say i am not great with development since i am in cybersecurity
Back to the void you go
i think now thaty ive reached 10K fps ill stop doing performance upgrades for a bit till i have something to improve again
im pretty sure i ran out of possible improvements i can do in C++
You still have asm and the spirv IR 
what are you doing, if you don't mind me asking i am curious
game-dev
it may only be a trianlge, but its the fastes triangle in this part of the galaxy
lol
what if i overclock my cpu? 
oh wait.. double bubbles are a thing. for even numbers i definitely figured out the algorithm.
so.. for even array sizes the algorithm is:
for half the array size:
submerge(1)
Surround(2)
submerge(0)
i guess for getting a clean array back you need to pop them. so...
pop()
submerge(0)
submerge(0)
and repeat this for half the array size again
but i can see the triangle, if it would be fast it should move at the speed of light

im in spain without the a
awa

Huh. Someone in a server was trying to make a government LARPing and called it skynet
wait.. no that still wouldn't work. due to the middle bubble being wrong
Skynet is the AI from Terminator
AB CD EF
BA CD EF
(BA) CD EF
CD EF (BA)
DC EF (BA)
(DC) EF (BA)
EF (BA) (DC)
(FE) (BA) (DC)
(BA) (DC) (FE)
(FE) ((BA) (DC))
BA (DC) (FE)
(DC) (FE) BA
DC (FE) BA
my current state of my notepad
I know I know
Oh I thought from how you worded that this was your first time seeing it lol
No it’s just someone recently in a server did a stupid system call skynet
¯_(ツ)_/¯

It was the most stupid thing ever
is this what you meant
I think someone is also working on the LSP for AWA5.0
I mean this(my code) doesn't work but I see the concept
i think there is gpu sorting algorithm that you can use that only takes in 2 elements in a recursive manner
Polyglot files pog
I made a file that holds different media depending on the file name.
Thank you for watching!
Here's a polyglot file with pictures of my cat: https://p2r3.com/meow
This project is available on GitHub: https://github.com/p2r3/beheader
Join our Discord server! https://p2r3.com/discord
Sound design by @dustyhob0. Music used, in order of appearance...
PR ran out of Portal/Portal 2 ideas 
Shhhh
This is better
I want a neuro polyglot file now lol
Reminds me of this
https://youtu.be/VVdmmN0su6E
Let's explore what a file format is, and provide a different view on it. We dive into polyglots, file format research and the impact on security.
Funky File Formats Talk: https://www.youtube.com/watch?v=hdCs6bPM4is
corkami/mitra tool: https://github.com/corkami/mitra
Guessing vs. Not Knowing featuring Steganography: https://www.youtube.com/wat...
LKJIHGFEDCBA
ABCDEFGHIJKL
BACDEFGHIJKL(sbm 1)
CD(BA)EFGHIJKL(srn 2; sbm 1)
DC(BA)EFGHIJKL(sbm 1)
(BA)(DC)EFGHIJKL(srn 2)
EFGH(DC)(BA)IJKL(sbm 5; sbm 4)
FEGH(DC)(BA)IJKL(sbm 1)
GH(FE)(DC)(BA)IJKL(srn 2; sbm 1)
HG(FE)(DC)(BA)IJKL(sbm 1)
(FE)(HG)(DC)(BA)IJKL(srn 2; sbm 1)
(DC)(BA)IJKL(HG)(FE)(sbm 0; sbm 0)
IJKL(HG)(FE)(DC)(BA)(sbm 0; sbm 0)
IJKL(HG)(FE)(DC)(BA)(sbm 1)
KL(JI)(HG)(FE)(DC)(BA)(srn 2; sbm 1)
LK(JI)(HG)(FE)(DC)(BA)(sbm 1)
(LK)(JI)(HG)(FE)(DC)(BA) (srn 2)
I'm now left with this
idk how to pop those double bubbles though
btw you can't double bubble a double bubble
well.. now you can just iterate again and after every 2 submerge(0) you call 1 pop
so:
(LK)(JI)(HG)(FE)(DC)(BA)
pop()
LK(JI)(HG)(FE)(DC)(BA)
sbm 0
sbm 0
pop
JI(HG)(FE)(DC)(BA)LK
sbm 0
sbm 0
pop
HG(FE)(DC)(BA)LKJI
etc.
well.. does it work for any size?
the reorder here works in any even size of array
mine only works on 12 elements, but it just need some fix to work on other sizes
the only variable is probably the sbm 5; sbm 4 part
it's supposedly halfsize - 1 and halfsize - 2
wait no

💔 washed
this reads like the ACLs on windows
Imma just get a sleep and see what my brain will come up when I'm half asleep
gn #programming


my college wants me to buy this course for self-study 
they have a coupon code to make it 10 bucks but still, id rather watch 12 hour youtube tutorials

if i wanted masterclass in a language i'd read the official docs available for free
euros*
bucks are euros to me
aight fair
i mean it is mostly used for usd i guess
these are the ones you cant buy yourself, they will buy it for you and send you the bill
like i cant even skip out on thsi one
Adobe Coercive Cloud
e
they are gonna make me learn github, litle do they know i already use github 
i will probably be forced to use a diffrent account tho
🍪
Om nom nom
You've given me 1 cookie! | I've received 239234 cookies total!
not necessarily, we got to use ours

the professor had us go through the github education thing (may not be how its named i don’t remember) and our usernames were associated with our names on his end
but that’s that professor, others just had us submit a repo/pr link
i love this dude
clean code 
clean code is a scam, nothing clean about it
this vulkan tutorial apparently had clean code
does this look clean to you?
cuz to me it doesnt
why the fuck are we storing 32 variables?
vulkan moment
i wonder if it’s possible to cram a webview onto the touchbar
huge for vk ceos
its been 84 years since ive heard from shiro about screeps
im beginning to think shiro is an ai too and vedal wiped his memory cuz shiro was playing too much screeps
ngl same

srgb evil 
his issue was srgb normalizing stuff while he was passing normals
honestly a skill issue on his side but still
srgb evil

could be 
@upper vine what is the meaning of this?
pull ghost
post removed
she's not even wearing the programmer socks, why is it here
I can still see the image but the link is invalid ye
Weird
artist zepped for rool for
the artist has most probably received a zeppelin bot warning for breaking rule 4 of this server
Should we ping a mod to remove the forward?

I saw the art in #starboard but it was deleted from there pretty sure
thanks for the catch
No prob
i like how google search doesn't let me search this image + "neuro"
?
what?
i mean
看起來是網址出問題

too sussy even for google lens
Watcha cookin?
What word is scribbled of?
this has to be some kind of a crime
not a word, an image
Of neuro? 
of evil
Oh 
We love the fake iPhones
I had my phone on mute when i downloaded that, what is that sound?? 
You use tiktok?
www
Why? Bro what did you watch on short? 
Yt shorts is my real personality, tiktok is my clean appearance
i just opened reels and i can’t send any of the first 5 i got
or i’ll get banned
be brave soldier

fwiw, the WKWebView class inherits from NSView (a generic cocoa renderable) which is very much displayable on the touchbar
i should try that
and cocoa is kinda like the ui part of win32
but macos
I know about cocoa, but man, why are you like this 
i choosed python for my project
i like doing things that you aren't supposed to
You should do more cyber-security instead lmao
i don't want to sound like i want to make a neuro two but i decided to try and make an ai so that it can play youmu cube adventures
a single player game that my friend made
cool
Did they not provide an abstracted enough API for the touchbar specifically?
Speaking of, Sam what kind of game are you trying to make?
rn we're doing the shovel game-jam game still. we've gone beyond the deadline to make it hopefully pretty good
setting up a cocoa app, then the private api wrapping (just having headers doesn't work for me, i have to... see the screenshot), then setting up the touchbar
the screenshot because i'm stupid
and you have to have delegates and controllers and yada yada
feels like i'm writing java from another planet, but still java
Yeah, Swift are closer to C# so it make sense that they are close to Java too
i guess it is kind of what it is
because all of the apis are objc
and objc is kinda java-ish
guy with darwin in their name typing probably an essay on how macos actually works and how i'm wrong 
My web dev education comes from 2017. But I feel like things changed since then. What do people use now? Like what does their backend run on.
I prefer to use TypeScript just because I like to use Cloudflare worker
PHP and ASP.NET is still used too
JS/TS is just popular because it is easy to write both front-end and back-end using the same language
i don't think things have changed much, yeah new frameworks came out, but like if you've seen express or flask you're not going to be confused on hono and fastapi
I was educated in php but I am considering js for a project because there are nice api wrappers for it.
Well, if you're doing a project to learn, then I'd just say go for it
It is going to get real users
Yeah I know, doesn't mean you can't do it while learning
then it does not matter even more than you think
you actually have income to cover compute costs
Just don't store PII and you're basically good to go
Uh, they are not paying users. It is more like a contribution to a community
I would not recommend rust for something that doesn't see high traffic lmao
But I guess if you're learning from 0 anyway, sure, maybe lol
Rust does have the api wrappers I am looking for...


By chance, is it something to do with interfacing with external device/embedded system?
Not this time
you asked about backend and the answer is simple. do you wanna hear what happened to frontend 
it's become way better tho
there's just sooo much change
They all converge to the same thing at the end yes, but still fractured somehow lol

Vue 3, especially with SFC should be a distinct framework tbf
It's even more insane than Python 2 -> 3
imagine taking a frontend class in 2025 and having to do react class components
But yeah, nowadays everything is having a compiler functions/expression to declare state and modify it imperatively.
React with State-Effect
Vue with Ref-Watch
Svelte with Runes
that'd be insane
"what is svelte"
That is the big 3 at least lol
angular
Oh yeah I forgot that thing existed at all 
This gist has data from SO up until 2024
https://gist.github.com/tkrotoff/b1caa4c3a185629299ec234d2314e190
Front-end frameworks popularity (React, Vue, Angular and Svelte) - FrontendFrameworksPopularity.md
i accidentaly made it faster again
now we're at 11K fps
i find it funny how the sampling rate of the profilers i have top out at 8Khz
You should not run the profiler at high sampling rate 

how is it supposed to get info on each part of the frame if it runs less than the frame?
Wtf there is a vue 3??? I have merely heard of vue 1
At some point the profiler will be the bottleneck, so the profiler is not run on every frame, rather you take a number of sample frame where the profiler is being ran on
I prefer to avoid vue. I want to keep client side js to a minimum
its fine if the profiler fucks up perofrmance a small amount, i just need to get an idea of what is currently the limiting factor of the program
Implementing things like buttons in js bypasses fundamental browser features
It is not a small amount 
as i said, good luck dude
11k -> 7K. small amount if you look at it in ms
For some browser feature, sure as the current HTML even provides <dialog>. But sometimes you need to dynamically change the nr of <input> element from other <input> element
as expected im fully cpu bound and the gpu doesnt even go above 40%
theo the annoying ex-twitch engineer now streamer did a great explanation of this phenomenon, if you've only ever worked on a blog, anything but vanilla js is mental illness to you. if you've seen ecommerce, everything other than js and jquery is. etc. etc. you get the idea
To be fair, that CPU usage can come from the profiler too
As I said, the higher your profiler sample rate is, the higher the bottleneck being the profiler
nah, the profiler is the bottom graph
about 5% usage
its probably running on a diffrent core? 
That's a lot of back and forth 
if you've never personally encountered the sheer complexity of some web apps you will never understand any of the modern frontend tech, because yes it is convoluted and bloated but it makes sense for a lot of problems, usually the complex ones
i bet we can sell them on svelte tho
More like astro or solidjs perhaps
solid? no chance
ive been running my keyboard and mice at 1khz when they could do 8khz cuz i never saw a reason for 8khz, but i guess it actually matters with these framerates 
too complicated
Depending on the things being profiled, it could be called low
you know what im thinking rn... the fps counter doesnt actually take the time it needs to calculate the fps into account does it?
Yep
Which is what I meant by the profiler may take more time calculating stuff the higher the rate is
Oh wait
nah wait does it actually?
every second
Nah, it doesn't matter then
It does affect performance tho since you are requesting time every frame from what I've seen
Yeah, you can implement it by counting the # of frame, then after say 1000 frame you see the delta time

i can't attach a viewcontroller to a touchbar item, it just always crashes
no touchbar webview
probably for the better

As if react native taskbar on windows is not cursed enough lol

"what if microsoft did the touch bar"
You know, if you release the RN binding for touchbar, people may write more things for it lol
Oh, you're not the first lmao
https://github.com/ptmt/react-native-touchbar
You are the one to speak 
oh it's just to add app functionality
i kinda forgot that's a thing already because i'm presenting system wide
with the private api
i suppose it was an easy port because it's just cocoa
wait 9 years ago rn for macos
didn't it officially get supported in like 2022
i remember watching the event livestream
i fucking screamed
cause like what
oled touch strip
I forgot that OLED is not new. It feels like ages ago that I still use CRT monitor 
these statistics are really interesting
that green bar with the numbered things like #486 is each frame, and the purple bar underneath it is how long itt akes to do vkQueuePresentKHR
vkQueueSubmit underneath shows how long vkQueueSubmit takes
i think that all in all this is maxing performance
most of the frametime is specifically those 2 vulkan functions
also i have discovered i can't read green text on top of a bright blue BG
ye nah
nvm
Dangit lol
you need to make the green and blue a bit darker
and smaller
that image is dispalyed on my screen at the size of my phone, believe it or not even elderly people can read that from a mile away
but i don't think there's a way to tap anything
not without a viewcontroller

Bro, even I struggle to read that kind of text lol. It's just bad contrast at that point
Yeah, that's about the same color I'd say
at this point its just the font
It's the size that makes it difficult not the font
and the size
Well, the font contribute, but mostly the size and contrast
the font here has less holes making it looks like a blob to me
Try to see the color contrast here if you want
https://www.achecks.org/apca-accessible-colour-contrast-checker/

It uses OKLCH color space so it's easier to make color of the same perceived brightness on different hue
now that i think about it, 11000fps means about 90 microseconds of frametime.
every time you blink 1000+ frames will have passed
it takes 5x as long as the frame to even make your brain procces it
A nerve impulse travels about 1 meter in that time, so the frame will have less latency than the nerve signals from your feet
thats crazy
all of this is bs tho since you're limited to gsync lmao
wait, it seems the default behavior for webkit webview is to ignore touch
i think i have found a way to make it work
delete clang
5
12
5
no
1059236278345936926
neuroSadge

Hey, why I didn't see that poll?
i love linux 
wonder why 1.20 crashes but 1.22 does not
tbf i rarely ever touch this ubuntu instance anyway so guess it doesn't matter
Does WSL even pass the GPU?
yeah
Interesting. Maybe embedded development wouldn't be such a pain anymore if they can pass USB
iirc they do support usb passthrough in some capacity, can't tell you how well it actually works though as i've never had to use it
Hmm, I'll still not rely on it I guess
they do give you an xwayland server (albeit a rather scuffed one) and pulseaudio so most things should work
damn
nurp
konii, this is midnight. What is a TV doing at this hour
wrong
2am
i woke up at 1:00
It's getting there at least lol

thats mine sleep schedule
we have four victims
proves that github scams work
what is this

full(not full) 
it has two different rickroll links
glados 
chat did I cook
#programming brain rot is on another level I swear lol
this seems like a loopable way to reverse arrays with arbitrary sizes
I'm just trying to print "Hello, world." in AWA 
AWA?
I know, and mlnt is playing with touchbar and webview now
But also, I think that is indeed loopable way in AWA5.0
we should make AWA but NEURO
instead of "awa" and "wa", we use "neuro" and "evil"

We could just redefine awa and wa 
VedalSCII sounds cool
with an array size of 12
proven to work

now I have a question, did we invent this algorithm or it has already been discovered 
although this method works, I have to reimplement my loop cuz well we have to have places to store the data of the loop itself
originally it just groups the array and sbm 0, but now we grouped the array already, so we can't do that anymore
someday if i get around to it ill make an AWA backend for hblang
Anyone know how in Godot I can store a mesh outside of the displayed game world? And I know anything can be saved as a "scene", but I need to have meshes generated while the game is running to have a storage space outside of the game world
just leave the generated mesh in the memory, instead of displaying it?
Okay, one day later on my laptop. Now it is time for asm
it's time for some AWAness
The test program copy pasted at least works this time
I feel like I'm back to the 90s
goodmorning
580.88 is out, yuppie
that is so poggers
I reversed an array with 4 elements
now it's time to scale it up to 12
what esolang is dis 💀
it's AWA5.0
awa
I thinking of making a step-by-step AWA debugger, it would be way more efficient than plotting each step by myself
remind me of chicken
get some lava chicken
it's 11pm i aint going anywhere 😭
awa awa awawa awawa awa awa awa awawa awa awa awa awa awawa awawa awa awa awa awawawa wawa awa awawa awawa awa awa awawa awa awa awa awa awa awawa awawa awa awa awawa awa awa awa awa awa awawa awawa awa awa awawa awa awawawa awa awawa awawa awa awawawa awawawa awa awa awawa awawa awa awawawa awawa awa awa awa awawa awawa awa awa awa awa awa awawawa awa awawa awawa awa awa awawa awa awawawa awa awawa awawa awa awawa awawa awa awawa awa awawa awawa awa awa awawa awa awa awa awa awa awawa awawa awa awawa awa awawawa awa awa awawawa awa awa awa awa awawa awawa awa awawa awa awa awawa awa awa awawawa awa awa awa awa awa awa awa awawawa awa awa awa awa awawa awawa awa awawa awa awa awawa awa awa awawawa awa awa awa awa awa awa awa awawawa awa awa awa awa awawa awawa awa awawa awa awa awawa awa awa awawawa awa awa awa awa awa awa awa awawawa awa awa awa awa awawa awawa awa awawa awa awa awawa awa awa awawawa awa awa awa awa awa awa awa awawawa awa awa awa awa awawa awawa awa awawa awa awa awawa awa awa awawawa awa awa awa awa awa awa awa awawawa awa awa awa awa awawa awawa awa awawa awa awa awawa awa awa awawawa awa awa awa awa awa awa awa awawawa awa awa awa awa awawa awawa awawa awa awa awawawa awa awa awa awa awa awa awa awawawa awa awa awa awa awawa awawa awawa awa awa awawawa awa awa awa awa awa awa awa awawawa awa awa awa awa awawa awawa awawa awa awa awawawa awa awa awa awa awa awa awa awawawa awa awa awa awa awawa awawa awawa awa awa awawawa awa awa awa awa awa awa awawa awawa awa awawa awa awawa awawawa awawa awa awa awa awawa
this is hello world in awa

yuh
I just saw the merge instruction in the AWA5.0 specs, so bleh
I gave up the idea of using loops now
without proper address management, and the limit of only able to access the topmost bubble is annoying me
I do have a demo of using loops done above, but it's too complex, mainly for debugging
maybe I'll work with it after I got a debugger on AWA
awa
awa 

i think i broke something
it work on image
transparency
jpeg looks better but for more than 10x the size its resonable
noe this is a different case
this one is the jpeg one
2mb png vs 83kb avif
discord probably re-encodes it so there's not much point in me sending it to begin with
bwa
the same
you could send a part of it like here
I see some difference
I think discord mostly strips metadata
i think if you open the original file it should be correct at least
though for image previews they also provide scaled images
Borders are smoother on left 100%
so yeah jpeg looks better
its not possible to tell on a 1080p screen from the image preview lol
there is no jpeg!!!
avif is noticeably less detailed and more blurry but thats what you get for 25x size reduction
lol
whats the other then?

its a very different comparison
I should see how well avif works on something that needs the 25x size reduction.
png compression is lossless
its especially noticeable here
#!/usr/bin/env sh
enc_args="-c:v libsvtav1
-pix_fmt yuv420p -fps_mode passthrough
-preset 1 -speed 1 -crf 25
-g 100000000
-svtav1-params tune=1:scd=1:scm=0:film-grain=0:enable-qm=1:qm-min=0:qm-max=9:enable-tf=1:enable-cdef=0
"
ffmpeg -i "$1" -y $enc_args -hide_banner -loglevel warning "${1%.*}.avif"
after playing with flags here is what i settled on
thats why it looks better
you can change the crf to be lower for higher quality
there are probably better options for aom or rav1e, but i could not be asked because of how long they take to encode
svt i can just churn out images
ty, I'm not waiting for encoding for this test when it is 8k res
luckily I can also compare because I got previous attempts at reduction logged
if you swap -preset 1 -crf 25 for -preset 0 -qp 0 it should make it lossless
i think
or, just the crf 25 to qp 0 is the main change
lowering preset just makes compression slightly better
but very diminishing gains
oh also you may want to remove -pix_fmt yuv420p if your image is HDR
otherwise it will get converted to SDR
For lossless I'm using JPEGXL, though WEBP has same results.
This image is a strong stress test anywayz
Don't have HDR to convert
im pretty sure jxl is a better format for static images than avif to begin with
but we shall see
a funny thing i learned earlier is it's possible to put av1 in webm, and the sizes are smaller than av1 in avif
quality was somewhat worse than avif though
which makes little sense since they are just containers

webp and webm are different
yeh
webm is a container format (like mkv) while webp is an image format (like png)
Step one is adding some padding because for some reason the height of this image is odd
did you modify the flags at all
Nope
ah
Resolution size complaints are valid
try changing colour space to yuv420p10le and the preset to 5
I'm just trying with what it had first then editing
I added a line of white to the test image so it is even
I figured I needed a copy anyways
should not be used for benchmarking until fully implemented
So anyways I started benching
Alright, from 126M to 16M in the test image.
The JPEGXL with lossy compression was 36M
how does it look though
I'm gonna need another colorspace or software because gimp is not having it
not colorspace I'm blind
HEIF support
can we please stop
Why is there more I swear to frick
theres also jpeg-based jng
are they useful at least
a codec is only as useful as its support 
I can only read that as autisms which I think is by design
replacing au with awa
theres even mjpeg (jpeg but video)
awful format name
Agreed, awful naming
Nixxxxxxxxxxxxxxxxxxx
Okay, so getting gimp to cooperate is going to be harder than usual. So time to find something else
you mean ship? trams dont go on water
(This is both 2.10 and 3)
More than one image viewer has claimed it is invalid
I guess that's the experimental support they were talking about
Got it
Discord loaded this, fml
Processing for a solid 10 seconds
If anyone has any idea how to get the over 100MB PNG for comparison I'm all ears
Real
zstd was basically a noop
huh, using http3 just breaks everything. Guess thats why its experimental 
lol
you won't believe this, but we need to develop one more universal standard that covers everyone's use cases
nice
Let me convert this to webp rq
I at least know gimp has a lossless option
theres even mjpeg (jpeg but video)
tied with jpegxl
oops sent it a second time
internet went off for a while so discord thought it didnt send it
multi processing 
See my message about not to trust filename lol
what is this picture about
looks completely normal
Lmao
Yeah, looking at the implementation interpreter doesn't help much 
The picture itself is something, It's pretty soulless looking despite not being AI generated. I'm using it because I've tried a lot of compression stuff on it.
I'm going just just make new structs of bubbles
I can't read this javascript thingy
if it works it works
oh my god theres jpeg.org with all of these
https://jpeg.org/index.html
the funniest part is when you open a page it shows a top menu bar where you can switch between jpeg standards
and it overflows
this is more referenceable

We are not getting this to the discord upload limit lossless
I'll first make it to be able to run each single instruction one by one, then make it able to run AWA5.0 scripts
I cannot wait to share the finished product with the community if you don't mind it 
Death
there is AWA5.0 community??
no way
It's basically just a bunch of the streamer fan 
I told you there's someone working on the LSP before 
@sage crag your getting JPEGXL at quality 1 because I've run out of options
awa
This is one pixel shorter because it is unmodified
It was good until JPEG decided to dip into their reserves
At what point do you think the compression quality will just be equal to an interpolated image from lower res one
who 
Not on this server
Let me check because I believe I noted this
Also, check out curl jorb.xyz/jelly if you wanted. There are people working on porting the server to AWA5.0 
:O
It's just this btw
If I had to guess the downscale would just be better in all cases.
JPEGXL is noted to be 5% and 1/4 scaling PNG is noted to be 15%
Granted I believe the PNG scaling was biased a bit
Okay, 1/4 res is a bit extreme, but considering the original size that image may already contain a lot of redundant information
We end at 2k res so that was my reasoning back then
I wanted to try and keep detail, but I had a different objective back then, which was to get a thumbnail version on web at all.
if only the situation was the same in the audio codec sphere
you have mp3 which is old crap, aac which you have to pay for and vorbis basically the only good one
Opus?
i forgor

opus is the best probably
OGG Vorbis is peak
My guess is that this is acting like 1/4th rez
This is the URL of the image

probably a person who thinks anything but vanilla js is the devil
i’m not saying they should use react on a blog
but seems like the type
I was about to give it F tier until it finished the second round of loading
I can excuse not exact for 10x compression
Because it does have all of the detail so 1/1
The tram's electric engine induces a current in my bt headphones 
I can hear the trams speed
Seems similar to the effect my image has, a lot larger tho
✨
will re-encode this to avif give me 3 years
I give it my seal of approval on working with absurd+ resoultion
It's the best file size to quality ratio so far
first 20 minutes will be allotted to downloading
they seem more like the type to use jquery
awif
psb 
If that is photoshop psb, gimp is not opening it
Speaking from experince
I would advice to not use image manipulation program to open it anyway lol
What is ram
Your copilot will have awatism by the end of the day
guess which is which format
awa awawa awa awa awawa?
vibe coding so you can feel like you’re management with a particularly stupid junior dev
i feel like avif maintains colours better than jxl but maintains structure worse
if that’s what that feels like i’m never taking a management position
JPEG, PNG, AVIF
jpeg - avif - lossless ?
2 and 3 could go either way for me
do you really think png would have chromatic blur and avif wouldn’t
idk how you actually call the effect so i made the name up
it looks bad though and png is lossless
I was told to guess, what do you want
lossless doesn't mean it can fix artifact that is already there. It can still be png of photograph and it may be indistinguishable from jpg
Aight i have arrived at the correct part of the beach 
That's what they'd like you to believe
jxl avif png
avif compression sometimes almost looks like ai artifacts
looks like it doesn tit

file sizes of them on lossless mode by the way
mhm, all three are good overall. Really have to look for it
As is tradition first to do is get drinks 
png: 2.0 MiB
jxl: 1.5 MiB
avif: 898.5 KiB
try webp with -lossless 1
SMOL, SMOLer and SMOLerer
matches jxl in my experince
Hold on a minute, if I was only 5MB off from being under 100 and this is a large jump. Then I might be able to make it afterall.
One last attempt time
Same size, not conviced the web is correct
It's 3:00 so I'll just call it a night with knowing I got a better format to use
wont have internet for a few hours 
It's not an image 
i ran the png through oxipng first, getting me down to 1.9MiB, but the size of the avif file doesnt change
psb is a photoshop project file, made when it gets over 2gb on disk
dw i have a plan
MY PC
Swap is being used
Always firefox
i wonder when it will be done converting
there's no ETA
also its nice to know that imagemagick can just convert psb to png somehow

Probably just rendering the project (which gimp should be able to do but can't for some reason)
processing realtime is very different from processing with no time limit
too much ram usage 
#programming and cursed side project
- Bime
- Promptboard
- AWA5.0
- L A R G E image processing
Throw in Sam Game Engine for good measure lol
Sam would approve
we hope
oh it died again
Now I am curious
i dont think i have oomd
It can't be OOM
If it's OOM you'll most likely notice a heavy slowdown before the assassins comes
I'm gonna do a funny
https://psd-tools.readthedocs.io/en/latest/
oh no
Imma note this for later because I have some legacy psb files I'll need to handle.
Do not do a funny and use this one if you need lol https://github.com/EmilDohne/PhotoshopAPI
Awa
Sooner or later, AWA5.0 will be used here, competing with hblang
I saw some code earlier, I'll wait to see how it goes
im doing i

this is going pretty quickly actually
i think
the png is already 1.5 gigabytes
nevermind 2.3
its done
nevermind
it just got a preview
It doesn't explode?
Now to actually check 
Why is the image rotated 69536 x 22230 
All i know is a 3:4? 16k image was 1.1GB when PNG
69K is above the allowed mak framebuffer size for every singl gpu 
69k? It's 1.545.785.280 pixel so even worse 
32K monitor when?
Idk
smallest Skyrim mod texture

and the classic https://www.nexusmods.com/skyrimspecialedition/mods/99094
I see a new Azaka pfp

I did my 2.0 today 
8K is the max of intel igpu's iirc at the time
So that tomato would be potentialy gamecrashing
i forget 30% of neurocord are vtubers

Is that so?
to be fair the stock tomato looks doodoo
no
theres a good number of regular chatter vtubers mostly in genchat

Thats not a tomato
Thats a plastic todler kitchen prop
good that they know to stay away from nn
5090
maxImageDimension1D 32768
maxImageDimension2D 32768
maxImageDimension3D 16384
maxImageDimensionCube 32768
Ye these specs have been the same for probably 10 years. Even the 1080ti could do thid
A minimal, quick AWA5.0 Awably(the name for the instruction language) to Awalang encoder. It does not have more checks than basic linguistic grammar issues, so be aware of issues on your own. Two examples of Hello World programs are included for testing the encoder. (Forgive my bad coding)
@opaque wharf 

The supposed output for print(convertCode(hello_world)) if verbose = True
oh discord doesn't embed .awa 
So, when are you debuting? 
Soon™
would that be a "virtual machine" if konii debuted
🥁
If we can have AI VTuber, why not TV VTuber 
1Hz stream 






peppy allow avif
