Star Wars: Republic Commando RTX

[modern isle]

that's the whole thing

here it is with simplified labels... you can see that this is very simple

[dry shoal]

Yeah that doesn't look bad at all

I feel like as I learn best practices and fumble less using instaMAT, that I'll be able to make some amazing stuff. It's so powerful that I really should be getting familiar with it

@modern isle would you mind sharing the link for that material?

[modern isle]

figured out what a cooked project is... it's just the selected graph with all the references embedded.

[modern isle]

let me know if this works for you

[dry shoal]

what's the difference between it and a project then? doesn't that also save the graph

then again I'm not quite sure what references is referring to

no pun intended

[modern isle]

a project has all your stuff as you set it. so if you didn't embed your textures, it's just a loose path reference. So if you were to share that file then all those textures break because it expects the absolute path to match

[dry shoal]

that makes perfect sense

TLDR, use baked exports lol

(when importing stuff)

[modern isle]

if you want to share the file

[dry shoal]

true but I figure even just for ease of use

[modern isle]

nah, no difference if you don't share the file

[dry shoal]

like if I use a texture from downloads bc im not organized, I can delete that file and not have to worry about it

I think I might be misunderstanding

worked great 👍

[modern isle]

when importing a file you can pick if you want to embed the texture or use the absolute path, the former will save that texture inside your project

[dry shoal]

that already looks better to me

decreased the roughness, right is new

not equivalent lighting conditions but it's a good reference to have

I need to make it a lot darker and a lot less rough if I want to get closer to the original

I'll probably have to use a different base texture entirely

[dry shoal]

Due to some (probably very good) advice I've gotten, I'm gonna scale back the scope of the project for the time being. I won't be doing anything more with meshes than I already have (anything more significant at least) to speed up the process. Instead I'll be focusing on making good materials, lighting, and various asset swaps without changing the underlying mesh. The exceptions would probably be baking displacement maps into meshes since that's relatively easy

In other words, I will be using blender as little as reasonably possible

[modern isle]

Some advice on scope. Focus on getting that first level done. That includes replacing all the textures with PBR materials, including just some nicer textures for the guns and clone troopers. Once you've done that and basically have the first level completely playable and remixed you can evaluate how much time you want to invest in new models.
Getting that base done should be your focus in my humble opinion, you can always improve from there.
But good enough is good enough for a start, trying to get everything perfect while not having most assets in place yet can be very difficult to work with.

[modern isle]

Not trying to rain on your parade, just don't want you to get hung up on stuff that prevents you from getting this at least done because it would be a shame if you'd grow tired of all the roadblocks and stop this project due to frustration. You made great progress and we're all really happy to see you make such strides in such a small time span, but I'm sure others agree that staying a bit smaller and then scaling up is probably the smart choice.

[dry shoal]

I don't take it as raining on my parade at all. The ones with experience know best, and the idea makes perfect sense. It's simply the right call

Sure it sucks that all of my ideas on models are being put on the backburner, but my progress slowed down dramatically trying to do it when it was still a bit too beyond me. By focusing on where I'm more comfortable I can accomplish more and probably have a more enjoyable time doing so

It's also fun getting actual screenshots which I haven't gotten much of in a bit lol

[burnt light]

i fully agree with this, this is what i've been doing and it gives you a great idea of what you'll be dealing with going forward, and what to aim for in terms of how manageable your project is and what you want the game to look like

for example, if you replace all of the meshes in the first level, but decide it's too much to do the rest of the game, you'll have wasted a lot of time on something that just makes part of the project look different than the rest (even if it's better)

[dry shoal]

It makes a lot of sense. I imagine you've also learned a lot more about instaMAT than I have within the same time frame for that reason alone. As for the scope, I've already basically decided I wouldn't be doing the whole game. When I started this project I already knew that would be a huge task, so I chose the first level. For the Remix project it's really just to see "what can be done" if the skillset and manpower was there to remaster the entire game with Remix. Unsure at how much I'll be doing with meshes when the time comes but likely quite a lot

Though in case there was any confusion, this project is a standalone one-off. Once I'm done with this and move to adapt the RC Remastered mod into Remix, everything I've done is gonna be scrapped

For the sake of efficiency and not being dumb, I'm gonna be working entirely with the stuff they've been making. So their modified level files, their new materials, meshes etc

By toning down the ambition and working with a group, a start to finish Remix mod can be made way quicker

[burnt light]

ya, that makes sense. that partnetship sounds great, and i hope it works out

[dry shoal]

Me too. Those guys know their shit though, they've been updating that mod for years

Funnily enough they're kind of doing a similar thing as to what you guys suggested. The scope of the mod grew with time, and only after the 1.7 release of their mod are they putting considerable work into modeling and making new materials to replace the previously AI upscaled ones

[dry shoal]

replacing the terrain with anchored ground meshes. I also found out how to get the original blends (see image!) to work well with the underlying terrain. I also added back the lighting in the tunnel

emissives for a nice bloom effect! otherwise the texture would be uniformly dark

doesn't make much sense for a lightsource

now that my priorities are changing for the project, I have a few things in mind I want to work on

I want to edit the existing storm trooper PBR textures a bit to touch them up, make a new ground material that would include covering this tunnel. The only mesh related things I have in mind at the moment is just a basic subdivide of the tunnel mesh since the angles are visible

if I keep it simple, surely nothing will go wrong

It feels good to be back in the editor actually making stuff happen :)

[dry shoal]

fun fact, these are the exact same mesh

:)

I've now discovered the power of normal baking

both 3k tris but you wouldn't believe it

it's not perfect but for a first attempt I'll take it

[dry shoal]

new clone texture has lots of potential :)

[modern isle]

This was THE major reason why everyone hammered you with making a high poly AND a low poly model 😛

[dry shoal]

Listen I always believe you guys but hearing about it isn't the same as seeing it ‼️

I will say it's a fairly unideal setup because there are some differences in shapes in parts of the mesh but I'll worry about that later lol

[analog wedge]

Will you concern yourself with wear? The clone armors are brand new in the first chapters and made out of some fancy plastic.

[dry shoal]

The original texture has boatloads of wear. I want to try and fight a balance between pristine armor and what's seen in the vanilla game. The game makes more than a few deviations from the canon look of different things for the sake of having a unique aesthetic

[dry shoal]

Heads up, I'm going to slow down/pause work on this project for a short period of time. Gonna do some work with Runestorm on https://discord.com/channels/1028444667789967381/1211038540218109953 to get experience working with a more diverse set of materials and I'll be able to get more direct feedback from him to improve my skills. When I take that experience back here, it'll be a benefit to the project

Don't have an exact timeline in mind but we're certainly not talking weeks here. Gonna go with the flow for now though

[dry shoal]

Semi update: now that I'm more familiar with the process of mesh baking, I should be able to retain a surprising amount of detail from that 20K poly clone trooper model swap I did while retaining the original 3K poly as the one in-use. This is helpful because it means I don't think I'll have to spend nearly as much time worrying about topology/retopology and I can more quickly make higher fidelity meshes in more calculated ways. Especially thanks to other tricks and tools I've learned in blender (not quite there with adding micro details yet, nor am I ready to learn it). My work process in general has been very slow lately but I hope to pick it up soon. For as painful as making a pile of hay for Unreal Tournament '99 was at first, it taught me a lot about how to properly use instamat. At least for environmental textures like the ground or rocky surfaces, I'll be making my own materials from scratch for Republic Commando. I won't have to worry about the same limitations as the original texture when being used as a starting point, and I can get much higher fidelity + more tuning to come out exactly the way I want it to. I'm excited to get back to work on this game, but for now I'm gonna do more for UT'99. I want to come back swinging to Republic Commando and that means getting familiar and comfortable with the material making process

^i say semi update because it's not fully related to RC progress

[dry shoal]

https://media.discordapp.net/attachments/1196836285545971712/1215151467502239784/image.png?ex=65fbb4fe&is=65e93ffe&hm=bbdad839b66983eb36e02dd651ff53ee3aa14a44cc71895fd2a02d48d484f308&

Started hard surface modeling for the first time yesterday. Current goal there is to remake the grenade box by hand in higher fidelity

[analog wedge]

dilta skwodd

[dry shoal]

doing more modeling stuff to brush up on my skills

btw the right side of the 2nd image isn't my asset, it's one from Republic Commando Remastered since they shared the blender project file

they seem to be open to using assets on a "which one looks better" basis lol

so if mine looks nice they don't mind replacing one they've already made, and if I can't figure something out I'm more than happy to use what they have

for now I'm trying to do everything from the ground up because I need the practice

[dry shoal]

WIP thermal detonator

[dry shoal]

https://media.discordapp.net/attachments/1098995005512765481/1216560441845616760/image.png?ex=6600d533&is=65ee6033&hm=074f8edd89864611743c1abc9664745002f75a8d5db3a1d69c9e512edbc9a395&

I'll likely follow the advice of others and remake it some time

[cobalt cypress]

Ok, this is actually SO COOL

I just joined to see what people were up to, and I am not disapointed

[dry shoal]

Thanks so much comments like this motivate me a ton

I've recently started learning hard surface modeling so that's what's currently been my focus. The thermal detonator is gonna be remade, and I was going to make the grenade ammo box but a friend working on RC Remastered had one already done, and at good quality! I tried myself for the practice if nothing else but eventually I gave up because I didn't plan ahead enough and it caused issues

[cobalt cypress]

Well best of luck anyways. I remember playing this game on xbox, and if this remix ever gets launched, I will get the game on PC

[dry shoal]

Currently my plan is to Remix the first whole level as a playable demo, similar to Xoxor's Nacht Der Untoten demo for World at War

Still have a lot ahead of me but it does make it 10000x more manageable than doing the whole game solo

After that demo is completed, I was looking to make a Remix adaptation of the RC Remastered mod which would be significantly faster given multiple people are on it and plenty of assets already exist that could be used in Remix

[cobalt cypress]

kindof odd question, but is there a limitation to the type of games that can be remixed, like DX version, or engine that its running on?

Like does the game have to be Dx9 or older to have this RTX Remix Support

[modern isle]

#remix-beginners-guide

[cobalt cypress]

Thanks

What a bummer tho

dx8?

that sucks

That sucks in the sense that you can't remake other games like Titanfall

Good graphics, but it would be cool to see the textures revamped

[dry shoal]

Republic Commando is a DX8 game 👍lots of DX8 games are working with Remix since the DX8to9 wrapper works super well

It's interesting, it's going from DX8 -> DX9 then DX9 -> Vulkan. Yet it works super well

Going from my experience and seeing the experiences of others, it seems like Republic Commando is one of the most stable/functional games with Remix right now, which is great for what I'm doing lol

But yeah, the way Remix works is by intercepting fixed function rendering pipelines. There's multiple versions of DX9 with slightly different names, and by the time DX9 was at the end of its lifecycle, fixed function pipelines had largely stopped existing

Since Remix is open source, including it's renderer, it's possible on a technical level that you could do an extremely complex and invasive mod that natively changes the renderer to a pathtraced one. The amount of technical skill required to pull something like that off isn't feasible by nearly everyone here

[cobalt cypress]

Lol the #remix-beginners-guide channel says OpenGL "May" not work, so does that mean that Mincraft java theoretically "Could" work. Although I read closer and it talked about shaders having to be below 2.0 so idk

idk, I guess im still pissed at Nvidia and Microsoft for hyping up RTX minecraft, and then never releasing an official universal pack to use.

[dry shoal]

If someone were to downgrade Minecraft's openGL renderer then theoretically it could be done. As-is, no clue. However Minecraft Java already has pathtraced shader packs that are similar to Remix. Wouldn't get stuff like DLSS and there's some additional fancy rendering stuff Remix uses, but they're incredibly stable and incredibly beautiful still

[dry shoal]

I remember seeing community made packs that made it RTX ready. They were always a hassle to find though

Biggest problem is how niche it is

[cobalt cypress]

Yeah, but the community ones are meh, also, dlss is stuck on like 2.5 or smthing, and dlss tweaks bug out on it so

[limber marsh]

Tbh, its almost certainly microsoft side

Nvidia wants "everyone" to use it [cause money for them].

Microsoft wants people to play versions that generate "them" money.

Its like that with this

Legality is a problem and decisions made are because of that.

Also remix is basically in beta despite it being released and even still its not a year 1 thing.

So dont have any wild expectations about game support.

[burnt light]

i mean, it was released as a beta. they stated so

[dry shoal]

They took out out of beta in under a year however

[burnt light]

i mean, no?

https://www.nvidia.com/en-us/geforce/news/rtx-remix-open-beta-half-life-2-rtx-trailer/

it's still in open beta

unless i missed something?

[dry shoal]

probably would've helped if I read all of his messages huh?

I missed the most important bit

[burnt light]

ah okay. i thought i was going insane and missed something important, lol

[limber marsh]

i must of misread it - i read the open source announcement originally as release but I kind of meant everyone involves knows its in "beta" despite what public/news may perceive.

[proven vessel]

looks like things are still coming along

[dry shoal]

yeah... swimmingly...

[proven vessel]

lol

[dry shoal]

Time to get organized

[dry shoal]

Gonna take a look at Ray reconstruction in republic commando sometime soon. There's definitely room for improvement when it comes to denoising and I think there are some instances where it will be a big improvement with the perf bonus as a cherry on top

[proven vessel]

noice

[dry shoal]

game's busted for now. I'll do a fresh install later but

the base of my thermal detonator mesh is complete

[dry shoal]

After this thermal detonator is 100% completed, I'm staying in my lane for a bit and focusing on significantly easier things

a big one I have in mind is redoing the surrounding geometry in the arena of geonosis

[dry shoal]

The meshes I wanted to work on most are too intricate for me to take on right now. Or even when they're not, I just can't be assed to work on another mesh that needs a lot of care to do well

[dry shoal]

Later today I'll be attempting to finish and texture the thermal detonator mesh. Specifically, making a high poly version that includes the ridges on the outer shell. For reference:

[modern isle]

I think there's two approaches to this, the one I mentioned earlier with the line bevel, and another one that might be easier for you. When you select a poly strip, detach it from the model so that it isn't connected anymore and do that for all the parts so that you end up with individual pieces for each segment. Then remove the faces that aren't on the outer surface as we won't be needing them and extrude everything inwards for all the segments. Then apply a bevel to the outer edges or just use a bevel modifier

[dry shoal]

I'm sure that's a pretty easy way but conceptually I have a harder time wrapping my head around it

[dry shoal]

Thanks a bunch. It also pains me to realize I could've done the middle pattern more easily 🥲

My dumb ass inset all of it, then and only then did I extrude bit by but each chunk up and down accordingly with auto merge on to make sure it attached

[modern isle]

[dry shoal]

did a fresh install of the game, using the latest available builds for dxwrapper, remix runtime + ray reconstruction files from portal RTX + swrcfix

it works better than it used to out of the box but some UI still needs to be manually tagged. Lightmaps seem to be ignored automatically so they might not need special treatment, but I'm not ready to say for certain. Sky textures still must be tagged, despite using auto sky detect with depth flags which I thought would make it "just work"

Looks like deleting the two intro video files may no longer be necessary to ensure stability

[dry shoal]

Note to self: by default the night vision mode spawns a point light when light translation isn't disabled

setting the night vision overlay to UI breaks the pathtraced lighting

[dry shoal]

lol I love running into issues I had the first time I tried SWRC with remix and not knowing how in the world I fixed it

it's not a big deal since I have 4 backup installs that have the exact settings/remix config that worked before thrown in

but still a little annoying

Kim offered to help figure out the terrain blending configuration since I'd love to get that working if possible. The first mission uses tons of blending in vanilla

[dry shoal]

https://media.discordapp.net/attachments/1196836285545971712/1218804296234172486/image.png?ex=6608fef4&is=65f689f4&hm=d93ae5d609d7dc250dba2a7bb43ca5ec75d94ff97e1af8b97100d970f2190b4d&

I'm gonna go insane if I don't finish this thermal detonator

Still WIP, that's just the last "best" spot I got to

I undid it after so we'll see

[dry shoal]

The thermal Detonator mesh is finally complete

Original mesh, another modder's mesh, and my mesh for comparison

keep in mind the middle mesh is gonna be/is used in the native game engine. He's also fairly new to modeling so at the time he could only do so much with the limited knowledge + avoiding too high tri counts

[dry shoal]

@past escarp when you were trying RC, were you using the rtx.conf from moddb?

[past escarp]

I used the one you had linked here: #game-compatibility message

[dry shoal]

well that's interesting lol. I'm running into an issue I had when I first started using Remix with the game where some UI elements existed in the game world behind me, so there were constant billboard shadows

but I haven't had that for a while so I'm not sure if there's something I missed

even on my moddb file it's messed up

[past escarp]

That's very strange. I'll have to try the game again and see if I spot that.

[dry shoal]

should be super hard to miss. Thanks, and take your time

@past escarp looks like I found the culprit!

if you have "capture vertices from shader" enabled, it does that

if you plan on still testing the game, would you mind seeing if you can recreate the issue on your end using the same setting? We're on different rtx.conf configs so it'd be interesting if it's working with it enabled for you

[past escarp]

I did try it out just now. It does only happen at certain angles for me, or at least only becomes noticeable. Turning off "capture vertices from shader does help with that, but it introduces a new issue, I have noticed. I will take a video and show you momentarily.

[dry shoal]

if you mean parts of the game world not loading in, yep. Got that too

[dry shoal]

I'm starting to wonder if this is a regression in the Portal RTX version of Remix. Are you using it too?

well check out what I just found

[past escarp]

Not with this game as I noticed that it currently lacks the coordinate toggle from the most recent DX-VK builds.

[dry shoal]

What coordinate toggle are you referring to?

[past escarp]

"Scene Left Handed Coordinate System"

[dry shoal]

I found the exact same billboard floating behind the player that we saw

in two captures, both times behind the player

I wonder if deleting this in the editor will remove it in-game (edit: it's the same mesh, deleting one deletes in both captures!)

[dry shoal]

I remember seeing that too. My build doesn't have it but I'm seeing none of the behaviors associated with not having that left hand coordinate system enabled

maybe they made it automatic?

[past escarp]

Hmm. Worth a shot. Maybe try ticking "Ignore Material" under Remix Flags in the editor.

[past escarp]

Hmm. I will give it a shot then and see if that is the case.

[dry shoal]

yeah this is a safer bet anyways than deleting the mesh. It's a generic screenshot of the rasterized game. If it's ever called again it should be avoided entirely

welp despite going to selection history I can't seem to undo it

oh well. If I'm fucked I'll find out later

[dry shoal]

Slowly building this list. This has unironically helped me stay more organized and on-task. Right now my focus is on 3D assets. Not sure if I'm gonna texture the thermal detonator just yet, I'm kind of dreading it lol

[modern isle]

Texturing it should be a breeze, it's made of fairly simple materials. Just needs some weathering on top and presto.

[dry shoal]

Honestly now that I think about it, half the problem with my trouble in layering is that I'm using stock materials. I really should be making each material in the node graph so it's exactly how I want it, then throw it into layering and add weathering layers on top of my custom materials. Does that sound about right?

[naive robin]

I'd suggest using the stock materials, tweaking them as needed, and then adding grime and such ontop

That's how I do it in substance painter/sampler

[dry shoal]

Problem is I've found adjusting existing materials to be pretty unintuitive but then again I'm also stupid

I think the thermal detonator will be easy just because there's so few materials and they're very consistent. It'll be the grime and wear on top doing the heavy lifting for detail

I think that'll make it easier to get the ideal exact result

[naive robin]

In instamat it's kinda okay, I don't really like how they're layering stack is compared to substance.

[dry shoal]

Tbf I don't like how the layering stack works in any software. It's just not intuitive to me with how the visual is. Blender is more readable for hierarchies and that's saying something

[naive robin]

If you've used photoshop, the layering is the same in all the adobe softwares

[dry shoal]

That said I have a feeling I'll be much more comfortable using the node graph and making each basic material. For metals it would be a solid color with very minor discoloration. Then I use all the different mesh masks to apply the imperfections in layering

[dry shoal]

I have but I have very little Photoshop experience

I was thinking about buying Affinity Photo 2 since it's on sale and it's a one time purchase alternative to photoshop

I found out about it today

[naive robin]

Would wait till they do a sale. They typically get disconnected pretty good.

[dry shoal]

Well I did just say they have a sale right now lol

[naive robin]

Reading can be difficult sometimes

[burnt light]

i just got Photo 2. it's a huuuuge improvement from v1, and it's good enough that i'm comfy using it over Photoshop

[dry shoal]

Don't worry I'm the worst at skipping important info in messages

[dry shoal]

Ooh that's good to hear

[burnt light]

i'd use their free trial before purchasing btw

[dry shoal]

Will do. Just gotta check when the sale ends

The thing I worry about with the free trial is if I won't have use for it right away and by the time I do I won't have it

Not related to RC in particular but there are some patterns that feel like they should be very easy to do but are messy to do procedurally via instaMAT. I had to finesse the system with two blend passes just to get the one basic result

I imagine stuff like that will be significantly easier on image editing software like this

[modern isle]

it's all about knowing when to use which technique

[dry shoal]

playing with the lighting a bit

I was able to successfully disable the stretching of the blue blaster projectiles (top right of crosshair) but didn't find the one for droids in the engine yet

only 1 projectile in my capture so I can't see if they all have the same ids lol

unfortunately this change didn't make the projectiles share the same mesh IDs. Gonna do more experimenting with time

[dry shoal]

in my experimenting I have broken the projectiles lol

I was able to disable the lighting from muzzle flashes which fixes a major visual issue

[main turret]

Man, can't wait to test this out!

[dry shoal]

Totally forgot to post this here!
left: original
middle: handmade recreation that wasn't a good idea
right: my new version based on the handmade recreation

Working on some of the surrounding geometry in the Geonosis arena

Way less stressful to work on than the thermal detonator

[dry shoal]

WIP

[dry shoal]

In an effort to not stay stock on one thing for far too long, I'm likely gonna stick with this for now^. I may come back to improve it more later since I know I can do better, but I really want to see results and that starts by putting together the progress I do have before worrying about perfecting it

[dry shoal]

Made some solid progress. Done with the thermal detonator texture for now, so next I'll be adding it to the grenade ammo crates so see how that looks

A big one was successfully disabling the muzzle flash within the native game so no more broken ground lighting in combat!! This was a big one for visual continuity so I'm glad I solved it

I started messing with various properties of the different particle emitters around the arena map to try and change how they function to better enable high quality flipbook replacements. If I use those flipbooks now, you get random standalone fires floating up and disappearing lol

I might be able to take advantage of the dynamic particle emitters system however. That I'll look into in the future as it'll get fairly complex

I've also begun working on making the blaster bolt projectiles have stable hashes so I have more control of them in the Remix toolkit. Ideally I could have full length capsule lights attached to each and every bolt to get the best possible visual effect. Something like that is only possible with pathtracing (rasterization can't do shadow casting lights of that shape) and I want to make that happen

The current system I had for adding lights to the blaster bolts was a hack that involved altering the Remix runtime to hardcode lights being attached to the texture itself. It's very difficult to tweak and at least that implementation was limited to adding sphere lights instead of capsules

I've been considering acquiring a blender plugin that should make redoing the environmental geometry quite a lot easier. I'll look more into that as it'll save tons of time and get better results than what my skillset currently can output

Lastly, I'll try looking into building a test map with native game modding to have easy access to everything I'll be modifying within Remix that isn't otherwise easily accessible

[dry shoal]

I'm so fucking inspired to work on these goddamn rocks

[dry shoal]

Small update of sorts

I recently had the idea to create a test map that would contain all the assets I intended to alter in some capacity to make the replacement workflow stupidly simple. Unfortunately I scrapped the idea because it would take enough work to be impractical. For example, you cannot place animated meshes. So the alternative is to edit the properties of a mesh that can be placed to then skin it with an animated mesh as a sort of model swap. It gets messy and might mess with Remix functionality if it's identified as a different hash

Putting the editing of particle emitters back on todo. There's lots of stuff to explore here but I've already broken some things exploring the settings for them. So I'd rather prioritize getting stable hashes for the bolt projectiles if possible. I've been told other weapons use proper static meshes for their projectiles instead of a stretched texture, so I may have to check those in the Geo hash view just to make sure it can be replaced in general, before I waste time on something that won't work

Something I forgot to add for artwork: I plan on taking another stab at the clone mesh I made previously. I've learned so goddamn much since getting the mesh that I know I can do it better now. I will go the whole way (within reason) and make a banger if I can help it. That will be for later due to the ambition of it

The DC17m primary rifle is another mesh I want to remake. I can replace that one in Remix directly so it'll be easier to work with

Trying to build up to the ambitious stuff rather than dive in head first though. So for now I'm gonna continue to replace the environmental geometry with significantly improved assets like seen here

[analog wedge]

Are these supposed to be natural formations?

[dry shoal]

kind of. Some of the ones I've replaced, yes. But one is a natural formation that's been altered by geonosians as a hive of sorts

this is one for instance. They would come out from the top in the lore

[dry shoal]

WIP rifle remake

[dry shoal]

WIP - base and barrel done

[dry shoal]

replacement ground texture. Will take a look in-game soon enough

[dry shoal]

previous material replacement vs new

[dry shoal]

in-game

it's times like this that I remember: I can absolutely make this look like a modern AAA game visually with Remix. Fucking incredible that this is possible at all

meanwhile some native game modders I've been talking to are constantly keeping an eye on their RAM budget and making cuts to fit everything in they're trying to make

deep respect to them. I'm happy not to deal with that though lol

[limber marsh]

People need to learn to ignore people that are on low spec

Its ironic because I actually remember the fixed function dx8-9 era of PC gaming.

Games were worst on PC because peoples hardware were generally lower then consoles

So in return many of those games are stuck as inferior versions then even console games to this day. If people have waited 20 years for a remaster of a game, then those people can wait another couple of years to when they upgrade.

[dry shoal]

maybe I worded it poorly, they're limited because of the engine itself. They can't push beyond a super limited amount of RAM else they risk it being unstable with constant crashing

I don't have that limitation with Remix

I think there's room for both oldschool style mods and remix mods. I really like what they're doing and there's an undeniable charm to working within the constraints of the original engine. Not what I'm going for but I'm glad it exists

in other news

I was able to get a mix of gravel and rock by using a seperate mesh placed within the surface of the rock!

decals aren't supported natively within Remix, so workarounds like this are necessary. So psyched I got it, it'll keep things from looking too samey and the original game did a lot of this in these same areas

in-game shot!

this project is brimming with potential. Every little addition thrown in motivates me more

[static mason]

oooooh That looks dope

[modern isle]

As auto mentioned, the modders are limited by the engine as it's only a 32 bit game they cannot exceed about 3 GB in total of ram usage. This isn't about low spec, but generations of computer architecture limitations that isn't solved by just throwing more stuff at it.

RTX Remix can circumvent this limitation with the bridge that it's using to essentially create a 64 bit application on top of the original game.

[limber marsh]

I missed the "native game modders" context but their are people looking to use remix for "card neutral" updates which feels like half measures so I reacted. I understand in-engine mods are very limited

[dry shoal]

In regards to Remix specifically I do agree we shouldn't concern ourselves with widespread hardware support. This is cutting edge tech and it's fully intended to be pushed in really cool ways. By the time Remix mods are more widely available, people will be on more capable hardware anyways :)

The GPUs of tomorrow should handle Remix just fine. Hell, we have people here on RTX 2060 cards! As long as you adjust expectations I think many people can have a cool experience with this tool. The only thing is I think it's worth keeping in mind VRAM limits, and that's more about being smart about tiling rather than be wasteful using unique textures for everything

[dry shoal]

RTX Off:

RTX On:

@main turret this has been my first side by side comparison after replacing materials through Remix, figured you might want to see

I ask that you don't show it off though since so much is subject to change. Like almost every single model in this screenshot needs a makeover

I'm working with the most suboptimal setup since I'm subject to whatever awkward UV mapping the original models used to show off my new materials in an awkward way lol

[main turret]

Lovely stuff!

And hey, I won't be checking it out until you're happy with a release demo

[dry shoal]

You got it 👍 hope you'll look forward to it

Btw, as I get further along in the project I'll be looking for feedback on various aspects to tune things

In particular, stuff like how warm the lighting is, how intense the eye adaptation, how strong blaster bolt light sources would be, stuff like that

I'm still learning as I go, even though I feel like a completely different person than I did two months ago when I started in terms of skillset lol

So learning and actually getting stuff into the game comes first

[weary radish]

@dry shoal is it normal that the mouse is really ... really floaty when having the remix menu open? Feels like I'm controlling the mouse via joystick 😂

[dry shoal]

unfortunately yes

also be prepared for the gun to shoot on its own, a lot

[weary radish]

I've noticed

Couldnt see any of the visual smearing you've mentioned in the opening scene .. which is good ig 😜

[dry shoal]

iff you mean Kamino then yeah it's surprisingly solid. Just super minor culling (when using anti culling) and the transitions between scenes not having the fade in so you see some weird things

[weary radish]

I mean the smearing issue I had in bo1. What is Kamino? 😂

[dry shoal]

kamino is the planet on the intro. The scene I was talking about is the flight cutscene with the briefing

if you're referring to that then..

[weary radish]

Yea that

[dry shoal]

really? it's fine for you?

[weary radish]

yea well I'm working on fixing that

Alex (nv) is guiding me and I think I've made some good progress

[dry shoal]

sorry I'm a little lost

you mean the issue in general?

[weary radish]

yes sorry

[dry shoal]

because if you haven't had the issue then there's gotta be something on my end

would you mind sharing your current rtx conf for republic commando? I've made changes from the moddb upload

[weary radish]

sorry for the confusion. I'm currently brewing a PR that fixes that issue (hopefully)

[dry shoal]

OHHHH

[dry shoal]

so the Remix runtime you have atm doesn't have that issue?

[weary radish]

^

[dry shoal]

implying this and the monkey one might be the same thing

[weary radish]

well the game itself was the main issue. It dew the ape in random draw order and remix just wasnt able to tell them apart. But I still had some flickering here and there and that is now gone with the code changes I've made

[burnt light]

that's great 🙂

[dry shoal]

so you think it's the same thing with RC? The clone being drawn in an inconsistent order

[weary radish]

Man input is hard in this game 😄
I think setting Unique Object Search Distance to 100 should also fix the issue for you

[dry shoal]

yeah you need the fix mod if you didn't install that. Should be a part of the instructions though

I'll try out the unique object search distance, thanks!

[dry shoal]

I hope it being this high already isn't a concern @weary radish

[weary radish]

Well thats what's causing the smearing issue. Try reducing it to a value that stops it from happening

try 150 first

[dry shoal]

remade mesh that more closely represents the designs seen in the films

made it a cross between these two, as you can see

super proud of it but next up comes actually doing it justice with a good texturing

I might also do a bit more wear on it

[dry shoal]

made adjustments to the scale and did a new break cut

might make the head shorter since it do be looking a bit chunky. It's fairly big in the original design though

[dry shoal]

shot of the original hallway lighting for self reference

why the light would be blue when the mesh is clearly yellow, I have no clue

[dry shoal]

Doing my best to stay on task and be organized 🙂

[dry shoal]

Made a little progress the other day, the hash stability reference was very helpful to give me an idea of what work had to be done, and more specifically what things I would have to implement into the game itself later on

I did a bit of testing with the ignore baked lighting tag in remix but got extremely mixed results. I had quite a lot of hope in it but ended up disappointed. That said, it's possible there's still some benefits I haven't found yet. I have a couple more ideas to try on that front

The DC17m priority is definitely being bumped down. It's such a significant model to upgrade and it would give an amazing first impression on screenshots. But it's a complex model by my standards and while recreating it (with a touch of my own flair) has worked so far, I hit a bit of a wall that slowed my momentum. There's more I can accomplish with it but it's not time efficient

I'm gonna try and pick up the pace with simpler but significant assets. I'm also looking at improving existing materials

[dry shoal]

before and after. Updated the ground texture to be darker and more wartorn, like the original. I also changed the ground composition to be more consistent. There are more cracks across the surface total, but they're dramatically more subtle

[burnt light]

looks great 🙂

[dusk trail]

@dry shoal Sorry for the ping but just noticed this project, if you want some help with props or some environmental art let me know, especially hard surface stuff, been doing it for years, and I'm utterly obsessed with star wars stuff

[analog wedge]

DILTA SKWODD

https://tenor.com/view/reaction-big-eyes-seriously-shocked-gif-12278669

[dry shoal]

Hi! Thanks for sharing your interest. I intended on largely doing this solo but if you'd enjoy making some stuff then I'd be more than happy if you contributed. Goes without saying but when the time comes you'll be credited ofc!

[dusk trail]

I'd be more then happy to contribute where and when I can

[dry shoal]

I'll have to look through and see which hard surface assets could use a facelift. I have some WIP stuff like the DC17m rifle and the thermal detonator (my 7th attempt...), and I'm really invested in working on the bacta dispenser in the future so I call dibs on that one :)

[dry shoal]

Sure, so how would you want to go about it? Would you like me to find a few hard surface assets to pick and choose from that you work on?

[dusk trail]

Haha yeah all good, I haven't done much in the way of weapons but I have done grenades, weapon attachments, and stuff like that before

[dusk trail]

Sure whatever works for you

I should be able to gauge if I could do it or not by looking at it

[dry shoal]

I ask because I have quite literally no experience doing collaborative modding just don't want to be too ignorant is all

[dusk trail]

Haha yeah no problem, I completely understand that

[dry shoal]

Sounds good. I think most of the assets shouldn't be too tough. You might not have seen it, but this project intends to cover the first Geonosis mission. geo1a, geo1b, and geo1c. I want to keep the scope limited so I can put my all into it and ideally not take years lol

[dusk trail]

If you wanna just make a list of stuff you could use help on I'll take a look and see if I think I could do them

Images would be the most helpful

[dry shoal]

Sounds good to me. I'll also share the assets directly since sizing/general similar shape is important. The asset is being swapped by Remix but the underlying collision is still there, so gotta keep that in mind

[dusk trail]

I scrolled up a bit and saw some of what you've done with geo, looks good so far

[dusk trail]

Alright, and yeah I'll keep that in mind

Feel free to DM me anytime as well

To my understanding it works a bit like how Halo MCC did the remaster for CE and H2, where all the underlying geo the is same and only the meshes and textures are changed

[dry shoal]

Still lots more to be done too. Much of my asset replacements so far have been a part of my learning, but may not necessarily be what is used in the end. I really want to nail this and I don't want to settle for good enough in terms of the final product

[dusk trail]

I get that, good to know where you want to take it

[dry shoal]

Exactly! But this project is interesting because not all assets can be added through Remix. Dynamic meshes like the bacta dispenser need to be added in-engine, with the PBR materials added through Remix

[dusk trail]

I can't say I'm the best with texture as it's probably my weakest skill but modeling I got down more then well enough

What's our poly limits and what not?

[dry shoal]

Wait I just wanna clarify, I'm not saying that for you, rather what you saw above isn't the best you'll see from me 🙂

[dusk trail]

[dry shoal]

Don't worry, I can handle that fine lol

[dusk trail]

Haha yeah I got you

[dry shoal]

For Remix replacements, technically we could have multi million poly meshes

[dusk trail]

Did not know that

[dry shoal]

Performance scales differently with pathtracing. Adding tris costs less performance as you use more and more, it's not a linear scale

[dusk trail]

I'm still going to optimize my meshes tho, and make them tris and what not, pretty much a habit as part of my workflow at this point

[dry shoal]

Well that's the limit, I don't think there will be multi million tri meshes in the end

[dry shoal]

Totally. Basically that is to say you can crank the detail without worry. We should always optimize but we should also take advantage of the tech, and maybe do a little less optimization than normal if you get me lol

[dusk trail]

Do it anyways for fun, the most random prop will have 4^10th tris for no good reason

[dusk trail]

Yeah gotcha

[dry shoal]

Hey, I was able to implement an 8 million tri haystack mesh testing unreal tournament

Performance drop wasn't the worst lol

[dusk trail]

So in that case I imagine having to bake details down into the normal map via high-low poly bakes would pretty much be unnecessary for this then

[dusk trail]

[modern isle]

could, not should btw. Please model responsibly and keep it reasonable 😛

[dusk trail]

I haven't messed with Remix enough to know exactly what it's limits are or exactly how it does certain stuff

[dusk trail]

Lmao of course

[dry shoal]

In a sense yeah. But if it's something that would be fully represented by a normal without any quality loss, might as well bake it. But we don't need any visible polygon edges

[dusk trail]

Fair

[dry shoal]

The TLDR is don't overthink poly count. Don't work as if there's a strict budget. Make something as detailed as you wanna make it and we go from there

[dusk trail]

I'll probably get marmoset again then, or I could try messing with Blenders normal baking again, it's gotten much better then it used to be

[dusk trail]

Fair

[dry shoal]

You said you don't have much experience texturing right? I fully planned on doing any and all texturing before this, so I can handle the baking of the assets you make if needed

Basically the boring stuff you don't have to worry about lol

[dusk trail]

I have experience doing texturing I'm just not great at it by any means

[smoky condor]

im gonna have to say it, but rc will be ass to mod sometimes lol

[dry shoal]

You're talking to the king of being not great

[dusk trail]

Or well more like I don't think I'm that good with it anyways

Hold up 1 sec

I'll give an example

[dry shoal]

Yep, TheMan is right. But since this is gonna be a hybrid mod, 95% of the asset replacements will be via remix. We'll only do in-engine stuff when needed and I'll do that side of things

[smoky condor]

^^^ you're lucky with remix

[dusk trail]

Not trying to do self promoting btw just showing what I meant lol, but here's an example of what I meant

https://www.artstation.com/artwork/D5l2LA

[dry shoal]

Don't sweat any of that stuff, I don't get that vibe at all. Lemme see this

[dusk trail]

https://www.artstation.com/artwork/dKPx2Q

Another example

I can model without issues but I kinda struggle with the texturing part, always have

[smoky condor]

you could definitely use this for the assault ship

[dusk trail]

Nah I'd rather redo it tbh lol

But thanks

[smoky condor]

yeah i ment as a reference mostly

[dry shoal]

Did you do this solo? Pretty incredible

[dusk trail]

Haha idk about that, I mostly just came up with a lot of it myself design wise

[dusk trail]

Ye, took me I think like a week or so to do, mostly because I kept going back and changing things over and over

[dry shoal]

Only a week

[smoky condor]

i mean parts like the windows could be used very well, it will fit for sure, but thats up to you ofcourse

[dry shoal]

[dusk trail]

@dry shoal Something like that ye

[dusk trail]

Fair enough

[dry shoal]

I see what you mean by the texturing. It's good but I'm really trying to push myself on that front with the help of a few experienced artists teach my dumb ass along the way

[dusk trail]

Yeah I get that

[dusk trail]

Probably would have been done sooner if I could have just made up my mind on stuff before hand

[dry shoal]

Oh boy I fully expect you to fly through stuff compared to me lol

[dusk trail]

I only really had a very vague idea of what I wanted it to look like

We'll see

[smoky condor]

yeah a week for that seems really fast

[dusk trail]

I suppose

[smoky condor]

and it looks really good

[dusk trail]

I think at the end of the day it's just because I've been doing it for so long, I think like 8 years now or something, started learning how to model in 2014

[dry shoal]

Yep, that would do it

[dusk trail]

Haha yeah

[smoky condor]

damn

[dry shoal]

I got a whole 2 months of experience baby

[smoky condor]

for me probably half a year

[dusk trail]

I was still in middle school at the time, said f it and that I wanted to learn how to make my own stuff, and fell in love with game dev and environmental modeling

Been doing it since

[smoky condor]

but never really did anything till auto came along and helped a ton with really nice addons

[dusk trail]

Same for me with Blender, I hated Max and Maya myself

[smoky condor]

idk i never used anything other than blender

but im modding the base game, and support for the format isnt really that great

[dusk trail]

Fair

[dry shoal]

Same and I've had at least 2 rants about it so far had a bit of a debate in this server about it with multiple people LMAO

[dusk trail]

I had to deal with a lot of old crappy software when I ported Kotor 1 into UE4 a few years ago

Was pain

[dusk trail]

Haha yeah it be like that sometimes

[smoky condor]

rc base is next level pain, im currently replacing brushes with static meshes so i can apply normal maps to them, but theyre tripping out somehow now

[dusk trail]

Oof

[dusk trail]

https://youtu.be/sMKlsbzhKOg?si=6AvECZpgnBHdPveh

[smoky condor]

not to forget we have about 2 gb of ram to work with, 4gb patch barely does anything

[dusk trail]

5 years ago...rip..

[dusk trail]

[dry shoal]

The limitations of native RC are certainly something

[dusk trail]

Yeah

[dry shoal]

There was a huge project on it a while back, but it was canned due to frustration with the engine. It was really unfortunate because they did some amazing stuff

[dusk trail]

I'm glad some companies are going back and patching their games to run on 64bit like EAW did recently

[dusk trail]

Ah

[modern isle]

They look pretty alright tbh, are you looking for C&C?

[smoky condor]

^^^ we did get some stuff of it though and implemented it in our mod

[dry shoal]

This is news to me

[smoky condor]

huh

wait i didnt tell?

[dry shoal]

[smoky condor]

short terms: they gave us custom scalers so the textures dont physically scale when their size is increased

so it stays small and crisp

which is how we got hd reticles and such

[smoky condor]

which you never saw before

[dusk trail]

I think I'm being dumb, C&C just means command and conquer to me rn lmao

[modern isle]

Critique and comments

[dusk trail]

Oh

[dry shoal]

Yeah uhhh might have to maybe yoink that eventually somewhat

[dusk trail]

Haha yeah that makes much more sense

[smoky condor]

welp you're in luck since i finally cleaned up my texture folder a bit

[modern isle]

I mean I'd also be totally down to throw some fists with NOD and GDI 🤣

[dusk trail]

You can if you like, I was just more or less showing that materials and textures are mainly where I'm lacking skill wise

[dusk trail]

C&C mod for RC when?

[smoky condor]

i have to say though, gotta ask metallica for permission, idk if you need it for remix tho

the dll that is

you can just grab the textures if you need them

[dry shoal]

I think UI is a complete no-go for Remix. Has to be game side. Anyways for the time being I wouldn't worry about it, hopefully by the time I do ask he'll see I put in enough work to be worth sharing it with

[smoky condor]

yeah just ask when its needed

tho

the download is in the rc discord

[dry shoal]

I do not see it

[modern isle]

The textures are a decent base, star wars is pretty grimy and worn in general, I've seen you've added some scratches to the metal edges (I assume with smart masks materials?) what your textures would greatly benefit from is more variation and dirt in the roughness texture to give the surface more depth, in addition to a full wash of light dirt and dust. I can whip up some quick examples tomorrow if you like.

[smoky condor]

actually

you dont have perms for the rchd channel i noticed lol

now you have

read through it and you'll know how it works

[dusk trail]

Fair analysis, I can agree on those points, I appreciate the input, and yeah sure I wouldn't mind looking at any examples

[dry shoal]

Runestorm is the master taught me the majority of what I know for material making

[dusk trail]

Actually never asked, does remix also work on character models?, like how does that work

[dry shoal]

The answer is pretty nuanced

[dusk trail]

[dry shoal]

The simple version is: anything that's a static mesh is usually fair game for remix, you can go crazy with the replacements

[dusk trail]

Soooo remix is generally for static stuff but anything dynamic will have to be modded in the normal way

[dry shoal]

The longer version is that most static meshes are remix replaceable, but few if any dynamic meshes (animated ones) are. For those meshes we'll have to implement those models into the native UE2 game engine. However, all models can have their textures changed, so we could put in a high poly clone but have shitty textures, and we apply the good PBR shit in remix

[dusk trail]

[dry shoal]

Basically, but the vast majority of what I've played is replaceable

Like off the top of my head, clones + droids + Geonosians + bacta dispensers are dynamic meshes

[dusk trail]

Fair

Yeah they're all gonna be skeletal meshes since they animated

[dry shoal]

That can't be replaced in remix. The laser bolts are also non replaceable but on the Todo list I want to see if I can turn the projectiles into static meshes specifically so remix can mess with it more

But like I said, textures are fair game. So I replaced the laser bolt texture and gave it an emissive

Example here from the editor

[dusk trail]

Neat

You may need to walk me through how to get that working on my end later

However that'll be a bit later as I am currently on vacation for the next week or so, I'll be available to start doing stuff around the 14th

[dry shoal]

Right now my project related stuff is super messy. I can totally help get the runtime up and going, but sharing my changes requires building a shareable mod package. Can be done, but I'm always messing with stuff and not everything is implemented together since I've done multiple Remix project resets to learn

So it wouldn't be all that interesting outside of a nice ground material and the lasers glowing lol

[dusk trail]

[dry shoal]

I plan on doing another reset of my project but I do have all my assets still on hand. I've never implemented 100% of what I've done all added together

[dusk trail]

I wonder if remix has any sort of live shared assets sort of thing for big mod teams

[dry shoal]

They have to do something like this for Half-life 2 RTX, they have a pretty big team understandably

[dusk trail]

At the very least I'd be able to test some stuff out on my end and see how it looks and get more familiar with it

[dusk trail]

[dry shoal]

I think they do a layers based workflow. So each person may work on a specific thing within their own layer, which can then be added together and toggled on and off

[dusk trail]

Ah, I was I wondering if they used something like perforce or something

[dry shoal]

Nvidia Omniverse has a bunch of live editing fancy bullshit but AFAIK remix hasn't gotten it yet

[dusk trail]

Hmm

[dry shoal]

In one of their other Omniverse programs you can literally edit within the same world live with multiple people

I really hope that's brought into remix at some point

[dusk trail]

Yeah

Because rn remix seems to be more situated towards solo devs but the amount of work to get something really good would require a few people to get anywhere sometime decent

[dry shoal]

It might be a good idea to start with your own unmodified captures. But I'm down to export the project so you have an example. I think you can edit packaged remix mods too

[dusk trail]

Good to know

We'll get to that when I get back tho

[dry shoal]

You're not far off. Most mods have one or two people, those with more are usually in the wild West exploring stage where they're still seeing what they can get working. TF2 has had some exciting progress recently on that front. But the need for speed underground 2 team has like 7 different artists? Don't know how they collaborate, I know they have a dev discord for more organized communication

[dry shoal]

Of course, and enjoy your vacay

[dusk trail]

Makes sense

[dusk trail]

[dry shoal]

well guys

I reset my rtx.conf to mess around and test

I may have made a breakthrough

HELMET POTENTIALLY WORKS NOW

[dry shoal]

ok I see a problem with the helmet

[dusk trail]

I hate how detailed that emote is

[dry shoal]

What can he say, bro is passionate

[dusk trail]

Fr

Lmao

[dusk trail]

?

Am I missing it?

[dry shoal]

Gonna drive, but I'll give the rundown after

[dusk trail]

All good

[dry shoal]

Didn't take a screenshot of it. But essentially I was able to get the helmet to work where it didn't previously without breaking the camera. Before the view model and direction of shots wasnt lined up and the entire screen would wobble in a very disorienting way

So I got it going now, but I've come across an old odd issue I don't remember much about where there are these giant billboards attached to the back of the player's head. They contain things like UI and oddly enough, a live feed of the game with normal rasterized visuals

You can't see them normally, but because all lighting is realtime they contribute to the global illumation and cast shadows in an obvious and ugly way

Previously the hacky fix was to enable Remix's dev freecam to back up, select the billboards, and set them to "ignore texture" in the Remix runtime. Before that did the trick. Now when I do it with the helmet enabled, the view breaks and everything beyond the helmet itself is just a solid blank screen

These leads me to believe whatever the billboards are doing they're necessary for the function of the viewport. I have a few workaround ideas in mind however that might alleviate the problem

The blunt, dumb but likely effective solution would be to grab those billboards in the Remix editor and just shove them really far away from the player

[dusk trail]

Lmao I was thinking just shove them into the body/neck of the player

Far away might also work however I'd be concerned that about any logic attacked to them having a stroke due to float point errors since UE didn't really have origin rebasing until UE4

[modern isle]

Keep in mind that any changes made with RTX Remix aren't touching anything engine wise. If you have to open up UEd to make the modification on the game then you might as well just scale the quad to 0.

or remove the passes wholesale but that's probably not that trivial if you're not familiar with rendering pipelines :P

[dusk trail]

Probably best bet yeah, just scale it to 0, that's if we can't find a fix with Remix

[dry shoal]

I can find someone who is dw

These aren't visible in Unrealed btw

It's probably the game storing layers of stuff to reference which aren't normally visible outside of remix

[dry shoal]

Like rune mentioned, Remix is its own renderer separate to the game engine. It's 64bit so floating point errors aren't really a concern. Shoving them X distance below the player would probably work fine

[modern isle]

I hastily recreated some of the panels with heightmaps and threw together a texture real quick. Note how I used much less scratches and instead opted for dirt and dust in addition to a lot more variation in the roughness texture. The key is laying several subtle layers on top of each other to give some variation, especially in crevices and on edges.

[modern isle]

I swear if you just turn around and immediately ask me 🤣

[dry shoal]

Lol one of the modders in the RC community is very intimately familiar with rendering stuff. Made his own DX11 renderer replacement for another game so

[dry shoal]

Rune strikes again

[modern isle]

doesn't necessarily mean that vertex position is stored as 64bit btw :) It just means the application CAN use 64bit, doesn't mean it does for everything. And there are memory and performance differences in how you store such data.

[dry shoal]

Wouldn't it have to be stored at 64 bit if you override the position through remix

[modern isle]

que?

explain

[dry shoal]

Because it takes the game's vertex positions and does a translation on them

So it doesn't have the game for that exact final placement

[modern isle]

that doesn't mean that remix stores that data as 64 bit though.

[dry shoal]

I see

Well regardless the maps are small enough that it should be a non issue

Like the playable space to the stop of the environment geo is pretty damn far

Wouldn't take even a quarter of that distance to make the billboards not visible by the player

[modern isle]

Yeah UE2 is pretty limited in map size but I'm wagering that remix only stores vertex data in 32bit full precision.

[dry shoal]

That makes sense

[dry shoal]

Btw I need to make a metal flooring that looks almost exactly like this so I might have to bug you in the future 🙂

Seems like the right use-case for doing it in instaMAT fully. Don't have to go through blender

I wanna use real geometry though for arbitrary reasons

Ok maybe not it'll be fine

[modern isle]

You go have fun UV unwrapping and texturing them while I continue with more parallax occlusion mapping :P

those are all original flat surfaces, are you sure you really need the geometry? 😛

[dry shoal]

You make a very strong argument

UV unwrapping sucks

But at least those you won't be seeing at awkward angles where POM breaks. I would make way more use of POM displacement if not for that concern

Small update on what I'm currently looking into

[dry shoal]

As seen here, I've had success retaining the helmet overlay without it breaking the camera! There are still some things to look into however, like improving the model quality or textures since it doesn't show in captures

[dusk trail]

My hours of work, days even, replicated in no time using height maps

[dry shoal]

Heightmaps are black magic

[dusk trail]

I blame AI

[modern isle]

Apologies

[dusk trail]

I use them too

[dusk trail]

All good

Just makes me rethink my life decisions

I joke, I do that a lot, generally when my text is in italics

[dry shoal]

Amazing that a flat ground can have this much depth (old material)

[dusk trail]

Since emotions are near impossible to determine via text alone

[dusk trail]

[dry shoal]

Smart

[dusk trail]

[dry shoal]

[dusk trail]

[dry shoal]

what is the hidden meaning here

[dusk trail]

[dry shoal]

[dusk trail]

[naive robin]

[modern isle]

chat pls

[dry shoal]

Chat is this real?

[dusk trail]

Huh

[dry shoal]

Does he not know the meme

[modern isle]

https://tenor.com/view/shut-up-shut-up-be-quiet-funny-gif-19370091

[dusk trail]

It would appear not brother, Rune does not know of our ways

[modern isle]

Oh I know about the meme, but I don't think emoji spam is an efficient way of communicating 😉

[dusk trail]

Uuh watta wunga

Idk

[dry shoal]

After lots of talking with native RC modders, I found out why the helmet has been fun to figure out. The original game is a clusterfuck with how it does render targets and the helmet pass

[dusk trail]

How so?

[dry shoal]

Cultured

[dry shoal]

Basically the way the game does its own pass for the helmet is very ass backwards

[dusk trail]

I am indeed

[dusk trail]

[dry shoal]

The billboard back there is actually a render target that's always right behind the player. It's a thing in the vanilla game but naturally it's not visible due to whatever properties they gave for it in-engine. But it is there

[dry shoal]

So one of those billboards is actually a realtime feed of the game without the HUD, and Remix is injecting into the final render target

When you tag a texture as UI in Remix, my understanding is all drawcalls that come before it become rasterized. Usually that's fine for UI, except in RC the UI is applied to the helmet at the very end. So some of the HUD textures getting tagged as UI totally break the graphics

[dusk trail]

Wait...so the game is capturing the environment from the player camera and displaying to a render target, and then the HUD is rendering over that and then is being displayed onto the helmet visor mesh inside of it?, did I get that right

[dry shoal]

Bingo!

[dusk trail]

Wha....why...

That is such a a convoluted way to go about doing it

[dry shoal]

When you're in the helmet view, what's in front of you is actually just the render target put into the helmet surface. It's not real 3D itself

[dusk trail]

Why not just display the HUD on the visor and have the helmet be in the 3d space as it would and attach it to the player character head/camera

Then use transparency so you're just looking out of the helmet itself

Might have had issues with near clipping maybe

[dry shoal]

I'm sure there were reasons for that not being viable. For example, the HUD needs to scale with aspect ratios and that would make a hard cutoff. Or the helmet has its own systems going on that need special attention, and keeping it in a different pass allowed for cleaner control

Spitballing ofc

[dusk trail]

That would kinda explain it I suppose

Makes some sense anyways

[dry shoal]

It had to have taken effort to make that work, I doubt they would've done it if the easier way worked better

[dusk trail]

I know UE has pretty much always had issues with doing separate render passes using just the built in render pipeline as well

[dusk trail]

Right

[dry shoal]

Maybe that's why there's a big TV screen following the player lol

[dusk trail]

I've messed with similar 3d helmet things before and between not being able to do separate render passes and issues with near clipping I'm not surprised they went this route

Does make you wonder tho if it would cause 2x render the render cost

Since it's more or less rendering everything twice, 1 from each camera

A render target is basically a camera in of itself

[dry shoal]

Right but the render target at the end uses the previous render target to display on the helmet

So it's not rendering again, it's taking an image that already exists

[dusk trail]

Yeah but it still has to render that image in real time as a render target and then display that in real time to the camera again

And a render target itself is similar to a camera in how it captures the scene

At least that's how it works in modern UE to my understanding

[dry shoal]

But if the render target has nothing but the UI and helmet, the processing load would only increase so much I imagine

[dusk trail]

I haven't used it too much since I've never had luck with resolution being good and having it run well

[dry shoal]

Disclaimer* I'm not very knowledgeable on this, just throwing out ideas that make some sense to me at least

[dusk trail]

I suppose, I guess if it's only rendering that in it's own area or they have some sort of culling to not have it rerender the entire scene around it

[dusk trail]

Right

And I myself only have probably half an understand of how it actually works in depth

[dry shoal]

Oh yeah, definitely doesn't have to re-render it. I feel like that would be counter-intuitive as well. The game is fairly visually advanced for the time period, so a doubling of the overall world rendering would make it probably unplayable

[dusk trail]

Exactly my point

So they must have found a way that either doesn't rerender at all or only renders what it needs and ignores the rest

As you have 2 cameras anyways, the one that captures the scene and the one that is inside the helmet

[dry shoal]

Unless I have a fundamental misunderstanding of how it works, the game world is always rendering through one render target, and it projects that onto the visor via a shader. So at a lower level the camera would almost be in a void of sorts?

But the whole screenspace is covered by that render target of the game world

[dusk trail]

I might be misunderstanding myself

Idk

[dry shoal]

Imagine the helmet mesh is in a void, nothing else there. Now the UI gets drawn onto the helmet, and via a shader they put the scene's render target running in the background onto the visor of the helmet via a shader or something

It's bringing forward something from before

[dusk trail]

Yeah I get that part, but the helmet still has to exist and be captured from somewhere, and the same goes for the camera capturing the scene itself

[dry shoal]

Oh yeah but I mean the camera in a void with only a helmet probably has a miniscule cost. Could be totally wrong though

[dusk trail]

Or at least that's how I'd interpret the part where Matellica says they swap the render targets

[dusk trail]

Thing is to my understanding UE can't just have a void as a separate place while also loading the actual playable world, it can either load one or the other or have them both in the same space but then you wouldn't have an actual void

Not back then anyways

[dry shoal]

Well look at his first two steps. "Copy that to the destination RT texture"

Could they not copy that again later in the chain

Conceptually I mean. Still unsure if that's actually how it works

[dusk trail]

I've done something similar with portals, render target captures the end scene, saves to an RT texture which then a material loads, and then via shaders display that on the portal mesh via said material

Which in this case I'd imagine would've similar but a bit different, camera captures scene and saves to an RT texture which then has another RT texture that another camera has captured which only has the HUD and helmet mesh in view, which is then used in the original RT texture material and the HUD is applied over that via shaders and then all that is applied and rendered to the player camera

I'm probably completely wrong with that, no idea

But that's why I'm saying it's probably 2 cameras, hence my intrigue on possible performance impact

But again I may just be misunderstanding how they actually went and did it

[dry shoal]

went down a huge rabbit hole today after screwing around with the remix runtime, editor, blender, and getting a bit more understanding of how the helmet renders in RC

and boy do I have a surprise

before I was able to rasterize the helmet exclusively. This leads to it foollowing the game's original behavior of stretching to fill the screen and the UI alongside it. Now the helmet is included in the pathtracing by successfully getting it to render in world space. It's also a stable hash, so I threw together a replacement helmet mesh with a PBR material and it fucking worked

first is a vid showing it in practice, right is an old screenshot of when I got it rasterized

screenshot of the replaced helmet mesh

@burnt light the rabbit hole totally ended up being worth it

my findings today have been an absolutely massive breakthrough

now the HUD doesn't stretch and the helmet can have proper materials with a high poly mesh

[dusk trail]

[burnt light]

wow, amazing work

[dusk trail]

Hell yeah

[dry shoal]

also by tagging the helmet's projected "screen" texture as a color emissive in the toolkit, there's a nice bloom effect and at the top you can see the HUD elements are reflecting off of the PBR helmet mesh

if I can get it lined up better with a future helmet mesh, the HUD should always be reflected subtly by the helmet itself

fucking cool

[dry shoal]

@royal sonnet I explored the stuff about the helmet's rendering setup in this game and got some real big progress. If you're at all curious how that went^

no question, just a followup after mentioning it before :)

[dry shoal]

there's more to come lads. Things are starting to come together

many meshes are ready to bring in, and then it's materials baby

first thing will be helping test Kim's PBRify before I get on that though

time to sleep because what the fuck it's 3:40

[dusk trail]

I'd be ready curious to find out exactly how they did it and how to replicate it/do it better for a project of mine I'm working on that also has a 3d helmet mesh

Really cool stuff so far tho

[dry shoal]

guys, I have the most insane idea known to man

if the helmet mesh is stable

then I can do something wacky with it

let me cook

[dusk trail]

Fear

[dry shoal]

[dry shoal]

time for a test run

[dusk trail]

Wait is that actually how the helmet mesh looks like?

[dry shoal]

this is it replaced by my own

but look

pathtraced glass LMAOOOOOOOOOOOOOO

there's reflections and distortion

if I want to try something like this then the helmet has to be closed rather than just a visor

but I could totally do it

[dusk trail]

That kinda how I did it in my current UE5 project, slapped a fully 3d helmet mesh I made onto the camera, does have some issues tho I've yet to fix

[dry shoal]

it has the potential to look really interesting as long as it's not distracting

[dusk trail]

Yeah, I'd probably make the metal thinner, especially on the bottom

It's a bit distracting

[dry shoal]

oh the helmet mesh itself I'm not messing with much yet, I just wanted to test the glass idea

[dusk trail]

And maybe turn down the reflections on the glass down a little if you can, idk

[dry shoal]

none of this is even close to final. Experimentation baby

[dusk trail]

Fair enough

[dry shoal]

the idea is if I make an enclosed helmet then there's nothing to reflect behind you

[dusk trail]

Right

That's how I did mine

Fully closed helmet mesh

Which also meant that the lights on my HUD would light up the inside of the helmet

And I thought that looked really cool

[dry shoal]

I was thinking something like that but probably different in execution

like having a tiny light always active within the visor

a real light source

[dusk trail]

Could look nice

[dry shoal]

'

this will forever be a legendary screenshot in this mod's history

it must be remembered

[dry shoal]

I refuse to let it be lost by time. Like tears, getting wiped off a clone's visor

[dusk trail]

Lmao

[dry shoal]

default blinding light in remix

ngl this looks like it could be really compelling if it wasn't jank as shit and had time put into it

[dusk trail]

Might do

[modern isle]

https://tenor.com/view/thats-no-moon-obi-wan-kenobi-starwars-death-star-gif-10170550

[dry shoal]

there's a lot wrong here but the giant floating orb shadow is peak content

[urban turtle]

Is it 32:9 compatible? (If not pls do🙏)

[dry shoal]

that's the plan. Not 32:9 specifically, but since I was able to get the HUD working in worldspace there's 0 stretching without any native game modificiation. By default that shit goes ridiculously wide lol. In theory as long as I have the right size for the visor glass then it should work at any aspect ratio

there's a lot left to explore with this stuff, but I don't want to tunnelvision on it too hard so I'm probably gonna stop with the experimentation soon. When I come back to it I'll give it my all to make it as seamless as possible

[modern isle]

may I suggest not using distortion on the helmet glass, gives plenty of people nausea as it's similar to a barrel distortion.

[dry shoal]

it's only like that because I fucked up the surface consistency before exporting it

if I design it with the camera FOV in mind then it shouldn't have any distortion for normal gameplay viewing

but it's true that motion sickness is something I need to be careful of if I go this route

[modern isle]

the FOV changes depending on aspect ratio however.

[dry shoal]

if it doesn't work in the end I'll drop the idea as a fun experiment

[dry shoal]

I mean if it's hor+ then it should be the same

[modern isle]

you mean unplayable at ultrawide?

[dry shoal]

no not at all. hor+ means the vertical FOV is the same. If I make the glass wide enough and curved perfectly with that given vertical FOV then it should look consistent across aspect ratios

that's just my guess though. It might be totally fucked up in practice

[dry shoal]

having another one of those "this shit is gonna look so beautiful when it's done" moments

using a CC0 ceiling material for a stand-in

ground is the same as my outdoor, transitions perfectly inside

[dry shoal]

dc17m progress

[static mason]

Ayo model progress!

[dry shoal]

I have lots of assets ready to go in-game. Just most of it is some of the more boring rock type stuff

just a matter of implementing it

[dusk trail]

auto doing all my environmental and prop assets before I can even get started

[royal sonnet]

sweet, glad to hear it! sounds like you have it working really well

[dry shoal]

don't worry, I'm doing the more boring stuff lol

unless you wanted rocks

[dry shoal]

yeah, and I've gotten a little further too. I've been messing around with it since it's temporally unstable being in world-space on a 3d plane. But with a bit of finesse I was able to make it 100% stable with virtually no softness added from DLSS

but since it's in-world with a temporal AA, the low res nature isn't as visible on the helmet UI bits

[dusk trail]

I might do some anyways, had a plugin for blender for good looking rocks that are shaped controlled that I've been wanting to try out

[dry shoal]

I got a plugin for that too, which one?

[royal sonnet]

I believe there's a mode to let a mesh show up in the camera but not cast shadows - it's used for the portal gun + arms in Portal 2. I think it's rtx.playerModelTextures, tho it may be rtx.playerModelBodyTextures

might want to try playing with that

[dry shoal]

thing is I'd be doing it on a replaced mesh. Is that even possible? Or do I just have to tag the texture associated with the parent

[royal sonnet]

I believe if the original draw call was categorized, the replacements inherit the category? not sure. I think it's also possible to manually set those flags on mesh replacement prims, but it's not exposed in the toolkit yet

[dry shoal]

got it, I'll give it a try some time

what's the difference between model and body textures?

functionally I mean

[modern isle]

I could be wrong but I think it was that viewmodel doesn't cast any shadows while body does cast shadows but is invisible to the camera.

[weary radish]

iirc one of the categories (the upper on in the list) defines the origin that is used for the bounding box (defined under the rendering tab - playermodel?) and the other category is a list of textures that are then affected by that (invisible but still casting shadows - within that bounding box)

[dry shoal]

That makes a lot of sense. And sounds like something I need to set up

[dry shoal]

@dusk trail just for reference, have you made any vehicles by any chance

[dusk trail]

I have not, but I know a few different people who could probably if we need some

[dry shoal]

I don't mind giving it a shot myself. The more I learn/the more varied my skillset the better. There was just one thing I was not looking forward to lol

I absolutely want to try making the LAAT though

[dry shoal]

I expect I'll need help with the many hard surfaces in geo1b btw. There's various unique crates, containers, etc

Idk why I replied to that again but oh well lol

[dusk trail]

Waiting for another message of mine to upload, my net here is shit so the images are taking forever

[dry shoal]

No worries

[dusk trail]

One of the dudes I know made these recently, these were him doing them for fun so they are a bit custom with his own take on some of the designs

There we go

[dry shoal]

Jesus fucking Christ almighty

[dusk trail]

Indeed

Also have a Venator another friend of mine made

[dry shoal]

These guys are machines. The hell

[dusk trail]

Are the assets in Remix able to be unpacked from the files once it setup?

[dry shoal]

That droid model is perfection. Only thing is I was hoping to do it myself 🥲

[dusk trail]

Like could someone get the source files after it's been packed

[dusk trail]

I'll have to talk to him to get any of those models, I just know he's done those recently

[dry shoal]

I think they can be pulled from the packaged file since they untentionally use a very open file format. I'd have to double check

[dusk trail]

Look at my discord profile, explains it all

[dry shoal]

Yeah I didn't mean it was a given. Rather if it was possible I'd feel like I'd be an idiot not to use it

[dry shoal]

Yep, I noticed it a while back. Incredibly cool. I haven't seen much of that mod but from the bits I have it seems very high quality in assets

[dusk trail]

Hmm, I ask as that may depend on if I can get those models from some people or not, since they sell some of these and I'm a friend I may be able to get the source files if the models can't be gotten afterwards by the main stream

Since that'd hurt their income

[dry shoal]

Wait a sec

For a model like the droid it'd have to be implemented in the native engine since it's a skinned

So it'd be packaged using UE2's format

[dusk trail]

Thx, to my knowledge we're the largest SW mod around, and we make pretty much everything ourselves from scratch or commission artist to, so we have a lot of model files and such I can use on pretty much anything, we do have some assets lacking quality but they're being redone rn

[dusk trail]

I mean as long as the files are even somewhat hard to get ahold of then they'll probably be willing to let us use them, they're mainly worried about the main stream downloads and such

[dry shoal]

Well the biggest thing is the odds of someone in the market for such a model knowing it's in a specific niche mod is very small on its own

The discover ability would be non-existent. And while the file format is pretty open, it's not as simple as just opening a folder to my knowledge. It has to be extracted and someone would need to dig through the mod's file structure looking for it

If needed I wouldn't mind putting such assets in incredibly obtuse locations lol

[dusk trail]

I'll see what they have to say on it, at the very least I could more then likely convince them to give me a major discount so they feel better with the possible risk and get something out of it

[dry shoal]

That's awesome and good to know. In the early days of this project I would've jumped at the opportunity. I'm more hesitant now since I've been learning so much and I take pride seeing my work in-game. I wouldn't want to go all the way down the easy route even if the meshes are 100x better than what I could make

[dry shoal]

In that case don't sweat it. I have an idea for one that I'd be especially interested in but it's seen only on Kamino at the start

One of the huge republic capital ships is in the background. And Remix is especially well equipped to handle huge tri counts like I mentioned. So it would absolutely do such a model justice in terms of what's used in-game

But it's just an idea. Unless there were 0 obstacles to them sharing it then I wouldnt mind leaving it

[dusk trail]

Honestly I'm probably going to anyways as I have my own UE5 SW RTS thing I've been messing with for fun and I need some models for that so if I do get them and you want to use them I don't mind, they probably won't mind either

[dusk trail]

It's either a Venator or an Accumulator, pretty sure it's the 2nd

[dry shoal]

That would work out great. I'm gonna try and be relatively picky about what I'd use if a lot was available just because I don't want to rely on outside assets too much

But something like a capital ship? Yeah I don't see myself making that lmao

But that's also outside of the scope of my project. I didn't plan on remastering the Kamino intro

If it wasn't too much trouble using such a model then I'd expand the scope of the project to include the Kamino intro

[dusk trail]

Lmao fair

Wait can you load multiple remix projects at once in game?

[dry shoal]

Yes actually lol

[dusk trail]

I was thinking if you made the helmet like you want and made that a separate mod then others could build off that as well as us

[dry shoal]

Oh yeah I was definitely thinking about splitting up parts of the mod. I was already thinking for the helmet in particular that I might need a choice between helmets made for different aspects ratios/FOVs

[dusk trail]

Right right, glad we're thinking alike in that regard

[dry shoal]

Lol in these last few months a common reoccurring thing is people thinking I'm more ignorant/naive than I am. I guess it's better safe than sorry but there's a lot more im thinking about than it might seem from the outside. Not saying you're doing that, just something I thought of and how someone else might have brought up the same thing but delivered very differently

[dusk trail]

I wouldn't say ignorant or naive, more like new to it all and is still learning, which there's nothing wrong with that

If you're not learning something you're doing it wrong

[dry shoal]

Oh sure. But I mean there's been many many instances where people have brought up concerns delivered as harsh reality checks that I already was fully aware of

I've had more than a few assumptions made on how I believed this project would be super easy, that I was trying to avoid native game modding like the plague (from the RC modding community), that me being excited about a result must think i believe its without flaws. Having high ambitions must mean I don't think it'll take time, work, and hours of frustration to make it

I make such a habit of asking why/understanding concepts I don't really need a deeper understanding of as a way to set my expectations more realistically than most starting out in modding lol

[dusk trail]

I mean as someone who's done a lot of modding over the years, a lot of people will get upset when you're not doing things the same way as them, which is really funny to me because altho not always but enough times the people who do things differently are the ones who figure out how to do stuff no one did before, or see an issue and have the knowledge/find those who do and make a tool or something to fix said issues

Saw it happen a lot with the Empire at War modding community

Just something that happens

It's good to be optimistic on what you wanna do and where you want a mod to go but also be prepared when things get hard or even impossible to do

[dry shoal]

Yep, knew this one too. In fact the way I got the helmet working via remix is something thought to be totally impossible via native game modding, even by the best in the community. But the entire time I was exploring it I was told "yeah this is dumb. You're wasting time, it won't work"

[dusk trail]

Honestly since we're here trying to get the most out of Remix, I'd probably do everything you can with it and fall back to native engine modding when it's something Remix just can't do

[dusk trail]

Yeah sounds about right, but that's the cool albeit annoying part is when you do something that couldn't be done before, new tools and new tech means new solution to old issues that were impossible to do before

Everything is worth a try

[dry shoal]

I'm of the same mind. Remix inherently has different limitations from native modding. In the game itself, you have access to about 2GB of RAM total. There was an ambitious remaster project way back that was canned because they hit that cap + faced various levels of instability with the engine. So my idea has been to prioritize Remix for asset changes and fall back to the game whenever it might be needed. Like for character meshes

This also has the side effect of not being very limited by the native modding either. Like a 20K tri character model works fine on its own in UE2, but if I made the whole game use higher poly counts I'd run into significant problems

[dusk trail]

Right

[dry shoal]

That helmet mesh I used to test the other day was 100K tris. Just because I couldn't be assed to optimize it in the moment. And in remix it worked flawlessly

Natively that would've broken the game on its own

[dusk trail]

Yeah UDK was pretty shit in terms of rendering abilities back then

It's come so far

[dry shoal]

Absolutely. At the start of the project I was trying to do things no one else would've wanted before just because the use-case was different. More specifically I was trying to have as little culling as humanly possible. I didn't do it because I thought that was the best way of addressing culling issues, I did it because it would allow me to take a scene capture within Remix that would cover significantly more of the level

[dusk trail]

Makes sense

[dry shoal]

I also looked into disabling vertex blending as it caused a few models to be unstable (non replaceable via remix). Tried baking the lighting on the maps with 0 light sources and everything unlit so I could skip needing to tag all of those textures by hand since they'd break the textures

One that was really important was disabling the light source muzzle flash on the gun firing. No one had any specific ideas on how to do it just because no one would've done that before lol

Lots and lots of stuff like that. I suspect that by the end there will be a not insignificant amount of work done in the game itself to enable Remix to reach it's full potential

[dusk trail]

At least RC is moddable enough to allow that, plenty of games that don't

So you picked the right game in that regard

lol

[dry shoal]

Some in the RC modding community would argue against that till they're blue in the face

[dusk trail]

Tell em to suck an egg

Idk

Lmao

[dry shoal]

But I think I'm in a unique position compared to what they've had to go through. Remix is really amazing in that sense, it does things that needed fan made software to do previously. Like reading certain formats, having to repackage higher res textures that can only work up to a certain res, etc

[dusk trail]

Fr tho until you figure out something new lots of people will be that way, hell even after you do some still will

[dusk trail]

Remix is pretty cool, altho basic in what it can do now, can't wait to see the new features and tech they'll be adding to it going forwards, but until then we should use it to the best that we can

And yeah it really does allow things you couldn't do before because of native game engine limitations

Super cool stuff

[dry shoal]

I've had more than a few convos with a couple of the people working on Remix and it's given me a good idea of what's theoretically possible and what's out of the question. Short version is there are things that will be added that don't exist yet. Not because they can't, but because Remix is in beta ofc

[dusk trail]

Right that's to be expected

[dry shoal]

Localized fog volumes, flashing/scripted lighting, among other things. One interesting idea was making game-engine like scripting within Remix as a whole. So like passing a volume and triggering something that only exists within Remix

[dusk trail]

That'd be really cool

[dry shoal]

Absolutely. There's a ridiculous number of things Remix could be capable of but the scope of implemention is big enough that it's a bit of a controversial topic within the Remix team. Some want to push it that far, some think the amount of resources needed is too much so they should stick to the basic premise (make old games look really cool) with visual features but no more than that

[dusk trail]

Right yeah makes sense, usually the case with these sorts of things, personally I'd love to see Remix being more of a the ultimate mod tool sort of thing but I entirely understand the concern with resource management

[royal sonnet]

One of the categories is for stuff that should be visible to primary rays coming from the camera, but not in reflections or shadows (i.e. the gun+arms model common in first person shooters). The other is for stuff that should be visible in reflections, shadows, or through portals, but not visible for primary rays (i.e. a full body model for the player character in most FPS games)

I don't remember which is which though, and unfortunately they seem to be lacking documentation

[dusk trail]

Altho if it was the go to mod tool it'd definitely give it that long term support and backing from the user base

[modern isle]

True, 32bit can be very limiting and iirc Large Address Awareness doesn't really help with the game. If you'd have to fall back to this, just keep in mind that all you need is the mesh data, textures can be just a simple 32x32 texture or something so you don't fill the address space with something that gets replaced anyway.

[modern isle]

Oh the joys of forward rendering.... it had some advantages, and some serious disadvantages when it came to scaling.

[dusk trail]

Yeah and there's a reason why deferred rendering is the standard now lmao

[modern isle]

Viewmodel is primary rays, Body texture is reflection/shadows. 😉

[dusk trail]

That's a good point, would really only need something as a temp placeholder texture, remix and do the rest later

[modern isle]

clustered deferred might soon be in unreal, so that's exciting.

[modern isle]

ay, as long as the texture has a unique hash, it should suffice for material replacement.

[dusk trail]

Can't wait

[dusk trail]

Ye

[dry shoal]

Yeah that was already my plan. If I'm gonna import a high poly mesh there's 0 point in wasting resources on it pre-replacement

[royal sonnet]

In general you only need a 2x2 with random color info to form a safely unique hash for replacement, but that might make it harder to visually recognize what that asset it supposed to be. though do also keep in mind that games may process vertices on the CPU, and the time it takes them to do that scales linearly with the number of vertices.

[dry shoal]

I wouldn't be surprised if it did do vertices on the CPU. Don't have enough background to say though. I've used a 2x2 texture on the anchor mesh I implemented early on but since one of the edges was out in a void of sorts it was easy to pick out haha

The game performed fine when using the 20K clone model, but I suppose it's a matter of scaling as I add more meshes through the engine

I figure if the only increase in poly count will be from character models that it's few enough in number to not have performance concerns. Just an assumption

[royal sonnet]

Yep - CPUs are faster nowadays, but single core performance isn't actually that much better. So old games written to be entirely single threaded can run into performance bottlenecks pretty easily, even on modern CPUs.

[modern isle]

and back in the day some devs made the wrong assumptions on where tech would go.... case in point: crysis

[dry shoal]

Fun how the remaster got dramatically faster in performance than the original just because it utilizes the CPU much better

[dry shoal]

Since the meshes I'd be importing are skinned meshes, and the CPU is doing the skinning, is that a significant concern? Or is that exactly what you were already referring to

And yeah, I had a CPU bottleneck at one point in my Remix project, on a 7800X3D of all things lol

[royal sonnet]

that's exactly what I mean.

In order to apply skinning, the game has to do some calculations on every single vertex in the mesh, every frame. If it does that on the CPU in a single threaded game, that means doubling the number of vertices also doubles how long it takes to do that skinning.

[dry shoal]

I want to say UE2 should handle it, but that's something I have to keep in mind. In this case the original mesh was 3K tris, with a 20K tri replacement. 5 or 6 of these character models can easily be seen at the same time. I don't recall it causing CPU perf to be hit but it's been a while since I had it in-game

[royal sonnet]

The only way to know for sure will be to try it out

[dry shoal]

Is there any recommended ways of working around it? For example, cutting the density of other in-game meshes?

Or would that not help the CPU

[royal sonnet]

keeping in mind that Remix also has to visit each of those vertices for hashing / bounding box calculations too

[royal sonnet]

well if you're going to replace them via remix anyways, cutting down existing meshes could free up some things. You can also push the game down to lower quality levels.

It'd be best if you can get the game to just do the skinning on the GPU, so you can apply the mesh replacement via remix

[dry shoal]

True but I'm sure 20k wouldn't be an issue for remix specifically. Otherwise that would break down pretty quick for ambitious mods I would think. Unless there's a different way of handling those meshes due to being CPU skinned

[dry shoal]

I have access to quite a lot of the game. It's not out of the realm of possibility I could use GPU skinning but I didn't find anything on it in my brief researching a while back

Google search also isn't very helpful because unless you make hyper specific quoted queries + exclude terms then you get nonstop UE4 stuff

[royal sonnet]

yeah, I don't really know much about unreal, especially back in the UE 2 era. It may not have support for GPU skinning build into it at all. It's entirely possible they jumped straight from CPU skinning to vertex shaders.

[dry shoal]

I checked old UE2 documentation and this seems like it might be relevant. Curious how this reads to you because I don't have the background to pick up something from it

Skeletal animation gives your characters smoother movement than was possible with vertex animation, but they must be designed and prepared properly to get the desired results. Generally, polygon counts can increase, because memory usage does not balloon as it does with vertex animations, and rendering power becomes the new bottleneck. The difficult part is that the artists no longer have frame by frame, vertex by vertex control of their animations. Each vertex can only be animated by the bones around it, which means that much more planning must go into each model, lest you find that your model isn't capable of certain motions because of a poorly considered skeleton.```

Copy pasted for easier reading

[modern isle]

things like that are hard to benchmark in isolation unless you have something like renderdoc or NSight (actually, does nsight even work with vulkan?)

[dry shoal]

I mean I do have a level editor on hand. Maybe I make a basic cube level and fill it with tons of the same high poly mesh actors?

[royal sonnet]

that's just describing the transition towards using bones and skin weights, instead of directly animating each individual vertex. the technique can be applied via GPU skinning or CPU skinning

[dry shoal]

Damn, understood. Thanks for the translation

[dusk trail]

Do you think a native mod that strippes almost everything down to it's basic stuff and using remix to make it all high quality would be worth the possible time sink?

I'm just wondering what the best course of action would be to get the most out of both native RC and Remix working together

[dry shoal]

I think it only would be if it can't be done otherwise. We could totally do it

[dusk trail]

I guess it just really depends on how much native can handle (even with Remix on top), if it can't then may have to

[dry shoal]

At least, I think what I'm going for should be doable.vespecially since things like the droid models should be much lower in tri count due to how angular it is

[dusk trail]

I suppose

What's the target tri for the updated character models?

[dry shoal]

Yeah I've never gotten a concrete answer for skinned meshes. In the RC modding community they said 30K is the upper limit for what's relatively easy to import

[dusk trail]

Hmm

Is that the total or just per mesh?

[dry shoal]

But I think that was for static meshes only. I'd have to ask but I already left them a question about possible switching to GPU skinning

Per mesh

[dusk trail]

Ah

[dry shoal]

My understanding is on the GPU side UE2 scales pretty well. With geometry there aren't hard limitations due to the technology. But in this instance we're taking CPU performance

[dusk trail]

Yeah I'd say find out if we can do GPU skinning for the skeletal meshes, otherwise that's probably gonna be one of our strongest issues to deal with

[dusk trail]

Right

[dry shoal]

Only one way to find out, and that's by trying. I was gonna redo my clone mesh because it was a piece of shit before (super inefficient topology, awkward edges here and there bc I had only just started learning blender). Once I get that done I can maybe create a new level that just has a bunch of clones running around and see how the CPU scales

[dusk trail]

Sounds good, let me know the results