Development versions (2.12)

[AluminumHaste]

5 minutes ago, stgatilov said:

Well, the previous dev build introduced a bug with noshadows lights. Their lighting can be optimized away in some cases, for faster performance and wrong results

The upcoming fix will probably reduce performance a bit back, although it seems to be rather small.

Yeah a little it seems.

Down about 10 FPS.

[AluminumHaste]

4 minutes ago, stgatilov said:

Maybe this: https://bugs.thedarkmod.com/view.php?id=6244 ?

If yes, please leave a comment there about how much it breaks existing mission.
Such circumstances greatly raise significance of an issue, at least in my eyes

I just opened that bug tracker

There aren't many missions that require player to swim against a current, I'll have to check some to see if any of them use forcefield.

[snatcher]

4 hours ago, Wellingtoncrab said:

Now we can tank performance with even more elaborate and unoptimized scenes.

[MirceaKitsune]

A reminder: Was playing with graphics settings again and r_shadowMapCullFront seems to give me an extra 10 FPS! Is there still a reason why it wouldn't be on by default for everyone?

r_shadowMapAlphaTested doesn't seem to have any performance impact for me. Was it needed to get shadows from surfaces with alpha channels, like having trees with leafs cast shadows? Then again those don't work since I think they're still "noshadows" by default... wonder if alpha materials could be given a special mode to support shadow with it.

Edited June 29, 2023 by MirceaKitsune

[nbohr1more]

16 minutes ago, MirceaKitsune said:

A reminder: Was playing with graphics settings again and r_shadowMapCullFront seems to give me an extra 10 FPS! Is there still a reason why it wouldn't be on by default for everyone?

It produces hairline light leaks at some recessed shadow boundaries.

Jonri was helping me create an erode shader to improve that but the result caused shadow acne because the shader was working in texture space. Needs to be screen space as I can grok.

[stgatilov]

2 hours ago, nbohr1more said:

It produces hairline light leaks at some recessed shadow boundaries.

Jonri was helping me create an erode shader to improve that but the result caused shadow acne because the shader was working in texture space. Needs to be screen space as I can grok.

A known appraoch is to do screen-space raytracing from rendered fragments as additional check.
But it is quite costly this would return some shadow boundaries to hard shadows... and also I'm sure it will bring even more artefacts as screen-space technique.

[nbohr1more]

yeah, i guess its all a game of tradeoffs.

right now, cullfront cures some selfshadow bugs on statues (and AI) so its probably preferable...

but those light leaks really annoy you once you start to notice them. tough call

[MirceaKitsune]

56 minutes ago, nbohr1more said:

yeah, i guess its all a game of tradeoffs.

right now, cullfront cures some selfshadow bugs on statues (and AI) so its probably preferable...

but those light leaks really annoy you once you start to notice them. tough call

Since they both improve performance and fix other edge cases, seems enabling them might be better than not. I already did so on my end... don't mind a few inaccurate edges even if I wish those didn't happen either, happy for the extra FPS plus fixing self-shadowing issues I have also noticed now that I think about it.

Actually I'm curious about something: What if we gave that option 3 modes, so only front faces or backfaces or both cast shadows? I remember once reading that in some circumstances, you can get better results by only computing shadows out of backfacing given they're closer to what's casting them and you don't need shadowing inside an enclosed model.

[MirceaKitsune]

2 hours ago, AluminumHaste said:

Can you go into the game and type 'getviewpos' into the console and paste the output here?

-5135.74 -1227.59 -215.75   89.0 60.2 0.0

Stand on the pillar and look down at its top while crouching. Oh, and enable g_showPlayerShadow since I think the effect shows in a player shadowed spot... if relevant I have shadow map quality set to Low (6 samples).

As is no surprise, I can confirm alpha textures such as leafs on trees produce the same sharpness as geometry without MSAA being able to address it. Makes me feel less bad about disabling AA entirely, given it's not just a performance drag but also a partial solution that fails to address various cases of sharp edges.

[AluminumHaste]

Ok yeah it looks like just the normal map has no mipmaps generated for it.

GIMP lists only 1 layer, whereas the diffuse texture lists 11 layers. So when you're far enough away you'll get shimmering and artifacts. The normal map is also a TGA file, so it's uncompressed and there's no embedded mipmap support.

Would have to convert it to DDS and generate mipmaps, which is probably a good idea anyways.

[MirceaKitsune]

1 hour ago, AluminumHaste said:

Ok yeah it looks like just the normal map has no mipmaps generated for it.

GIMP lists only 1 layer, whereas the diffuse texture lists 11 layers. So when you're far enough away you'll get shimmering and artifacts. The normal map is also a TGA file, so it's uncompressed and there's no embedded mipmap support.

Would have to convert it to DDS and generate mipmaps, which is probably a good idea anyways.

Is that a vanilla texture, or should I post about this on the FM thread? I think I saw it before so I presumed it's default.

Didn't know mipmaps were still so important: I always thought they're intended as LOD or for progressive loading. Good to know since if I made any textures I'd have likely used tga or png, still had this stereotype that dds is a more proprietary format though it works with Gimp and even the compression modes were open-sourced at some point.

[AluminumHaste]

Yeah it's vanilla

[Wellingtoncrab]

2 hours ago, AluminumHaste said:

Ok yeah it looks like just the normal map has no mipmaps generated for it.

Unless it is has changed recently practically all normal maps in the game are uncompressed as dds compression causes visual artifacts on them.

Though I imagine how noticeable this is somewhat dependent on the asset and method of compression.

[AluminumHaste]

Only really at lower resolutions. After 1024* the blockyness becomes hardly noticeable

[datiswous]

You guys are just going to ignore the topic move? No respect for mods?

[AluminumHaste]

Huh?

[datiswous]

 

[MirceaKitsune]

Sorry! The forum didn't make the move obvious... glad it happened since yeah this branched into its own discussion. Please move my last posts on it there as well.

[datiswous]

Hi, did you see this post?

I thought I mention it for a possible fix.

[wesp5]

1 hour ago, datiswous said:

Hi, did you see this post?

I thought I mention it for a possible fix.

I removed this in my latest Unofficial Patch so it should be easy to do this in the core game as well.

Edited July 7, 2023 by wesp5

[stgatilov]

dev16818-10434 is available.

As for renaming Stealth Score, I see lengthy discussion but no resolution yet...

[wesp5]

38 minutes ago, stgatilov said:

dev16818-10434 is available.

As for renaming Stealth Score, I see lengthy discussion but no resolution yet...

I found a very simple solution for my patch ;): Stealth Score (lower is better):

[snatcher]

6 hours ago, stgatilov said:

As for renaming Stealth Score, I see lengthy discussion but no resolution yet...

Try finding members supporting a positive score (vs negative) and more importantly, let's see the arguments.

The total score value must be negative. Such tiny change illustrates on its own a "penalty" vs a "positive who knows what". The only remaining decision is what happens when the score is zero. In this case zero is better replaced with a "Perfect", "Ghost", "Master"...

As for the name, "Stealth Score" is fine and it's secondary to me and I am neutral.

[MirceaKitsune]

I actually thought of the stealth score label myself: Score implies achieving something positive, but getting caught isn't something to be desired... a more correct name would seem like "stealth penalty". Having it decrease instead of increasing might not be possible though since you can trigger any number of alerts, there isn't a standard amount you can relate to and subtract from so you can come up with a percentage, like for loot where you know the maximum amount on the map.

[datiswous]

On 7/15/2023 at 1:46 PM, stgatilov said:

dev16818-10434 is available.

As for renaming Stealth Score, I see lengthy discussion but no resolution yet...

I was only talking about my specific post and it does include a solution. It's a minor thing though.

On 5/24/2023 at 11:38 AM, datiswous said:

I just tested and you first get a mission succes screen, where you can click on the Mission Statistic screen.

But if you click on it, the button stays there, but it could be easilly removed from the statistics screen.

By adding code:

			set "statistics::visible" "0";
			set "statisticsH::visible" "0";

to windowDef statistics_select under onAction

Edited July 17, 2023 by datiswous