perf(pbr): improve glints

[doodlum]

No description provided.

[Pentalimbed]

actually, just put the noise in GlintCachedVars

[doodlum]

actually, just put the noise in GlintCachedVars

It's not specifically made for glints so i dont think it should be in there because it might be reused in the future.

[Pentalimbed]

actually, just put the noise in GlintCachedVars

It's not specifically made for glints so i dont think it should be in there because it might be reused in the future.

Do that when you actually need it in the future.

Edit: nvm

[Pentalimbed]

Change the texture size here

https://github.com/doodlum/skyrim-community-shaders/blob/7be08c17b5c2eb76a2fc28344291513b50f11692/src/TruePBR.cpp#L272

and here

https://github.com/doodlum/skyrim-community-shaders/blob/7be08c17b5c2eb76a2fc28344291513b50f11692/package/Shaders/Common/Glints/Glints2023.hlsli#L225

and see how the appearance and performance trade off (The original code uses 512)

[doodlum]

Change the texture size here

https://github.com/doodlum/skyrim-community-shaders/blob/7be08c17b5c2eb76a2fc28344291513b50f11692/src/TruePBR.cpp#L272

and here

https://github.com/doodlum/skyrim-community-shaders/blob/7be08c17b5c2eb76a2fc28344291513b50f11692/package/Shaders/Common/Glints/Glints2023.hlsli#L225

and see how the appearance and performance trade off (The original code uses 512)

no measurable performance impact but also no visual difference.

[Pentalimbed]

no measurable performance impact but also no visual difference.

Okay then I think it's better if you make it 512

[Pentalimbed]

Additionally, here https://github.com/doodlum/skyrim-community-shaders/blob/7be08c17b5c2eb76a2fc28344291513b50f11692/package/Shaders/Common/Glints/Glints2023.hlsli#L536
you may incerase the clamp value (20) until you see weird glitter lines (the sand beach in front of broken oar grotto is a good location to test)

[doodlum]

no measurable performance impact but also no visual difference.

Okay then I think it's better if you make it 512

I'm not sure because it will matter when you have a lot of lights.

[doodlum]

Additionally, here

https://github.com/doodlum/skyrim-community-shaders/blob/7be08c17b5c2eb76a2fc28344291513b50f11692/package/Shaders/Common/Glints/Glints2023.hlsli#L536

you may incerase the clamp value (20) until you see weird glitter lines (the sand beach in front of broken oar grotto is a good location to test)

That is outside the scope of this PR

[Pentalimbed]

Additionally, here
https://github.com/doodlum/skyrim-community-shaders/blob/7be08c17b5c2eb76a2fc28344291513b50f11692/package/Shaders/Common/Glints/Glints2023.hlsli#L536

you may incerase the clamp value (20) until you see weird glitter lines (the sand beach in front of broken oar grotto is a good location to test)

That is outside the scope of this PR

No? because you stochastized the sampling, you are now temporally averaging the clamped value instead of clamping the average value, which means there's more bias towards the tamer side.

[doodlum]

Additionally, here
https://github.com/doodlum/skyrim-community-shaders/blob/7be08c17b5c2eb76a2fc28344291513b50f11692/package/Shaders/Common/Glints/Glints2023.hlsli#L536

you may incerase the clamp value (20) until you see weird glitter lines (the sand beach in front of broken oar grotto is a good location to test)

That is outside the scope of this PR

No? because you stochastized the sampling, you are now temporally averaging the clamped value instead of clamping the average value, which means there's more bias towards the tamer side.

done

[Pentalimbed]

done

great, also check out the change request