Bright light vs Colored lights additive math reality vs game play-ability

[nephranka]

I agree it's not a bug, but I'd love to see it work the way Silent Ruin suggests. I gave up on using colored lights long ago because they never get seen by the players who invariably have something that will make them show up white.

Same here!

[pindercarl]

light_blend.jpg

Okay. You've convinced me to take another look at it. I may have a solution that meets your needs. It preserves the color in overlapping light areas while ensuring that the light value is represented in at least one color channel. E.g. when full intensity white overlaps with full intensity red, the result is RGB -> 1.0, 0.5, 0.5. (light pink). In the attached image you can see the same lighting in the current version (left) and the new method (right). I'll chat with John about it next week. It would likely be a per image option with existing images set to the old method and new images potentially set to the new method.

[rocketvaultgames]

light_blend.jpg

Okay. You've convinced me to take another look at it. I may have a solution that meets your needs. It preserves the color in overlapping light areas while ensuring that the light value is represented in at least one color channel. E.g. when full intensity white overlaps with full intensity red, the result is RGB -> 1.0, 0.5, 0.5. (light pink). In the attached image you can see the same lighting in the current version (left) and the new method (right). I'll chat with John about it next week. It would likely be a per image option with existing images set to the old method and new images potentially set to the new method.

This looks promising!

[SilentRuin]

light_blend.jpg

Okay. You've convinced me to take another look at it. I may have a solution that meets your needs. It preserves the color in overlapping light areas while ensuring that the light value is represented in at least one color channel. E.g. when full intensity white overlaps with full intensity red, the result is RGB -> 1.0, 0.5, 0.5. (light pink). In the attached image you can see the same lighting in the current version (left) and the new method (right). I'll chat with John about it next week. It would likely be a per image option with existing images set to the old method and new images potentially set to the new method.

That looks...

AMAZING!

Thank you for reconsidering as I know this was brought up by others in the past.

I'd even take it as optional (as I'd always have that new way option ON).

[MrDDT]

Awesome work! Looks good, I think this is pretty much the best of both worlds without messing up either.

[MrDDT]

light_blend.jpg

Okay. You've convinced me to take another look at it. I may have a solution that meets your needs. It preserves the color in overlapping light areas while ensuring that the light value is represented in at least one color channel. E.g. when full intensity white overlaps with full intensity red, the result is RGB -> 1.0, 0.5, 0.5. (light pink). In the attached image you can see the same lighting in the current version (left) and the new method (right). I'll chat with John about it next week. It would likely be a per image option with existing images set to the old method and new images potentially set to the new method.

What about dim colored lighting, with bright white lighting? How would that play out? Will we be able to see some of the color from the dim?

[pindercarl]

What about dim colored lighting, with bright white lighting? How would that play out? Will we be able to see some of the color from the dim?

light_blend_half.jpg

Same lighting as previous example, but the colored lights are at half intensity.

[HywelPhillips]

That's certainly an improvement, although it has got some unphysical features still (the arc of paler light through the white light source, for example).

The ideal would still be to do the math in a log space or LAB or HSL colour model to avoid this immediate clipping to white. Maybe even just trying doing the calculations with an inverse gamma transform, then applying the gamma transform again?

https://gmshaders.com/tutorials/tips_and_tricks/

//Useful on textures because they are already gamma-encoded.
vec3 decode = pow(color.rgb, vec3(0.4545));
//Useful with linear gradients or lighting. It should be the last step.
vec3 encode = pow(color.rgb, vec3(2.2));

But that does mean doing computations in float rather than integer which may not possible the way FGU is set up to render, I don't know. I thought Unity always used float for shaders (https://docs.unity3d.com/2020.1/Docu...rformance.html ) but of course I've no idea how you are actually doing the processing.

[SilentRuin]

light_blend_half.jpg

Same lighting as previous example, but the colored lights are at half intensity.

I'm completely satisfied as this solves all the issues I presented (bright white light wipes out all other colored lights, dim white light wipes out all other dim colored lights).

Look forward to update.

[mlbrown]

NVM I was replying to an earlier comment, and see that this is being addressed to correctly show light sources.