Use the Light3D Indirect Energy property in LightmapGI

[Calinou]

Follow-up to #52252. See also #51705.

This implementation currently only works if the light's bake mode is Dynamic, not Static. I'm not sure how to make it work for static as well, but I documented this limitation nonetheless.

This closes #52257.

Testing project: test_lightmap_normalmap2.zip

[jcostello]

@Calinou the plan is to make it work for static as well, right? BTW static baking is not entirely working right now

[Calinou]

the plan is to make it work for static as well, right?

If someone manages to find a way, yes 🙂

BTW static baking is not entirely working right now

Could you describe what's not working? Here, it works as expected as long as Directional is not enabled.

[jcostello]

Could you describe what's not working? Here, it works as expected as long as Directional is not enabled.

Baking spot lights doesn't work even if directional is not enabled.

[Calinou]

Baking spot lights doesn't work even if directional is not enabled.

Please open a new issue with a minimal reproduction project attached.

[JFonS]

The current changes look good.

I haven't tested it, but adding support for static indirect light may just be a matter of adding the factor multiplication here:

godot/modules/lightmapper_rd/lm_compute.glsl

Line 358 in b611e8b

static_light += light;

Or better yet, multiply directly in the light variable initialization.

[Calinou]

I haven't tested it, but adding support for static indirect light may just be a matter of adding the factor multiplication here:

I've tried this, but unfortunately, it multiplies both the direct and indirect light:

DirectionalLight and OmniLight Indirect Energy set to 2

DirectionalLight and OmniLight Indirect Energy set to 1

DirectionalLight and OmniLight Indirect Energy set to 0

(The red light source is an emissive cube, not a light node.)

[JFonS]

@Calinou Right, that makes sense. In this case, I think we need to have two separate buffers for the direct light pass. The currently used one should stay the same, as it's the final light accumulation buffer, and we should add a new one that stores the "scaled" direct light depending on the light's indirect energy value. This new buffer can then be used as the source buffer for the first bounce and discarded.

[YuriSizov]

If this is a relevant fix for a regression in 4.0, it needs a rebase. cc @clayjohn

[Calinou]

If this is a relevant fix for a regression in 4.0, it needs a rebase. cc @clayjohn

The way this is implemented is incomplete, and it's too difficult for me to do it in a complete way. This can be done for a future 4.0.x release anyway, as this property isn't relied upon very often (at least with lightmaps).

[jcostello]

If this is a relevant fix for a regression in 4.0, it needs a rebase. cc @clayjohn

The way this is implemented is incomplete, and it's too difficult for me to do it in a complete way. This can be done for a future 4.0.x release anyway, as this property isn't relied upon very often (at least with lightmaps).

It is use a lot by artist. Most of the UE4 lighting tutorials I've seen use indirect energy when baking lightmaps

[Calinou]

This was resolved for all light types and bake modes by #82068, so this PR is no longer necessary. Both per-light and global indirect energy controls work as of 4.2, closing.