Automatically increase Light3D shadow bias when shadow blur is above 1.0

[Calinou]

This makes it possible to adjust shadow blur without having to tweak bias to avoid shadow acne. The formula automatically increases shadow bias, normal bias and transmittance bias when shadow blur is above 1.0:

shadow_blur_bias_factor = max(1, 0.5 + shadow_blur * 0.5);

At the highest shadow blur value allowed in the inspector (8), the shadow bias and normal bias will each be multiplied by 5.5.

Automatic adjustment is not performed for PCSS-style shadow blur (light size > 0) yet.

This partially addresses #53534.

TODO

• Leave shadow bias unaffected when shadow quality is set to Hard, since constant blur will not affect the shadow's appearance with this setting.

Preview

Note: The bias values used in the demo below are from #55757.

DirectionalLight3D

Visible lines between PSSM shadow splits will be attenuated if #48776 is merged.

Before

shadow_blur_directional_old2.mp4

After

shadow_blur_directional_new.mp4

OmniLight3D

Before

shadow_blur_omni_old2.mp4

After

shadow_blur_omni_new.mp4

SpotLight3D

Before

shadow_blur_spot_old2.mp4

After

shadow_blur_spot_new.mp4

[Ansraer]

Would be nice if this was somehow displayed in the inspector. Right now user input is just ignored if it is below the min value.

[mrjustaguy]

Does this change affect the total shadow blur (so after Quality blur multiplier is applied) or only the setting blur?

It should affect the total because higher quality softness tend to create more shadow acne because of their internal higher blur multipliers.

[Calinou]

Would be nice if this was somehow displayed in the inspector. Right now user input is just ignored if it is below the min value.

This acts as a multiplier, so I don't see how the value can be ignored (regardless of the blur setting you're using). The multiplier itself is an implementation detail (like the shadow Z near/far ranges), so I didn't mention it in the class reference.

Does this change affect the total shadow blur (so after Quality blur multiplier is applied) or only the setting blur?

It should affect the total because higher quality softness tend to create more shadow acne because of their internal higher blur multipliers.

Right now, this PR doesn't take the shadow quality blur factor into account. This should be feasible, but I think the base automatic adjustment will work well enough. I fear that further increasing the shadow bias factor automatically will result in too much peter-panning at high shadow quality settings.

[Calinou]

Superseded by #68339, which is a more robust solution to the original problem.