Missing Light Probes, Cube Maps, or both.

You should have at least one light probe covering any space that might contain a dynamic object (e.g. anywhere you're allowing a man-hack to go, etc).

I find t's best to cover most of the map with some low density probes first (maybe even just a single extremely low-density probe, aka low "resolution"), then add default density probes to each area as desired. (You can also use "env_combined_light_probe_volume" at this stage, since this also adds cube-mapping and has similar placement and workflow)

And In case you're wondering:

Light probes are really just a way of telling the compiler what density to "voxelize" the world...(this affects baked fog, and the the accuracy of dynamically-lit objects as they move throughout a baked scene)...and...as far as I know, higher resolution probes always have precedence...meaning: every region of space is voxelized only once, and always to the highest density that was called for in that particular space.

Edit:

It's also worth noting that the compiler is excellent at ignoring space that shouldn't be voxelized, so in some cases its easier to place some default res areas that cover a very large area in additional to your catch-all.

(this is usually "good enough"...Just don't waste higher density in open air as at best that's wasted compile time).

Publicado originalmente por ...BadLucky:

Missing Light Probes, Cube Maps, or both.

You should have at least one light probe covering any space that might contain a dynamic object (e.g. anywhere you're allowing a man-hack to go, etc).

I find t's best to cover most of the map with some low density probes first (maybe even just a single extremely low-density probe, aka low "resolution"), then add default density probes to each area as desired. (You can also use "env_combined_light_probe_volume" at this stage, since this also adds cube-mapping and has similar placement and workflow)

And In case you're wondering:

Light probes are really just a way of telling the compiler what density to "voxelize" the world...(this affects baked fog, and the the accuracy of dynamically-lit objects as they move throughout a baked scene)...and...as far as I know, higher resolution probes always have precedence...meaning: every region of space is voxelized only once, and always to the highest density that was called for in that particular space.

Edit:

It's also worth noting that the compiler is excellent at ignoring space that shouldn't be voxelized, so in some cases its easier to place some default res areas that cover a very large area in additional to your catch-all.

(this is usually "good enough"...Just don't waste higher density in open air as at best that's wasted compile time).

Thank you so much! It's fixed now :D