General-purpose texture painting sounds very convenient to have, but in reality it is a tool that requires a sense of balance, as there are many areas that can be improved by what it can do, but can also be replaced by DCC.
Using RE ENGINE, we will answer the question, "How does texture painting work within an in-house engine?"
Is this how all the models work in Resident Evil Village?
Or is the method different?
As explained in the earlier parts of the session, any parameters that can be painted in RE ENGINE can also be set using DCC tools.
so not all things have necessarily been painted in RE ENGINE.
If you can manage it with DCC, including previews, that can be a viable option, and ultimately it boils down to which workflow is more cost efficient.
I have a question about picking.
I think you're getting the depth value and world coordinate position from the ZPass, but how do you get the model information?
It is not retrieved after the Zpass, but stored directly in RWStructuredBuffer at the time the pixel shader is executed.
Therefore, any semantic information that comes into the pixel shader can be stored.
Is the pixel shader for picking drawn in a dedicated pass separate from the geometry drawing?
Yes, it is.
It is implemented in a dedicated pass that is different from the part that draws the graphics.
I think you mentioned that you "had the textures in an intermediate format," does that mean that format is supported by other DCC tools?
The intermediate format is only a volatile binary that is created to temporarily interpret the original png, dds, etc. saved in the asset browser as a common format.
Therefore, when the editing is finished and saved, it is written back to the original image format so that it can be loaded in DCC.