I have been working on my project for a while now and started bringing all my 3D models into UE5. However, I have noticed that when I export my textures from Substance Painter with the UE4 (packed) settings and import them into UE5, the textures appear washed out and don’t look the same as they do in the IRAY preview in Substance Painter.
To address this issue, I began searching for solutions and came across a video by William Faucher.
Although the video was helpful, it seemed a little outdated, and the suggested solutions didn’t work well anymore. I suspect that Lumen, the dynamic lighting system in UE5, might be part of the problem, but I’m still unsure.
Digging deeper into the matter, I started learning about color profiles and ACES (Academy Color Encoding System) over the course of two weeks. Now, I have a better understanding of the subject and have set up Substance Painter to use OCIO (OpenColorIO) with the ACES 1.2 configuration. I also learned about using EXR textures instead of formats like PNG, JPG, and TIFF. Accordingly, I set up my textures as 32-bit EXRs, with L16F for height and L16 for roughness, and so on.
Everything seemed to work great when I imported these textures into UE5.2, starting with an empty level containing just one building with UDIMs 1001/1004 for each. I adjusted the texture settings for each setting to HDRcompressed, enabled linear encoding override, and set the color space to ACEScg. This made the textures look much closer to how they appeared in Substance Painter.
However, I encountered a problem where the performance suffered significantly. When I moved closer to the building, which has less than 7,000 polygons, the frame rate dropped drastically to 16 FPS or lower for a second.
I am currently stuck and seeking suggestions. I suspect that the use of EXR32 textures might be causing the performance issue due to their large file size. Is there a way to achieve a balance between the desired appearance in Substance Painter and performance optimization? I am relatively new to this field and still in the learning process, so any ideas or feedback would be greatly appreciated.