The texture I get in Reality Capture is noticeably different from the original photo resolution.
Here’s what I tried:
- Change the resolution to 16k.
- Create multiple 16k textures per model and project all those textures onto the model with the 16k texture.
- Change texel to custom and adaptive.
- Change the gatter.
- Unwrap the model in third-party software.
And in general tried everything I could find in documentation, on forums and open sources.
In the end, I came to the conclusion that Reality Capture has no way to prioritize the creation of textures from certain photos. What I mean by prioritize, when creating a texture, a certain group of images had a greater impact on the projection from the cameras.
Now to the point, why I came to this conclusion. I have 30 cameras of the same model, but the lenses are not the same on all cameras as they are on the others.
26 cameras are 55mm;
4 cameras are 80mm.
On those cameras that are 80mm, the detail is higher because the frame is optically closer. All cameras have identical resolution and settings, only the focal length changes. As far as I understand, the parameter “texture quality” is measured relative to the texture resolution and does not take into account the focal length, and when building a texture the program takes into account only this conditional parameter “texture quality”.
Image showing the difference in quality, the model in the image has a single 16k texture with a “texture quality” of 200%
