If possible, adjust Diffusion to generate images with greater precision to the typed command prompts. Because when i enter the prompt, the loaded images come with inaccuracy to the typed text.
Diffusion is not meant to be an accurate rendering engine, at least for now, it’s more for early stages of a project, for proof of concepts or even inspiration, if you want a more accurate rendering engine that uses AI, Veras is the best I know, or for absolutely accurate rendering with all the materials, geometry and assets, V-Ray is one of the best options but there are a lot of rendering engines on the market that can do what you’re wanting Diffusion to do.