Hello Maru,
To help you understand better, here's my workflow.
During my production renders, I create different layers: beauty, Interactive Lightmix, different light, reflection, refraction, Cmasking, etc., which I use during my post-production in Photoshop.
I save all these elements in .png format.
My working base is my Interactive Lightmix layer, which contains the correct light settings I configured in the VFB.
When I use Project Dream or Magnific AI to rework and improve my characters, I use the most balanced, contrast-rich image, i.e., the Interactive Lightmix layer or a good-quality post-production layer.
In the tests I did with AI Enhancer, the image created by Corona and used in EXR format was not of good quality. It's overexposed, like the beauty layer usually created in my renders... so the result with the AI Heanser isn't really good quality and isn't usable in my workflow.
as an example two images, quick rendered, the interactive layer that I want to use for AI enhancer and the result of the AI enhancer live from the VFB via the EXR format created by corona
