When the subject of the picture is in low light conditions and has long hair, a noticeable distinction can be observed between the areas where the hair has been restored and those that have not. This distinction appears as an unappealing artifact. Using The face recovery,
I’ve noticed this problem as well, and from many versions of the program: basically it seems that when you turn on face recovery, the program actually enhances ONLY the face, creating a sort of “rectangle” of adjustment on the subject’s face. outside of that box instead everything is left untouched, creating these problems for women’s hair, often!
I’ve confirmed through several tests with different images of women with long hair, that the software restores the hair around the face accurately, but it struggles with the rest of the hair. Could someone kindly update us on whether this issue will be addressed? Thank you for your assistance!
This is expected behavior at the moment as face recovery is fixing the face but there is a sharp border with the rest of the image.
The face recovery is built to work on only the face and part of the neck as the model was not made to fix the rest of the body or the rest of the image. It is a known issue that the boundary is sharp as we are not feathering the face with the rest of the body.
We are working on ways to improve this behavior so that the change in quality is less drastic.
For now I would recommend decreasing the face recovery strength so the face is blended with the rest of the image better.
Let me know if that helps or if you have other questions about this.
Or is it even possible to extend the Face-Algorithm that it enhanced also the complete Hair, Neck etc? And that the „CutPoint“ is the start of the body? That would help so much