[Documentation / Explanation] How can it work? #221

New Issue

claunia · 2026-01-29T21:45:48Z

claunia commented

2026-01-29 21:45:48 +00:00

Originally created by @rubyFeedback on GitHub (Aug 1, 2022).

Would it be possible to add an article explaining how it can work?

Let me explain this question:

Recently some websites pointed at GFPGAN, in part due to a promo-video on youtube
showcasing some examples.

Now - the examples look nice. But when I looked at it, I wondered whether the images
at hand were also "polished" to look smoother or better than the original video, e. g.
if you had a not-so-pretty person's foto that was old, that the digitally restored image
was not only looking better, but perhaps facial features were auto-upgraded or
auto-polished. I have no idea if this is the case or not, which is why I think an article
may be helpful that explains what kind of restorations are done by GFPGAN. People
reported different results - some great results, some not so great results. Right now
I have no idea whether GFPGAN is doing an accurate job or not, which I think
it would be helpful (in english please) to explain how it works, and whether it does
a good job on these images, but not on these images / fotos.

Originally created by @rubyFeedback on GitHub (Aug 1, 2022). Would it be possible to add an article explaining how it can work? Let me explain this question: Recently some websites pointed at GFPGAN, in part due to a promo-video on youtube showcasing some examples. Now - the examples look nice. But when I looked at it, I wondered whether the images at hand were also "polished" to look smoother or better than the original video, e. g. if you had a not-so-pretty person's foto that was old, that the digitally restored image was not only looking better, but perhaps facial features were auto-upgraded or auto-polished. I have no idea if this is the case or not, which is why I think an article may be helpful that explains what kind of restorations are done by GFPGAN. People reported different results - some great results, some not so great results. Right now I have no idea whether GFPGAN is doing an accurate job or not, which I think it would be helpful (in english please) to explain how it works, and whether it does a good job on these images, but not on these images / fotos.

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: TencentARC/GFPGAN#221