preview with "Face swap frames" checked vs rendered image via the "Start" button

[pablo-888]

Is there a difference in quality between the two when using a single target image with only 1 face in it?. For example if I upload multiple faces (Source Face Image), and I check the "face swap frames" option I can quickly select the source face I want and the preview panel gives me a fully rendered image. If I were to not have that option checked and use the normal orange Start button, I would only be able to use the first face in the source gallery (unless there is a way and i'm just missing it).

I compared the two final images and they look identical. Are there any drawbacks to just doing it the first way and skipping the Start button entirely?.

[C0untFloyd]

Are there any drawbacks to just doing it the first way and skipping the Start button entirely?.

No, it's identical.

If I were to not have that option checked and use the normal orange Start button, I would only be able to use the first face in the source gallery (unless there is a way and i'm just missing it).

Of course there is a way to use multiple faces, but this is only true for the 'Selected' swapping mode. If you have 4 input faces you can map them to 4 target faces. Please take a look at the wiki page https://github.com/C0untFloyd/roop-unleashed/wiki/Face-Swap-Tab

[Fikka7]

In the previous version, one could add faces by dropping files one after the other in the 'source files' drag & drop area. In the current version, one can only add multiple faces the first time the drag & drag operation is performed. Subsequent drag and drop don't work and open the file in a new tab on the browser.

This is not a suitable behavior in my opinion since it makes ordering of faces particularly challenging. I would request to revert this behavior as in previous version. Thanks.

[C0untFloyd]

Wrong, you can click on the x in the upper right corner to clear it and add another batch of images just like before. Yes I know, some kind of manual is needed. Ordering of faces would be easier to do with the target faces btw.

[lysxelapsed]

I miss the old behavior too to be honest, where you could add source faces one by one by drag & drop. Is there a downside to re-implement that? Yes, adding source faces first and assigning the target faces after would work. Just gotten used (and made more sense for me) to the other way around like Fikka7.

[C0untFloyd]

Guess what, I liked the previous one better too 😄
However the Image Component I used didn't support multiple files nor my custom facesets. That's why I had to change to the more general files box.

[Fikka7]

Just a suggestion but can the clear function of the Source images be triggered pragmatically and executed once drag and drop is done. Having that file in the list from my understanding doesn't serve any purpose as it is since the face gets added already.