Does Photo AI have "image sequence" capabilities?

I have found several threads on batch processing, but this is different from “image sequence” processing insofar an image sequence (derived from video footage) can contain thousands, tens of thousands, even hundreds of thousands of images.

Before I install to test Photo AI, does anyone know about this?

It will not make a video of your image sequence if that might be your question. :eyes:

no, that´s not my question.
i just want to know if photo ai can be commanded to work on a folder with so many images in it in a reasonable workflow manner (and without crashing ofc)

The largest amount of images I batch processed in one go were 430 jpeg images in 3840 * 2160 size and that worked very well.

that translates to about 7.1 seconds of 60fps video footage :slight_smile:
i´m merely asking this, because as of now, i´m working on footage which video ai cannot meaningfully improve, but which, so i hope, photo ai could…

are you sure video ai can’t improve it properly? maybe upload a file so we could give it a try!

well, windows defender says “virus found” and then deletes the file before I can even open it. :eyes:

mhmokay? it`s a zip file with a .mov prores movie clip of 10seconds in it…

anyway, i put it on cloud webspace. this should work.
download

1 Like

at least you can get rid of the strong noise: 10sec_1_iris.mp4 - Google Drive

yes, sure… but everything else is in shambles… now i´ve tried out taking a partial screenshot from photo ai, and here´s the result…


so the face recovery is much better than in video ai

yes, but improving a still image is different from a motion picture. what looks good on a still image might not look great in motion. not to mention the immense amount of time photo ai would need to process all images of a movie file. the source material lacks of so much details that even video ai can’t do wonders here i am afraid.

(none of which are actual reasons not to give it a shot… it can´t be worse than what videoai produces)

edit:
some good news. with the help of hybrid (via degrain,denoise,deblock) i managed to prepare the footage in a manner that allows proteus and iris1 to produce something clean. not great, but quite alright.