To sharpen video in the past, I have had to resort to exporting the entire video out as still images, then running them through Sharpen AI. I then re-assemble back in to a video.
I had some success with Video 3.x sharpening one video, but another I am trying this weekend does zero – even with the same settings. I can’t even tell that there’s a pixel different in original vs preview screens.
I am trying to do the Sharpen AI de-blur to video – is there some method to do this?
I’ve also had endless problems over the past day trying to import a batch of video frames. Even 786 images causes Sharpen to hang for ours when I try to apply settings to all videos. Thus, that is no longer an option unless I want to manually do about 100 or so frames each pass. (With video shot at 60fps, each pass is about 2.5 seconds of video so I’d be doing this for weeks for a seven minute video.)
Hoping there’s some secret in the new or old Video to help.
You can try the Theia model. Right now it’s broken so only the sharpen setting works—but it’s tied to the Reduce Noise slider!
Give it a try and see if it will do anything like what you’re hoping for.
Topaz is working on a motion deblur option, but it is still in a very early alpha state: Reduce motion blur in video option? - #5 by Imo
I just gave that a shot, and 100% Reduce Noise in both Theia Fine Tune Fidelity and Fine Tune Detail don’t change the image any noticeable amount – just looks the same on both sides.
Welp, that’s the only suggestion I’ve got. Sorry it didn’t work.
Back on December 9th, whatever version of Video AI was available allowed me to sharpen video using these settings. The current version doesn’t seem to noticeably change the video with these same settings.
I have now tested this on the current Mac version and one under Windows 11.
Still photos processed through Photo AI are wonderful, but it doesn’t allow batching more than a few hundred photos at a time before it crashes (also tested on both Mac and Windows 11).
Is there a CLI for Photo AI? I see Video AI has one.
VEAI is not good at this.
But there is a way, similar but better, to what you have been doing by now: Convert the video to an image sequence (or output it as such from VEAI) and then do a batch processing job in a deconvolution software: Blind deconvolution is the best way to sharpen an out-of-focus blurry image.
A good software to do this is AutoQuantX: https://www.mediacy.com/autoquantx3
A much more affordable solution though is SmartDeblur: SmartDeblur - Buy
When the batch job is done you can reassemble the image sequence, e.g. by using ffmeg or by sending it through another round in VEAI.
I will look in to that. Part 1 of the video I have been working on completed today, but it’s very time consuming even on an M2 Mac.