I use Linux and I created a video that’s pretty low resolution, by zooming in. It’s so pixelated that it really isn’t going to work as-is, so I tried to use a few of the AI video upscalers out there–TensorPix and others, but none really helped.
I included one of the frames from the video to show how pixelated it is, so I’m hoping for something that will fix it. Since I’m on Linux, I’m hoping to either find something I can use in the browser, or something that’s compatible with Linux.
Does anyone know of any that would be able to greatly improve the quality and “create previously unseen details” (which is what TensorPix said it would do) to a great extent?
I found it
I love that the AI just decide to remove the white shirt collar completely.
“create previously unseen details”
AI upscalers can’t reveal information “hidden” inside of a digital image. The way they work is they guess and fill in the blanks. I can’t find it but a good example was a blurry image of Obama that was upscaled into a normal looking dude with a tan…
You are going going to need to give way more info. What did you use to make it, what was the original resolution, what is your target resolution
Rule 5. Locking.
I am an IT guy and a hobby photographer.
Lets break it down, there are ways to encode a file that makes it capable to handle some data loss or corruption.
Most file formats does not use these techniques, simply because they tale up more space, and will slow down the processing of the file.
The basic fundamental fact is that no computer algorithm or AI can recreate data that just isn’t there.
You won’t find any secrets or “recovered data”, what you will find are random details placed where the statistically belong according to the datamodel.
