ScaleUp: Innovative Video Resolution Enhancement Software

Seprentes@lemmy.world to Technology@lemmy.world – 147 points –

ScaleUp is a cutting-edge software that enhances video resolution up to 4 times and provides extensive codec support, all thanks to the powerful RealESRGAN AI.

Now, transforming your low-resolution videos into high-quality, crystal-clear images has never been easier! ScaleUp incorporates advanced RealESRGAN AI to significantly improve the clarity, details, and vibrancy of your videos, delivering realistic and more captivating results.

Download link: https://github.com/Seprentes/ScaleUp/releases

25

Remember when everyone was laughing about how in crime shows they “enhance” digital photos?

I would still laugh because you're just as likely to get a completely BS image as you are to get an image that is close to reality

Upscaled photos are essentially fiction that is based on actual events. Any details that are not clear in the original should not be treated as real.

For the sake of, say, improving the visual quality of a movie, this is fine.

For the sake of forensics...aw hellllll no. The first time a prosecutor tries to use AI-enhanced images in court, they'll be absolutely destroyed by a competent defense attorney, and possibly disbarred.

I feel like AI isnt ubiquitous enough for a lawyer using it improperly to be seen as acting in bad faith and disbarred. I don't know, but disbarring seems like a more "fatal" punishment.

You're probably right. I'm sure it would depend on specifics. An extreme case could be considered falsifying evidence, which I think could be grounds for disbarment. In practice it might just be a matter of providing the original photos and clearly presenting any edited images as such.

It still depends on the training data, so the originals importance will only get prominent as better upscaling methods will arise.

They should really name this software "Enhance!"

Some example video?

I just tried. It's basically https://github.com/xinntao/Real-ESRGAN expect OP's program upscales video, frame by frame, and combines those frames into a new video.

Video looks like it's made by AI. Tried on old video - not sure if old or new video looking better. Contours in new video are difderent with each frame lol.

OP did a great job, but overall technology is not there yet. Videos aren't looking good tbh.

Does it configure it's own venv or do I need to to that?

Edit: doesn't look like it. That should be just common practice, Like how oobabooga/SD work. Makes life easier.

What python version?

Edit:

I just downloaded from the releases, giving it a test now

Here we are in 2023 and people still have to ask for venv and which Python versions. I don't understand why these AI devs on these smaller scales seem to be container adverse.

I just downloaded from the releases, giving it a test now

Did the same. :)

It's very ai-upscaley. I think if there is some extra processing steps, then this could be really good.

Quality was bad because ScaleUp had bitrate problem. Now i fixed that. Results now much better.

Ad?

Is it really an ad since the tool is open source and free?

"a notice or announcement in a public medium promoting a product, service, or event or publicizing a job vacancy."

definition by Google

This is an ad

Counter point:

Product (noun) - something that is made to be sold.

The software in question is not for sale, therefore is not a "product", and this post is not an ad.

I feel like if OP had worded it in a more community-oriented manner it wouldn't have read like an ad. Not into it