The VideoLAN Forums

Hello there

Today I had an idea, and I think it could be nice to contribute the VideoLAN project with this idea.

The idea comes from an issue I actually have.

I bought a fantastic TV SET of 32" Inches, a SAMSUNG LE32B550A5W which is amazing. 1080p resolution in that TV is a dream, Playstation 3 games are amazing there, the PC resolution there is also fantastic, the wallpapers woooow incredible... but... when you are going to reproduce movies... then... you see the defects of the pixels.

That is normal,you are trying to reproduce a movie of a very reduced resolution in an amazing screen of 1080p. If you are accustomed to watch movies in a 19" inches monitor... and now you watch those movies in a 32" inches monitor... in a big screen there is more physical space and normally... you'd see the defects... of the DivX movies you may download from Internet.

Also I played original DVDs... and yes... you see that the TV is so amazing that the only way to find quality is reproducing a BlueRay, Using a computer in 1080p or playstation 3 or xbox games in 1080p. A normal DVD has lower resolution, and always, you see the defects in the picture.

However, I've been thinking that there is available a magic technology, called Fractal Resizing, and seems to be, thru some complex mathematical algorithms... you can increase a picture without much loss in quality.

There are programs that right now are using those technologies for picture resizing...

onOne Genuine Fractals Print Pro 6, is one of them for photoshop. If you go to the website, just google it... you will find examples of resized pictures with amazing quaility... so I thought, why VideoLAN don't implement this technology in the future?

No one did it yet!!!!!!! Neither Microsoft!!! so would be something amazing, a point of reference in the world, and something prestigious

I had the idea today, and I thought it would be nice to share it here. Just imagine that we have a video player that don't produce quality loss in 1080p in movies with lower resolution (even original DVDs), because it's using a revolutionary technology called fractal image improvement. That would be the most of the most!

If someone here is interested to read about the subject, I'd recommend you to visit these links:

http://avaxhome.ws/ebooks/programming_d ... ssion.html
http://avaxhome.ws/ebooks/transform_and_data.html
http://avaxhome.ws/ebooks/journal_of_ma ... ision.html
http://avaxhome.ws/ebooks/the_data_comp ... elson.html

I think it's a subject really interesting for research and development.

Thanks

I don't wanna know what kinda CPU you'd need to run that algorithm real-time on a video That technique might work well with single static images but probably cannot be applied to live video upscaling for desktop end users in the next few years. Image upscaling is a very difficult field and there exist only a hand full of algorithms that produce convincing results. And these are very complex, thus they need much CPU power. Too much for being able to process 24 images per second, which is the least frame rate for smooth video.

Would the core developers be interested in my research team working on fast anti-aliasing CPU/GPU-based image enlargement code for vlc?

For one, I think that we may be able to make enlargements of destructively compressed small videos look better. We have some experience working on multi-core and GPUs, to wit the following article which describes an early version of the methods (which have absolutely nothing to do with fractal methods):

CPU, SMP and GPU implementations of Nohalo level 1, a fast co-convex
antialiasing image resampler, by Nicolas Robidoux, Minglun Gong,
John Cupitt, Adam Turcotte and Kirk Martinez, published in the
Proceedings of the Canadian Conference on Computer Science &
Software Engineering, C3S2E 2009, Montreal, Quebec, Canada, May
19-21, 2009, ACM International Conference Proceedings Series, pages
185-195. http://doi.acm.org/10.1145/1557626.1557657

which is already outdated (our current implementations for all three platforms are now noticeably faster, and we now have slower but smoother and/or more strongly anti-aliasing methods).

A quick description of some of our recent work in the context of GeGL is found here.

http://www.mail-archive.com/gegl-develo ... 00794.html

Nicolas Robidoux
Universite Laurentienne

PS

Of course, saying "yes" just means to us that you may be interested if the result is better than what you have already.
I just want to get an idea of how happy you are with the current quality of vlc video enlargement.

nr

Yes, VLC upscaling and downscaling is far from being the best possible.