3 ms·

I'm interested in a 'super lossy' (deep learning based?) compression. You should be able to compress movies down to a screenplay and a few stage directions. A

I'm interested in a 'super lossy' (deep learning based?) compression. You should be able to compress movies down to a screenplay and a few stage directions. A picture should of course be worth 1000 words.

lfowles 10y ago
Google Photos already has an incredibly lossy compression, right? Images organized under Dog, Cat, etc. :) Might take a bit more effort to get the extra 999 words.
mcbits 10y ago
I started something like this long ago but didn't get very far with it. This was before "deep learning" and I was groping in the dark, but I think the concept is sound up to a point. The idea was to train a neural net and build up a database of features (maybe on the order of 1-10 GB, or whatever is just small enough to ship) to estimate the missing details from downscaled and extremely over-compressed JPEGs. If it worked, I think it would also improve the quality of all the 10-20 year old images out there where the uncompressed source is long gone. Sort of a Blade Runner-style "enhance" tool, but of course it would only be filling in aesthetically plausible details.
- kardos 10y ago
  Sounds like the "content aware fill" algorithm [1], only for small scale features that are somehow stitched onto the low-resolution image [1] http://www.logarithmic.net/pfh/resynthesizer http://www.logarithmic.net/pfh/resynthesizer
eon1 10y ago
Sounds a bit similar to the fractal compression experiments[0][1] that were eventually repurposed in stuff like Perfect Resize. IIRC, it worked a bit like RLE, but by generating an IFS on the fly that could be used to recreate the image, instead of just storing pixel sequences. [0] http://www.cs.northwestern.edu/~agupta/_projects/image_processing/web/FractalImageCompression/ http://www.cs.northwestern.edu/~agupta/_projects/image_proce... [1] http://www.math.psu.edu/tseng/class/Fractals.html http://www.math.psu.edu/tseng/class/Fractals.html

lfowles 10y ago

Google Photos already has an incredibly lossy compression, right? Images organized under Dog, Cat, etc. :) Might take a bit more effort to get the extra 999 words.

mcbits 10y ago

I started something like this long ago but didn't get very far with it. This was before "deep learning" and I was groping in the dark, but I think the concept is sound up to a point. The idea was to train a neural net and build up a database of features (maybe on the order of 1-10 GB, or whatever is just small enough to ship) to estimate the missing details from downscaled and extremely over-compressed JPEGs. If it worked, I think it would also improve the quality of all the 10-20 year old images out there where the uncompressed source is long gone. Sort of a Blade Runner-style "enhance" tool, but of course it would only be filling in aesthetically plausible details.

kardos 10y ago
Sounds like the "content aware fill" algorithm [1], only for small scale features that are somehow stitched onto the low-resolution image [1] http://www.logarithmic.net/pfh/resynthesizer http://www.logarithmic.net/pfh/resynthesizer

eon1 10y ago

Sounds a bit similar to the fractal compression experiments[0][1] that were eventually repurposed in stuff like Perfect Resize. IIRC, it worked a bit like RLE, but by generating an IFS on the fly that could be used to recreate the image, instead of just storing pixel sequences. [0] http://www.cs.northwestern.edu/~agupta/_projects/image_processing/web/FractalImageCompression/ http://www.cs.northwestern.edu/~agupta/_projects/image_proce... [1] http://www.math.psu.edu/tseng/class/Fractals.html http://www.math.psu.edu/tseng/class/Fractals.html