Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Locked thread
Max Facetime
Apr 18, 2009

the best as in most backwards-compatible choice is PNG

1) grab a frame to use as a preview, save as PNG
2) add all I-frames and save as an APNG, rename to PNG
3) add P-frames, B-frames and uncompressed WAV-pieces between chunks containing the I-frames

and there you have it, a file that's an animated PNG in firefox, a normal PNG in other browsers and a video

Adbot
ADBOT LOVES YOU

Max Facetime
Apr 18, 2009

Suspicious Dish posted:

https://www.youtube.com/watch?v=mvqakws0CeU

lets remove all iframes for visual effects!

looks coo, there's 2 shots in <1min that don't match the rest of the video

Max Facetime
Apr 18, 2009

MrMoo posted:

Beyond that not forgetting there is fractal compression which is very compute intensive, produces better compression rates and has a nice blurry artifact when scaling. There was Fractal Image Format (FIF) back in the late 90's, and I believe some US government agencies used it for ISDN based video conferencing at very low bit rates.

I remember getting so very excited when I heard about this. Bitmaps! Fractals! Resolution independent! Zoom without pixelation! it hit all the right points to capture my imagination. then it came out but was wrapped up in so many patents that it was impossible to find a version so you could try it yourself

Max Facetime
Apr 18, 2009


I've had the paper about this on my reading list for a while, read it just now. It's unfortunate and a bit ironic that all the demonstration images have horrible jpeg compression, but this side result illustrates the technique well, I think:



though i'm not sure about using contours as the building block, especially for a video codec. with video wouldn't it make more sense to consider a single frame as a 2D sampling in a grid pattern of a point-cloud in 3D and then try to reconstruct the point-cloud with time and depth from motion, like Kinect? the reconstructed point-cloud could then be approximated using an animated 3D mesh, which would be almost trivial to decode with hardware-acceleration

using a highly detailed 3D mesh as a storage for your raw video would have other advantages too. like in film you could adjust a camera's position within some limits e.g. during editing. or you could use another camera to capture anything that was occluded in the first camera, combine two or more of these together and do impossible matrix-style moving cameras without special effects

Max Facetime
Apr 18, 2009

I proposed using 3d geometry for compressing images earlier, turns out homeworld 2 did just that for their background images:





how it's done:





also I guess it only gains up to a certain amount of high-contrast detali, which is part of why those images are more about style than accuracy

more details here, also there's an animation showing what the background objects look like from outside the box

Max Facetime
Apr 18, 2009

ahmeni posted:

This just looks like its a regular image projected onto a sphere? Where's the storage part?

the sphere, which is laid flat in the wireframe gifs, contains RGB color data from the original bitmap only at each of its vertices. the colors for each filled triangle come from a linear gradient which is calculated from the 3 vertex colors

you can see in the wireframe how the size of the triangles changes where smaller details from the original need to be kept

Max Facetime
Apr 18, 2009

Doc Block posted:

So it's a plain old 3D model...? Storing the color information at each vertex instead of texture coordinates isn't exactly a big deal.

yeah I guess, though usually vertex colors are used for adding color variations or shadows and ambient occlusion to existing models rather than the opposite like here, the model adding details to the coloring

here's what a "compression artifact" looks like:

Max Facetime
Apr 18, 2009


uncanny reality

Max Facetime
Apr 18, 2009

Nomnom Cookie posted:

is there a simple way to dynamic range compression on mp4 files. i have a poo poo setup and that's how i like it, gently caress cinema mixes

media player classic -> options -> internal filters -> audio switcher, turn on normalize

Max Facetime
Apr 18, 2009

Wheany posted:

tbf, there really ought to be some non-poo poo free video editors by now, the year 20-almost-14

there is!

with one editor for the movie
and another editor for the sound

you too can create a youtube masterpiece! like this:

https://www.youtube.com/watch?v=CR0aswHBPq0

Adbot
ADBOT LOVES YOU

Max Facetime
Apr 18, 2009

Caged posted:

lol is that a popup ad

free as in freedom fries

  • Locked thread