| |
 |
|
|
Science Forum Index » Compression Forum » methods for better than h.264 compression ratio...
Page 1 of 1
|
| Author |
Message |
| recoder |
 Posted: Sat May 03, 2008 5:53 am |
|
|
|
Guest
|
Dear All,
We are starting an academic project to build a codec which should
have much better compression ratios than h.264.
Computational complexity of the new codec is assumed to be infinite
for now.
There must be methods that have been rejected for having high
computational complexity. Any hints for us.
Thanks in advance... |
|
|
| Back to top |
|
| Industrial One... |
| Posted: Mon May 05, 2008 2:48 pm |
|
|
|
Guest
|
On May 3, 9:53 am, recoder <kurtulmeh... at (no spam) gmail.com> wrote:
Quote: Dear All,
We are starting an academic project to build a codec which should
have much better compression ratios than h.264.
Computational complexity of the new codec is assumed to be infinite
for now.
There must be methods that have been rejected for having high
computational complexity. Any hints for us.
Thanks in advance...
There are tweakable configurations in x264 such as more B-Frames or
exhaustive motion-prediction searches that would be in the DAYS range
to encode. As for concepts, I can only think of some object-oriented
scheme instead of a rigid macroblock technique that has been in use
since the first video encoding standard. Most today are vaporware. I
doubt any practical one can be built in another 50 years or so, but if
you're looking for advancement then stepping out of this cockwash
matrix is the direction to go. Drop the idea of applying a constant
compression rate to the whole frame and find a way to discern
visualities and the relatively linear relationship with proceeding
frames.
I got flash movies up to an hour long, perfect quality at full screen,
60 fps and only a couple megs. Take a guess why. Now, if I capture all
those frames to raw video and encode with XviD, the result is about
400 megs if I wanna keep a fair quality. With H.264, i can retain the
same quality with a more fucktabulously economic size of 100 megs or
so. But HOW would any video be given a fair trial, such as this one,
to determine that it can be more efficiently encoded in only 3.5 megs?
No automated program can do that. At least until one can write an
algorithm that can intelligently convert photographic text into plain
text.
Or maybe I'm exaggerating here. Nobody thought anything could be
overlooked when XviD was out with its superior quality in 2004 and
only a year later it was replaced by a standard that compresses twice
as better...
They're working on a new one as we speak. So either way, you're
fucked. Your project ain't gonna outclass H.264. |
|
|
| Back to top |
|
| Jim Leonard... |
| Posted: Tue May 06, 2008 3:56 am |
|
|
|
Guest
|
On May 5, 7:48 pm, Industrial One <industrial_... at (no spam) hotmail.com> wrote:
Quote: I got flash movies up to an hour long, perfect quality at full screen,
60 fps and only a couple megs.
I think you meant flash *animation*. Big difference (no real-world
video content). |
|
|
| Back to top |
|
| Industrial One... |
| Posted: Wed May 07, 2008 11:22 am |
|
|
|
Guest
|
On May 6, 7:56 am, Jim Leonard <MobyGa... at (no spam) gmail.com> wrote:
Quote: On May 5, 7:48 pm, Industrial One <industrial_... at (no spam) hotmail.com> wrote:
I got flash movies up to an hour long, perfect quality at full screen,
60 fps and only a couple megs.
I think you meant flash *animation*. Big difference (no real-world
video content).
I'm aware it ain't the same as generic video, but there are ones that
come very close and not instantly noticeable to be computerized
graphics. However, they take up over 1 GB for a couple hours,
extremely CPU intensive, but at least perfect quality. |
|
|
| Back to top |
|
| |
|
Page 1 of 1
All times are GMT - 5 Hours
The time now is Thu Oct 16, 2008 2:20 am
|
|