theora ?

[achtungfeuer]

A new version of theora:
http://athos.leffe.dnsalias.com/ffvfw-20031028.exe
What's about quality and speed?
Begin test next week...

[bond]

does ffvfw offer theora encoding?

[Sirber]

yes.

[RadicalEd]

I wouldn't really call October new... but sure. In my tests the ffstuff implementation of theora is kind of unstable, which is understandable since theora is pretty undeveloped.

[BoNz1]

I did some quick visual tests with theora today and certainly doesn't seem to have any problems quality wise in my short tests done at 1000kbps and 704x304 at constant bitrate (I couldn't get the 2pass to work in ffvfw). Also it is reasonably fast as well approximately 17fps on my P4 1.8.

[bond]

hm i wondered because if i remember it right from what i read on the maillists the specs/bitstreams arent finished at all atm, so you cant be sure that your now encoded files will be compliant to final theora

but for testing and playing around its nice

[Latexxx]

There is some recent progress on theora: http://www.vorbis.com/ot/20040225.ht...ec-ogg-support

[achtungfeuer]

what the difference between theora et ogg tarkin?

[Sirber]

tarkin was (is?) the first try to have a "patent-free" video codec. Thoera is based on on2 VP3 technology.

[virus]

Originally posted by Sirber
tarkin was (is?) the first try to have a "patent-free" video codec. Thoera is based on on2 VP3 technology.

Erh...*cough*, no. Both are intended to be patent-free.
Theora is a "standard" codec based on the DCT (like many other popular codecs) while Tarkin (right now it exists only as a proof-of-concept, probably it will be developed in the future) will use the DWT (Discrete Wavelet Transform).
On2 Tech. released VP3.2 open-source, so there won't be patents on Theora, at all (as a matter of fact, this is the only open+free video format available)

here's a useful link: http://www.theora.org/theorafaq.html

virus

[bond]

Originally posted by virus
so there won't be patents on Theora, at all (as a matter of fact, this is the only open+free format available)

similar to vorbis on the audio side

i love the ogg framework idea

[SeeMoreDigital]

I tried it just before Xmas.

But if I remember correctly I kept getting 'application error' messages when using other tools and apps, such as Gspot!

Shame really because it had a lot of features!

Cheers


EDIT: Features such as 16:9 anamorphic signalling - which I really like...

Cheers again

[Latexxx]

Originally posted by bond
similar to vorbis on the audio side

i love the ogg framework idea

ON2 has some patents on vp3 but Xiph negotiated a deal which gives everyone the right to use the technology covered by the patents for free.

[On2Tech]

On2 put the VP3 codec into the open source domain, where it was picked up by Xiph.

[Rash]

They could have renamed the ffvfw file

Is that the same encoder that comes with the latest ffdshow?

[bond]

Originally posted by Rash
Is that the same encoder that comes with the latest ffdshow?

what do you mean? ffdshow (formerly known as ffvfw) offers with a theora encoder, but not a vp3 encoder (altough theora is based on vp3, they are different codecs!!)

[FoboldFKY]

I just downloaded the ffvfw encoder and did a quick (read: incredibly uninformed, baised, and about as reliable as a German minister's word on software patents) test against an XviD encode I just did (thanks to all who helped!!).

First of all, I managed to do a two pass encode, so yay for me. I used VirtualDubMod, and I encoded the same video using XviD 1.0 (it was the intro of Full Metal Panic -- the first season's intro). ffvfw didn't seem to have a way to limit the bitrate (only target filesize), so I set what bitrate options I could find, and set the filesize to approximately what XviD generated.

That said, the encode was slower than a dog that's been pop-riveted to the ground. It was going on around 1-3 fps most of the time. This is on an Athlon XP 2800+. I dunno if that was because I was doing a two pass (BoNz1 reported up around 17fps for single pass) or something I configured wrongly, but it took forever.

Anyway, the quality was quite good. It didn't `seem' as clean as the XviD encode, but I will say this: it handled reds (especially in the title of FMP) really well. I've seen lots of codecs choke on red before, but no sign of that here. Also, it managed to pick out details just as well as XviD. One scene that's shot underwater, I could pick out the individual waves on the ocean's surface, and they looked just as clear (maybe marginally worse) as in the XviD encode, so score one Theora.

Unfortunetly, it seemed to have some issues with blocks appearing in gradual colour (I'm Australian, not English, btw) graduations, again in the ocean scene, while XviD was smooth as... uh... something really smooth.

In any case, Theora's looking very cool. Speaking as someone who has worked (and IS working) on software projects that would benefit from a really good, free video codec, I'm definetly looking forward to Theora's development.

Also, it should be noted that even though I gave it some extra headroom compared to XviD, it still produced a file about 2.5 meg smaller than the XviD one, so I think they may have some problems with size estimation/bitrate estimation. That, or I have no idea what I'm doing...

That said, I've got some screenies if you wanna see them, but I need sleep :P

[aketon]

I made some tests using ffdshow 20040520 and I can say that I am quite impressed from the results! Theora seems to be a pretty good codec! The only problem I had is when I tried to make a two-pass encode! I didn't had any encoding problems, everything went fine! Where I was really confused, is when I saw all these options under the two-pass second-pass! Is there any quide that explains all these options, or should I start guessing????

[guada]

Hello everyone,

The discussion is very interesting but, i don't understand that we are talking about Theora while other codec broke away from the batch.

SKAL's MPEG 4, a codec supported by FFDSHOW (apparently of Xvid source) offer a quality of picture better than Theora.
At once, well-balanced, precise and faster, it will captivate advanced members of ultra low bitrate.

My settings: an avi file of 45.5 mo..objective 2.55 mo
The last FFDSHOW

*Encoder:Skal's mpeg 4 - Fourcc:Xvid - Bitrate:2690 kbytes

- generic: max I frame int: 240
min i frame: 1
- quantization: h.263 - treillis search quant activated
- mode two pass: 2nd pass int max bitrate: 10000kbts
max overfow improve: 5%
min......... degradation: 5%
I frame boost by: 0%
Below I frame distance of: 10 frame
reduce I frame bitrate by: 20%
Curve compression
high bitrate ..: 0%
low ...........: 0%
Bitrate payback delay: 240 frame
payback proportionnaly activated
- input configure
Post processing activated - preset middle - automatic quality control activated
Sharpen: msharpen - high quality activated -
strengh: 20 - threscold: 75
- awarpsharp Blur mode: high quality 3-pass -
deph: 16 - threshold: 0.5 - blur: 1
- noise activated
new noise algorithm
luminance noise: 15
chroma noise: 0
- setting resize setting: lanczos

That's all.

[Neo Neko]

Originally posted by guada
Hello everyone,

The discussion is very interesting but, i don't understand that we are talking about Theora while other codec broke away from the batch.

Well that's easy to answer. It is the topic of this thread. And it is a royalty free and pattent unencubered format unlike any form of MPEG4.