Windows vs Linux decoder quality comparison

[DocAliG]

Hi all,

i'm looking for a fair comparison (in term of quality) between windows build and linux build. I'm actually working on a transcoding tool (from scratch).

If someone heard about this.....

Thanks a lot !

[Sergey A. Sablin]

Hi all,

i'm looking for a fair comparison (in term of quality) between windows build and linux build. I'm actually working on a transcoding tool (from scratch).

If someone heard about this.....

Thanks a lot !

If decoder compliant with spec then output has to be exactly same for each platform decoder support.
So there shouldn't be any difference between decoders in term of quality if their are compliant (except if some of them implement any post-processing which is out of the spec)

[DocAliG]

So there shouldn't be any difference between decoders in term of quality if their are compliant (except if some of them implement any post-processing which is out of the spec)


Does it means that all decoders which are MPEG compliant output the same quality? :rolleyes: Then MainConcept, Dicas, Ateme, Vanguard.... etc etc, are all doing the same thing ?

So everybody should use the JM.....

[Hellworm]

No, the idcts used in mpeg decoders can differ in precision. But the integer transformation used in mpeg-4 avc is always bit exact.

[Sergey A. Sablin]

Does it means that all decoders which are MPEG compliant output the same quality? :rolleyes: Then MainConcept, Dicas, Ateme, Vanguard.... etc etc, are all doing the same thing ?

So everybody should use the JM.....

I've said any avc spec compliant decoders, that means if all of these decoders are compliant then yes - if their aren't then this is a _bug_ in software - not just different quality (like for mpeg-1/2/4p2)

[nm]

So everybody should use the JM.....
Well, personally I prefer watching movies at framerates higher than 1 fps ;)

[DocAliG]

Well, personally I prefer watching movies at framerates higher than 1 fps ;)

lol... :p

Are you saying that the JM is slow ? Noooo......

[Sergey A. Sablin]

Well, personally I prefer watching movies at framerates higher than 1 fps ;)

latest JM software changelog says decoder was improved twice - so it is already 2 fps at least ;)

[bond]

first of all define quality

for my definition of quality check the decoder comparison

[BetaBoy]

first of all define quality

for my definition of quality check the decoder comparison

Bond... Speaking of.... and chance for an update to add CoreAVC?

[DocAliG]

first of all define quality

for my definition of quality check the decoder comparison

Well, err..., it's true that's a hard definition. But any comparison would be appreciated, objective (PSNR, MSE,...) or subjectively. Like people feedbacks who are using both, or one of them....

:thanks:

[foxyshadis]

PSNR? MSE? The only valid comparison you can make for AVC decoders is featureset, speed, and bugs. It was already said that the output of a compliant decoder (barring bugs and speedup hacks) has to be identical to the reference decoder. Not within a certain error margin, like other MPEGs, but identical.

ffmpeg has some speedup hacks, and I don't think coreavc does but I know both have their share of bugs, but the latter are quite rare and the former you have to explicitly enable, knowing it'll reduce quality.

In other words, MSE is always going to be 0 and PSNR is always going to be infinite.

As bond says,

imho the quality of a decoder is defined by four things:
A) price
B) supported features of the format
C) decoding speed
D) postprocessing