mpeg2dec3 with ConvertToYUY2(interlaced=true) vs mpeg2dec

[DoW]

The new decomb that Neuron is working on currently does not have YV12 support in it. As such, some form of conversion from YV12 to YUY2 must be done before it hits decomb. I was wondering if there would be any speed or quality reasons to use mpeg2dec3.dll and then ConvertToYUY2(interlaced=true) versus just using vanilla mpeg2dec.dll. Make any suggestions that you wish. Im seeking max quality, but if they both result in the same quality and one is faster than the other, let me know.

Edit: I really should learn to spell check what I write.

[trbarry]

If you are going to convert to YUY2 before deinterlacing then MPEG2DEC2 will probably do it better. It knows which frames are interlaced and progressive.

The advantage of deinterlacing YV12 is that by the time you are done ALL frames are progressive before any color conversion.

- Tom

edit: Plus of course the obvious speed and quality advantages if you can just stay in one color space all the way through, without conversions.

[DoW]

heh, gimme a dust and convo3d that work in YV12 and do just as good a job as the YUY2 versions. Ahhh.... a man can dream, cant he?
Another question to add because of what you said would be: why would it matter that mpeg2dec2 knows which frames are interlaced and which are progressive?

[neuron2]

Originally posted by DoW
why would it matter that mpeg2dec2 knows which frames are interlaced and which are progressive? YV12 Chroma sampling is different in the progressive and interlaced cases. To make a proper upsampling of chroma to YUY2, you need to know the sampling type.

[DoW]

I assume the same applies the the YV12toYUY2(interlaced=true) included in mpeg2dec3.dll by markFD.

[neuron2]

No, the same does not apply. YV12toYUY2() does not have access to the MPEG flags so it has to be told what kind of upsampling to use. The MPEG decoder has access to the per-frame flags.

[trbarry]

There's not a C3D yet for YV12? I didn't know that.

- Tom

[DoW]

There is a C3D for YV12, but it doesnt remove as much noise as the YUY2 version.

[sh0dan]

Remaining ontopic:

Donald: Is there a nice (fast) pretty accurate comb detection algorithm that can be extracted from decomb 5, you'd recommend? (My intensions with these should be pretty obvious ;) )

[neuron2]

@sh0dan

Please explicitly state your intention so that I may unambiguously determine whether the code will be useful for you. Sure, I can make a guess, but I'd like to know whether you are trying to determine whether a clip as a whole has interlacing, or whether a specific frame has interlacing. If you are trying to automatically determine the chroma sampling type for a convert function to be used for the entire clip, the answer is different than if you intend to adjust on a per-frame basis. I doubt that the latter is even required, but please state your intention.

[sh0dan]

... development stuff moved to a new thread there.

[kassandro]

if a movie with original 24 fps is broadcasted via PAL it is usually done by simply increasing the speed to 25 fps. Consequently the mpeg2 stream is essentially progressive but as all PAL-DVB streams flagged as interlaced. To have the correct chroma upsampling, do I have to deinterlace? Of course I would like to avoid interlacing. Probably, for old black and white movies I do not have to deinterlace?

[trbarry]

If it was properly mastered then you don't need to deinterlace but there is still a chance the fields will be out of cadence, matching like:

A B C ...
B C D

So you still might want to use UnComb() or one of the simpler Telecide() options.

- Tom