Blocky conversion DV-->Mpeg2 @ 9000Kbps using TMPGEnc 4.0

[manono]

Notice the DV sample (leftmost) looks just like yours. What VFW DV decoder are you using?

Beats me. How do I find out? That diagram you included doesn't mean anything to me, either. I installed Cedocida just yesterday, just so I could work with that DV AVI.

Edit Later: I guess by installing Cedocida it became my decoder. I kind of wish I had back whatever was being used before. Anyway, I took a look inside VDubMod->Video->Compression->Cedocida->Configure, which I should have thought of before. As you suspected, under YV12 Chroma Sampling, it was set for MPEG-2 Non-Interlaced. By changing it to Interlaced, now that frame 173 shows 2 of the encodes with those ugly stripes, the source DV and the Main Concept ones. My CCE and HC encodes are still the same as before. Not sure what that means, if anything. I double-checked that they were really BFF, which I remember doing yesterday before encoding them, and everything checks out. I hope the frame 173 problem doesn't invalidate the other 2 pics I posted of frames 195 and 212.

[jagabo]

Edit Later: I guess by installing Cedocida it became my decoder. I kind of wish I had back whatever was being used before.

If you uninstall it your system should revert to whatever you were using before. Maybe Panasonic DV codec?

Anyway, I took a look inside VDubMod->Video->Compression->Cedocida->Configure, which I should have thought of before. As you suspected, under YV12 Chroma Sampling, it was set for MPEG-2 Non-Interlaced. By changing it to Interlaced, now that frame 173 shows 2 of the encodes with those ugly stripes, the source DV and the Main Concept ones.

So the DV decoder is now preserving the interlaced chroma channels.

My CCE and HC encodes are still the same as before.

Did you reencode them with the corrected Cedocida settings? If not, you're just looking at your existing encodes with the essentially blend deinterlaced chroma channels. If you have reencoded them then the chroma channels are being deinterlaced somewhere else along the chain.

I hope the frame 173 problem doesn't invalidate the other 2 pics I posted of frames 195 and 212.

Well the chroma channels were messed up so they may indeed not fully represent the codecs' handling of the interlaced sources. It would be best if you reencoded them. Depending on how you opened the files with the MPEG encoders you might be using a different DV decoder. CCE and TMPGEnc use VFW decoders if you open the DV file directly with them. If you use an AVS script they will use whatever decoder AVIsynth uses: AVISynth's AVISource() uses VFW decoders, DirectShowSource() uses directshow decoders.

Regarding my earlier diagram: I guess it wasn't clear enough. The issue is the flip side of the "chroma bug" that plagues many DVD Players:

http://www.hometheaterhifi.com/volume_8_2/dvd-benchmark-special-report-chroma-bug-4-2001.html

[manono]

Thanks for the detailed explanation. I read it carefully, as well as the link. I'd read the link before, but only as it applies to DVD players, and made certain to buy one that doesn't have the problem (my beloved Oppo). I never thought I'd run afoul of the problem myself.

If you uninstall it your system should revert to whatever you were using before. Maybe Panasonic DV codec?

Cedocida was the first DV codec I'd installed on this computer. I think in previous computers I had installed the Panasonic decoder. I had to install Cedocida just in order to open the DV source. As I said, I don't see much if any DV AVIs, since I don't cap and don't own a digital camcorder. What I don't quite understand is if DV is by definition interlaced (isn't it?), why Cedocida, by default, installs with progressive chroma sampling, just so dummies like me get tripped up.

Did you reencode them with the corrected Cedocida settings?

I hadn't, because I didn't realise progressive sampling would mess up both the encoding as well as the decoding. I've reencoded both my CCE and my HCEnc versions, and now see the pic of frame 173 as in your pic of it, with the yellow stripes in the light, and the blue stripes for most of the way down, for all 4 pics. I've replaced both of my main pics back in my older post, and removed the one without the stripes.

Well the chroma channels were messed up so they may indeed not fully represent the codecs' handling of the interlaced sources. It would be best if you reencoded them.

Maybe because the predominant color was blue in frames 212 and 195, there were virtually no differences I could detect in my new frame 212, and for my frame 195, the one where HCEnc and CCE were using B-Frames, HCEnc improved very slightly for that frame, and CCE, while very slightly different, didn't seem any better or worse to me.

If you use an AVS script they will use whatever decoder AVIsynth uses: AVISynth's AVISource() uses VFW decoders, DirectShowSource() uses directshow decoders.

I always use AviSynth to open videos, and always AviSource or MPEG2Source, unless I can't for some reason, and am forced to use DirectShowSource.

I got a big kick out of the link, because some of their pages I use as my Bible for a number of things, especially as regards DVD players. But I found it very interesting to see them fumbling around and slowly working their way to the source of the problem, and then the solution. I especially enjoyed this (about 2/3 of the way down the page), in a section named 4:2:0 Interlaced: Fundamentally Broken

But as soon as something moves, a strange artifact appears. The chroma information from field 1 leaks into field 2 and vice versa on every pair of fields, and the result is a bizarre pre- and post-echo in the chroma channel. On slow movements, it manifests as a smearing of all the chroma information, and on fast movement it looks like people and objects are split into two parts - a high-resolution luma object and a low-resolution chroma object. It's difficult to describe. We were unable to get a screen shot for this article, but trust us when we say it's just as distracting as the chroma bug, if not more so.

They should have asked me, because I encounter this problem with some frequency. If you work with Indian DVDs, you'll come across virtually everything that can possible be screwed up. I call this one coloring outside the lines, like a first grader and her coloring book. Here's a pic where some of the chroma for one frame comes from the chroma for the previous field:

She's dipping her right shoulder and moving from left to right, and some of the chroma is where it was in the previous frame, to the left of where it should be for this frame. After coming out of RePAL, it shows up like that every fifth frame. Sometimes, on other DVDs the chroma is just shifted over by 4 or 5 pixels in every single frame. That one's fairly easy to fix, though. Anyway, thanks again for your help.

[jagabo]

Cedocida was the first DV codec I'd installed on this computer. I think in previous computers I had installed the Panasonic decoder. I had to install Cedocida just in order to open the DV source.

Just FYI: Windows ships with a DirectShow DV decoder. So you could use DirectShowSource() to open DV AVI files without having Cedocida (or any other VFW DV decoder) installed.

Did you reencode them with the corrected Cedocida settings?

I hadn't, because I didn't realise progressive sampling would mess up both the encoding as well as the decoding.

If the Cedocida decoded the DV to YV12 incorrectly the MPEG encoders were starting with bad frames.

[45tripp]

Aside from the aforementioned unfairness in comparing:
single frames,
sections of a whole frame actually,
different frame types...

there's also the fact that Tmpg Plus was used as the tmpg representative.
Xpress is vastly superior to Plus.
a fact many seem oblivious to.

frame 212 from Xpress 4, as an I frame: