Is it harder to encode DV into MPEG2???

[jeanl]

People,
I'm sorry if this post is redundant. I haven't found an answer to my problem looking through the forum.
I'm encoding home videos shot with a DV camcorder, using CCE-SP, VBR, 3 passes, with an average of 7500kb/s (max 9000 min 0). Typical home video: people, scenery etc.

I'm puzzled by the fact that even with a high bitrate like that (7500 kb/s) there is a very substantial degradation between the original DV movie and the MPEG2 result no matter how I tweak video quality etc.
I compare the result with, say, an excerpt from a commercial DVD, and I'm baffled at how much better the DVD quality is compared to my encoding even though the bitrate is fairly comparable. The background is still very sharp, there's no blockiness, tessalation etc, whereas my encoded movies show very large amounts of these! Before you ask, I have made sure the field order is right.

In any case, I'm puzzled. Is this the results of largely superior encoders used by professional studios, or is it more the nature of the original video material?
Am I the only one to be puzzled by that?!

Any pointer/explanation would be greatly appreciated!

Jeanl
P.S. I've tried TMPGenc too, which yields similar quality (not quite as good as far as I can tell)
P.P.S. While I now know quite a bit about the DVD format, I'm still a noob when it comes to encoding, so bear with me! ;)

[Boulder]

Some points to think about:

Camcorder material is often shaky. Even the slightest amount of motion can raise the bitrate demand substantially due to the nature of MPEG2 encoding. This combined to interlaced material makes bitrate hungry stuff;)

Camcorder material can be noisy. You should clean the video a bit, there are good Avisynth plugins which can denoise without removing excessive amount of details. I recommend LRemoveDust, it's a very good one. Just make sure you treat the interlaced material properly, there are some threads about it in the Avisynth usage forum.

Fizick's Avisynth plugins should help you (DePan etc.) do some motion-compensated denoising.

Try some custom quantization matrix, you might be pleasantly surprised. At that bitrate I would experiment with Didée's SixOfNine, even though it was designed for XviD.

[jeanl]

Thanks for the pointers. I will make some more experiments!
jeanl

[hendrix]

also take in consideration that your source DV25 now im not sure if your a PAL person or NTSC...DVDs have a 4:2:0 color space the same as a PAL DV25...NTSC DV25 have a 4:1:1 color space and when you encode to DVD it'll result in a 4:1:0 color space which will result in a blockier image. although you can export as an uncompressed .avi file (4:2:2) or/and use avisynth then encode.

"In any case, I'm puzzled. Is this the results of largely superior encoders used by professional studios, or is it more the nature of the original video material?
most DVDs are not produced using DV25 tapes but rather from Digital Betacam, IMX, HDCAM SR, etc. and all have higher quality. I usually work with IMX masters as our archive masters are all IMX but can encode Digibeta, Betacam SP, DVCAM and S-VHS/VHS. have you tried a OPV and see how it looked? Q1 Min/Max @ 0~9000

[Boulder]

Originally posted by hendrix
also take in consideration that your source DV25 now im not sure if your a PAL person or NTSC...DVDs have a 4:2:0 color space the same as a PAL DV25...NTSC DV25 have a 4:1:1 color space and when you encode to DVD it'll result in a 4:1:0 color space which will result in a blockier image. although you can export as an uncompressed .avi file (4:2:2) or/and use avisynth then encode.

Would any of those Avisynth reinterpolating functions/filters help in that case?

[jeanl]

Thanks for the additional input. It looks like I have a lot of things to try! Thanks a bunch to you all.
jeanl

[hendrix]

Originally posted by Boulder
Would any of those Avisynth reinterpolating functions/filters help in that case?
i can't speak from personal experience but i would think so

at work i encode from IMX and Digital Betacam masters...at home i have the Avid Mojo w/Avid Xpress Pro HD 5.1.2 which lets me ingest at 1:1 which gives me a 4:2:2 color space so 4:1:1 to 4:2:2 interpolation isn't required

[jeanl]

People,
I did a number of tests over the week end. I found that I could get much better results if I de-interlaced the video before encoding. In my test, I used HC (not CCE so this is a bit OT), and the difference between encoding the original interlaced video or the de-interlaced one was quite amazing. Even using a simple Bob() yielded much better encoded results, when watched on a frame by frame basis on my PC. There was very littel (or no) blocking, and no combing. The individual frames looke nearly as good as the original DV ones. I encoded using VBR with an average of 7.5Mb (but even at 6Mb it was still pretty good).
But I have some questions:

- Does it make sense to watch an interlaced video on a frame-by-frame basis? I use PowerDVD, and I wonder what algorithm is used to reconstruct the full frame from the 2 fields when the input is interlaced and you're in "pause" frame-by-frame mode...
- Does it seem normal to you guys that I should get so much better results on a de-interlaced video than on the original interlaced one?

Thanks a bunch for any pointers. I'm still pretty much a newbie in the area of encoding, so I'm not sure I'm not missing something quite obvious!

jeanl

[Boulder]

It's natural that you don't see combing if you use Bob(). What you see is the two fields of one frame being separated into individual frames. Thus, the framerate of the clip is doubled.

Progressive material requires much less bitrate. However, you throw away detail and motion will not be smooth.

My opinion is that if I'm encoding for viewing on the TV, I keep the material interlaced. For PC, deinterlace.

What comes to bitrate, I have MPEG2 stuff encoded at avg 1500kbps at 704x576 and it looks good. So, bitrate doesn't tell everything -- it depends on the material.

[jeanl]

Boulder,
I'm not sure I understand you 100%. I also used TomsMoComp() which does not double the frame rate, and the results were even better (and you're right, there's a bit of smoothing).
In any case you answer prompts another question: when I encoded the original interlaced material and watched the output on my PC, I could see a lot of combing (that was not present in the original). Are you saying that if I watched that clip on my TV, I wouldn't see the combing?

jeanl

[Boulder]

Exactly. The combing will not be visible on the TV provided that the field order is set correctly. For DV it is bottom field first, I believe.

[jeanl]

AH! I never thought of that!
OK, but then can you explain to me why I don't see the combing on the original DV video (when watched on my PC), but I do see it on the encoded version? Is it a problem in PowerDVD?
jeanl

[Boulder]

PowerDVD probably does automatic deinterlacing, in my version I can disable it. Try opening the original clip in VirtualDub and see if there's any combing.

[jeanl]

Boulder, I'll try that. But then the question is, how do I check the quality of the conversion, other than by burning to a DVD-RW and checking on the TV? When I watch the output on PowerDVD, the combing is so obvious that it's hard to actually evaluate the quality of the conversion... Can I watch MPEG videos with virtualdub?
What puzzles me is that PowerDVD correctly displays the original DV video (which is interlaced) but does not seem to do such a good job on the output interlaced MPEG.
In any case, thanks a bunch for the help. I'll do some more experiments tonight!
jeanl

[Boulder]

I'm afraid the only solution is to watch the video on the final viewing device.

You could see if you can force PowerDVD to deinterlace the MPEG file as well, which it apparently isn't doing. There's other media players you could try, for example VLC.

[jeanl]

Thanks a bunch for all the info. Things are getting clearer for me, I have a few things to try when I get home. I'll check whether the output MPEG video looks fine on my TV... Thanks a whole lot!
jeanl

[guada 2]

" Can I watch MPEG videos with virtualdub? "

Hello jeanl,
Have you tried to read your file with Vdub Mpeg2?

[jeanl]

Yes, since my last post, I've been using virtualdubmod which can open avis as well as mpegs. Very handy to make comparisons. However, I'm finding that in motion frames where there's a lot of movement, the combing that results from showing the 2 combined fields prevents me from seeing real encoding artifacts. My next step (tonight) is to include an avisynth filter to separate the 2 fields and visualize only 1 of the two (at half the vertical size) in the original and encoded videos. This should allow me to better detect artifacts in the encoded video for moving frames...
jeanl

[guada 2]

Good continuation and with soon.

;)

[jeanl]

To visualize fields individually (so I can compare the original AVI and the encoded MPEG), I ended up using vdubmod, and inserting a de-interlacing filter (drop top-field for example), which gave me an output that's half as high, just what I wanted for my comparisons. I was then able to compare original and encoded at various bitrates, on a field by field basis. Pretty cool.
My tests seem to indicate that HC does as good a job as CCE at 5-6Mb/s, for my material. At 6Mb/s, the differences between original and encoded are pretty minute, and it would be hard or impossible to see them during regular playback on a TV. I'm now pretty happy with the results!
Thanks to all who helped!
jeanl

[kolak]

For interlaced DV format from miniDV cam I use ProCoder2 and ( in my opinion ) that is the best encoder for this type of source.

[jeanl]

Originally posted by kolak
For interlaced DV format from miniDV cam I use ProCoder2 and ( in my opinion ) that is the best encoder for this type of source.
Have you done extensive tests? Did you see big differences? Which encoders did you compare it to, and at what bitrate??
jeanl

[guada 2]

Procoder 2 is a software of conversion of very good quality, only when the bitrate works to 9800 Kbps.

Nevertheless, the software CCE 2.70 pro excel in all the 4 figures.

[kolak]

HI Jeanl
I use as a source PAL DV from Sony VX2100. I tested CCE 2.70 , ProCoder2 and MainConcept 1.42. Bitrate was avg=5Mbit, max=9000mbit.
Everyone from my friends and I say taht Procoder2 was the best. The picture is detailed and stable. I tried many filters settings in CCE but any settings gave better result then ProCOder2. That's my opinion.

BTW... when I use Divx HD as a source (from www.divx.com/hd) the CCE clearlly won.

[jeanl]

I see, thanks for the clarification. My test was far from exhaustive, in particular, I didn't test many different scenes. You results confirm what seems to be a general feeling that procoder2 is better for interlaced DV material...
jeanl