Using TMPGEnc Master Works 7 to encode DV-AVI to mpeg2......Best settings.

[davexnet]

Constant quality is a form of vbr because the rate will vary depending on the video's complexity.
But with straight vbr you know the bitrate (and filesize if you calculate it) with cq you do not

[Purple2112]

I have already done this before, way back in 2006 using Nero 7 to encode and author to DVD. I was never happy with the results of having to compress them to fit onto DVD. I just bought TMPGEnc Master Works 7 and I'm hopping to get better results by encoding to file this time, bypassing the DVD limitations.

DVD limitations were not the problem, it was Nero.
It is perfectly possible to make very high quality DVD-compliant encodings from a shaky home video DV source, although it certainly does come with some pitfalls.

But if you are not aiming for DVD or some broadcast delivery format, there is barely a good reason to use MPEG2!

All my footage was shot using a tripod, no shake what's so ever.

[Purple2112]

I'm not sure what you mean by my H.264 has only 1/8 the bitrate of the mpeg2.

You initially reported you encoded a video at 15 Mb/s and that the MPEG 2 file was about 8 GB and the h.264 file was about 1 GB. It they were both encoded at the same bitrate they would both be the same size. If the h264 video was 1/8 the size of the MPEG 2 video then the h.264 video has 1/8 the bitrate of the MPEG 2 video. You've cleared that up now by indicating you were looking at the wrong file.

I just encoded two of the same video, one as H.264 and the other mpeg2. I set both to VBR(Constant Quality) @ 15Mbps Max and (Quality) to 100, see settings.

That doesn't mean both videos were encoded at 15 Mb/s -- only that the max bitrate used anywhere in the video was limited to 15 Mb/s.

Yes, I understand that.

The average bitrate of the video is 1456 Kbps. But many parts have more or less than the average.

Yes, I understand that too. The peaks represent a change in the video, for instance, going from a person standing totally still (would use very little bitrate) to all of the sudden dancing very fast would cause a spike in the bitrate used.

But it's the average that determines the file size, not the max.

Yes, the peaks use the most bitrate and the average uses much less and the constant fluctuations help to save the amount of bitrates used and keeps the file size smaller,........I'm not worried about the finished file size.
In the past I've always just used the default VBR(Average Bitrate) setting because I was authoring to DVD. This time I'm converting to file with no size limitations. I sent a support question to the makers of the software asking which of the three (Rate Control Modes) should I use for best possible picture quality. (The response I got back was in slightly broken English because the software is made in Japan, but was still able to understand most of what they said) They told me to use the CBR (Constant Bitrate) or the VBR(Constant Quality) and change the Auto bitrate setting to manual and set higher for better picture quality. They said to only use VBR(Average Bitrate) if I want to keep the file size small for posting on the internet or authoring to DVD.
I then replied back asking which of the two mentioned would provide the best picture and what exactly does the Quality Setting do when using VBR(Constant Quality), but never heard back from them.

What you are doing by setting the max bitrate is limiting the peaks.

When using VBR(Constant Quality) the only options are Max Bitrate and Qaulity.

[Attachment 64054 - Click to enlarge]

[Purple2112]

As for resing etc it is a little more complicated than I inferred. The image width of 720 pixels on the DV include 16 (some argue 18) of black. Only when the 16 are cropped away will you get the true 4:3 image (720-16=704)/11*10 = 640.

Simple cropping away from 720 to 640 actually makes the image slimmer since you still have the PAR of 10/11 and has you already stated you lose detail were you to attempt a compromise with a PAR of 1:1.

When I changed the resolution to 640X480 the PAR automatically changed to 1:1, I didn't have to change it from 10:11 to 1:1

The main issue keeping at 720*480, despite what your eyes tell you, is that not all players recognise the original PAR/DAR and you will end up with a stretched image.

This is my player https://www.oppodigital.com/blu-ray-udp-203/ and this is my TV https://www.samsung.com/ca/support/model/UN49KS8000FXZC/ and this is my monitor https://www.asus.com/us/Displays-Desktops/Monitors/ProArt/ProArt-Display-PA279CV/ and the picture is not stretched. The clarity is not the best on the monitor, but that's because flat screen monitors are Progressive only. You have to rely on your video card or your playback software to deinterlace interlaced videos and deinterlacing is seldom done right. I have a CRT monitor connected to my PC right for converting the interlaced videos and my TV and player do a good job at playing interlaced content without having to deinterlace it.

Even then a maximum bitrate of 15 mb/s for SD material (6-7 gb per hour) is way too high - throwing bitrate at the lossy compression of DV is not going to save the overall quality. But at the end of the day it is your perogative. And it is a shame that you can not share even a short sample - 20 to 30 seconds would be ample - just to show what could be done with your footage.

I'm going to get you a short sample, I just found the video camera, not even sure if it still works, haven't used it since around 2006-2007. I have to find the tapes, I know I made it a point to not store them in the attic to keep the heat from degrading them, just have to go thru a bunch of totes. I mainly want to provide you with a sample so you can see that they are not stretched looking.

[Purple2112]

Constant quality is a form of vbr because the rate will vary depending on the video's complexity.
But with straight vbr you know the bitrate (and filesize if you calculate it) with cq you do not

Yes, the rate is constantly changing depending on the content, but it should be changing to give you the best quality instead of best use of bitrates as straight VBR does as way of keeping the files within the size limit. That's why I'm curious about what exactly the Quality Setting 1 thru 100 does? Yes, I know 100 is the best, but how does it improve the quality.

[DB83]

There still remains something I do not understand.

I am sure in one post you mentioned that connecting the camera direct to the tv you saw no distortion. Now you state that the camera, which you have recently 'dug-up' plays no part in the playback.


Camera or player can not, surely, display native DV on a tv - a pc player can - so do enlighten me on what the player is receiving.


By all means upload a short sample from the original camera (Shame that 'Grace' will not be 'amazing' ).

[Purple2112]

There still remains something I do not understand.

I am sure in one post you mentioned that connecting the camera direct to the tv you saw no distortion. Now you state that the camera, which you have recently 'dug-up' plays no part in the playback.


Camera or player can not, surely, display native DV on a tv - a pc player can - so do enlighten me on what the player is receiving.


By all means upload a short sample from the original camera (Shame that 'Grace' will not be 'amazing' ).

I'm not sure what you mean by "plays no part in the playback"?
I use to connect the video camera directly to my TV using a Composite 3-RCA Cable, yellow-video and red & white-audio. That's how I use to watch all the videos before I got around to capturing them to my PC and authoring to DVD. The only point I was trying to make is that they never looked stretched, whether connecting the camera directly to my TV at the time https://www.crutchfield.com/S-ta5r3tSP3eH/p_15834XB800/Sony-KV-34XBR800.html or converting to files or DVD and playing thru a DVD/BD player.
I know you want convert the sample video to H.264 to show me how good it can look. I want to convert to both Mpeg2 & H.264 and I'm going to need to know if your using any special settings so I can convert all my other videos myself. I can convert them and have them look okay, but what I'm really interested in is all the little tweaks in the advanced settings. The advanced settings in Mpeg2 aren't that bad, but the H.264 has a lot of little advanced settings and is over whelming. That's why I want to know how to use all the mpeg2 settings first, most of them carry over into H.264 settings which will make it easier for me and not be so over whelming when I try converting to both Mpeg2 & H.264 and be able to compare the finished file conversions.
People are telling me to use VBR so that there is an average bitrate being used. I've been using the VBR(Constant Quality), but I think I may be getting better results from just using CBR(Constant Bitrate). AFAIK, the only reason to use a VBR is to keep from wasting bits thus keeping the file size small.

[Attachment 64110 - Click to enlarge]
So far, the only advanced settings that I'm using for Mpeg2 is the (DC Component Precision), changing from 8 bit to 10. Also, I'm changing the (P &B Frame Relative Quality) from 0 to 2X.

[Attachment 64111 - Click to enlarge]
I should also be able to do away with using B Frames if using a high bitrare. Unfortunately, I don't know how high of a bitrate.
I found these two good reads that cover why my 720x480 footage does not looked stretched. http://dvd-hq.info/faq.php#Resolution
http://dvd-hq.info/dvd_compression.php
Here is a link for the sample video, there should be a download button in the top left hand corner. https://www.dropbox.com/s/6m4xpkm3qw5kh41/.03-02-09_15-12.00.avi?dl=0

[jagabo]

Using Main Profile @ Main Level is limiting you to 15000 kbps. Changing that to Main Profile @ High Level should allow you to raise the Max Bitrate as high as 80000 kbps. When I did that with an old version of TMPGEnc Plus (probably the same MPEG 2 encoder as the version you're using) at Constant Quality 100, MPEG Quantization Matrix, GOP size 18 (IBBP...), interlaced BFF, I got a file that had an almost constant bitrate of 27000 kbps (using your clip). Using only I and P frames was a tiny bit higher. Only I frames a tiny bit higher than that.

[Purple2112]

Using Main Profile @ Main Level is limiting you to 15000 kbps. Changing that to Main Profile @ High Level should allow you to raise the Max Bitrate as high as 80000 kbps. When I did that with an old version of TMPGEnc Plus (probably the same MPEG 2 encoder as the version you're using) at Constant Quality 100, MPEG Quantization Matrix, GOP size 18 (IBBP...), interlaced BFF, I got a file that had an almost constant bitrate of 27000 kbps (using your clip). Using only I and P frames was a tiny bit higher. Only I frames a tiny bit higher than that.

AFAIK the MP@HL setting is for higher resolutions, but I did try it and it did allow me to break thru the 15Mbps ceiling. I set the VBR(Constant Quality) to 28Mbps and my average being used was between 21 and 22Mbps. I couldn't tell if there was a difference in picture quality between using 15 vs 28Mbps. However, I did notice an improvement when using Slow Motion. It played smoother and there was less artifacts/fuzziness around the eyes. (which was cool because I do use slow motion a lot).
How did you manage to only use I frames? These is no option to turn off P frames, only B frames.

[Attachment 64188 - Click to enlarge]

In the advanced settings I did notice a Check box setting labeled: (Encode Key Frames into I Frames)

[Attachment 64189 - Click to enlarge]

I tried unchecking it and then went to the GOP page to see if any of those setting had changed or disappeared, but everything looked the same. It would be nice to use only the original Key Frames and do away with the whole GOP compression, if that's even possible.

[jagabo]

Set the GOP size to 1.

[Purple2112]

Set the GOP size to 1.

Nice! That makes perfect sense. The first frame in a GOP is always an I Frame so if you limit the frames to only one.....you get only I frames. I don't know why I couldn't see that.
Thank you

Let me go try adding this new setting to my others and see how it works out before I move on to any other questions. Thanks again

[Purple2112]

Set the GOP size to 1.

Do you know the minimum bitrate needed for using only I Frames? I found this statement (The above is true for the majority of use cases. For very high bitrate encoders where maintaining high picture quality is more important than saving bits (typically 50Mbps and higher), a GOP length of 1 can be used (that is, if every frame is an I-frame!). This is typically only used for broadcast or production quality and archival encodes)

This statement can be found underneath the two pictures of trees and yellow grass on this page.
https://aws.amazon.com/blogs/media/part-1-back-to-basics-gops-explained/

[jagabo]

Do you know the minimum bitrate needed for using only I Frames?

It's entirely dependent on the particular video. The more detail (and things like film grain and CCD noise count as detail) the more bitrate a video will require.

[davexnet]

Constant quality is a form of vbr because the rate will vary depending on the video's complexity.
But with straight vbr you know the bitrate (and filesize if you calculate it) with cq you do not

Yes, the rate is constantly changing depending on the content, but it should be changing to give you the best quality instead of best use of bitrates as straight VBR does as way of keeping the files within the size limit. That's why I'm curious about what exactly the Quality Setting 1 thru 100 does? Yes, I know 100 is the best, but how does it improve the quality.

It increases the bitrate, makes the file larger. I assume the setting relates to some kind of quantization calculation that the encoder targets.

[Purple2112]

Constant quality is a form of vbr because the rate will vary depending on the video's complexity.
But with straight vbr you know the bitrate (and filesize if you calculate it) with cq you do not

Yes, the rate is constantly changing depending on the content, but it should be changing to give you the best quality instead of best use of bitrates as straight VBR does as way of keeping the files within the size limit. That's why I'm curious about what exactly the Quality Setting 1 thru 100 does? Yes, I know 100 is the best, but how does it improve the quality.

It increases the bitrate, makes the file larger. I assume the setting relates to some kind of quantization calculation that the encoder targets.

Thank you

[Purple2112]

Do you know the minimum bitrate needed for using only I Frames?

It's entirely dependent on the particular video. The more detail (and things like film grain and CCD noise count as detail) the more bitrate a video will require.

DV-AVI doesn't use a GOP structure for compression......correct? If it only uses I frames and all my original DV-AVI files all use a constant bitrate of just under 28Mbps then I should be fine just matching that bitrate when converting to I Frame only mpegs..........No?

[Attachment 64473 - Click to enlarge]
New settings question, my original AVI files all show the chroma sampling as being YUV 4:2:2 in the clip properties, but when converting to mpeg it automatically gets downgraded to 4:2:0. Is there a way to keep the original 4:2:2?

[Attachment 64474 - Click to enlarge]
I tried changing the Profile & Level setting to 422P@HL, but when I do that I get a file that doesn't show a photo on the thumbnail and when I try playing it I get the old "format not supported" error. My player supports up to RGB and 4:4:4

Thanks

[jagabo]

DV-AVI doesn't use a GOP structure for compression......correct? If it only uses I frames and all my original DV-AVI files all use a constant bitrate of just under 28Mbps then I should be fine just matching that bitrate when converting to I Frame only mpegs..........No?

Every compression with a lossy codec reduces detail and creates artifacts. So keeping the same bitrate will not give you the same quality. It will give you less quality. It's up to you if that's good enough.

my original AVI files all show the chroma sampling as being YUV 4:2:2 in the clip properties

NTSC DV AVI is 4:1:1. Most decoders produce 4:2:2. Some by duplicating the chroma samples, some by interpolating.

but when converting to mpeg it automatically gets downgraded to 4:2:0. Is there a way to keep the original 4:2:2? I tried changing the Profile & Level setting to 422P@HL, but when I do that I get a file that doesn't show a photo on the thumbnail and when I try playing it I get the old "format not supported" error.

Yes, you can keep the video as 4:2:2 and encode that way. But as you've seen, support for 4:2:2 MPEG 2 is not widespread.

[ConsumerDV]

I tried changing the Profile & Level setting to 422P@HL, but when I do that I get a file that doesn't show a photo on the thumbnail and when I try playing it I get the old "format not supported" error.

Considering that most DVDs are encoded at about 3 Mbit/s, and Superbit DVDs were about 6 Mbit/s (Desperado was encoded at whopping 8 Mbit/s), I think that 15 Mbit/s ceiling for standard def MPEG-2 is more than enough. But can you play 4:2:2 on a DVD player or even on a BD player? Maybe on a 4K player.

I think that you don't need HL though, because ML should be enough. It seems, that the level limits are different depending on a profile. For MP@ML the limit is 15 Mbit/s, and color subsampling is 4:2:0. For 422P@ML the limit is 50 Mbit/s.

MPEG-2 4:2:2P@ML is defined as SMPTE D-10 (SMPTE 356M, SMPTE 365M) and is implemented in professional formats like Sony's MPEG IMX. Sony offers 30, 40 and 50 Mbit/s MPEG-2 4:2:2P@ML intra, claiming compliance with ISO/IEC 13818-2000.

[Purple2112]

Yes, you can keep the video as 4:2:2 and encode that way. But as you've seen, support for 4:2:2 MPEG 2 is not widespread.

(Just so you know, I'm trying all the settings info that your giving me on both Mpeg2 & H.264, not just on Mpeg2) I wish I knew why my player won't play the converted 4:2:2 videos, it shows that 4:2:2 is supported. Oh well, no use beating a dead horse, onto the next thing.

[Attachment 64539 - Click to enlarge]

Every compression with a lossy codec reduces detail and creates artifacts. So keeping the same bitrate will not give you the same quality. It will give you less quality. It's up to you if that's good enough.

I guess I'll have to play around with different higher bitrates. I did a search for lossless codecs and found out that there is a H.264 lossless codec, compressed, but lossless.(At bottom of this page)
https://www.archivalworks.com/blog/lossless-video-compression
Maybe I'm misunderstanding and it's all about changing certain settings to make it lossless or almost anyway. This is basically why I started this thread, to find out about all the settings and which ones to use for the least amount of compression. We already got the compression down a lot by doing away with B & P frames and using only I Frames as you suggested. Apparently, I can also set the Quantization Parameter(QP) to 0 to not lose any data.
https://www.anymp4.com/video-editing/lossless-video-compression.html
Unfortunately, in the Advanced setting for H.264 there is no setting that just says (Quantization Parameter(QP)), there is Min/Max Quantize Value and the Quantization Matrix (which is something I was going to ask about another time) See settings

[Attachment 64540 - Click to enlarge]
And then there are also these other setting related to Quantization/QP

[Attachment 64541 - Click to enlarge]
Any idea which one is the actual Quantization Parameter (QP) setting? I could just zero out and disable anything remotely related to Quantize, but would like to know why I'm doing it, what the settings actually do.
Thanks for all your help, surprised you haven't bailed on me yet like the others.

[jagabo]

To get lossless encoding with x264 you need to switch to quality based encoding, not bitrate. Then set CRF or QP to 0.

[Purple2112]

To get lossless encoding with x264 you need to switch to quality based encoding, not bitrate. Then set CRF or QP to 0.

I found this old post of yours where you went over quality based encoding and Remuxing.

Regarding quality based encoding vs. bitrate based encoding:

With quality based encoding you're telling the encoder what quality you want. It will use whatever bitrate (and hence file size) is necessary to deliver that quality. With bitrate based encoding you're telling the encoder the bitrate (and hence the file size) you want. The encoder will deliver whatever quality it can for that bitrate.

Different videos require different amounts of bitrate to maintain visual quality. So there's no way to know what the file size will be when you encode with constant quality. And there's no way to know what the quality will be when you encode to a specified bitrate.

Regarding remuxing:

Remuxing is the process of taking compressed audio and video out of one container (MKV in this case) and putting it in another (or the same) container. It does not change the audio or video at all. Like taking a cake out of one box and putting it in another. If you do it carefully there's no damage to the cake.

Are you saying to simply use VBR(Constant Quality) in the Rate control Mode?

As for Remuxing, that would not work in my case because DV is not supported by blu-ray/media players regardless of what container it's in.......correct?

[johns0]

Correct.

[davexnet]

Perhaps the OP should produce a test case for comparison:
Encode a clip with FFmpeg using Libx264 @ CRF 0 and then encode the same
clip in Tmpgenc and compare the two.

Command line similar to this (-crf 0 for lossless)

ffmpeg -i source.mp4  -pix_fmt yuv420p -c:v libx264 -crf 1  -flags +ilme+ildct output.mp4

I've lost sight of the objective, I's the OP trying to find the best quality that will
Play on a blu- ray disk? Something else?

[Skiller]

I wish I knew why my player won't play the converted 4:2:2 videos, it shows that 4:2:2 is supported.

4:2:2 does not play on any consumer hardware. It is a format used in the professional field and all consumer formats use 4:2:0 (or 4:1:1 in case of NTSC DV). You won't be able to play back 4:2:2 anywhere but on a computer.

The settings you show are the output settings of the HDMI connection. This is not related to the color space of the media to be played.

[jagabo]

I know of one device that can play 4:2:2 AVC: the Chromecast with Google TV. That device can also play DV in AVI, MOV, and MKV. It's not as smooth as playback on a PC -- I think it's deinterlaced to 30p for display.

And, by the way, most devices cannot play x264 lossless either. TheChromecast+GTV can.

[davexnet]

I know of one device that can play 4:2:2 AVC: the Chromecast with Google TV. That device can also play DV in AVI, MOV, and MKV. It's not as smooth as playback on a PC -- I think it's deinterlaced to 30p for display.

And, by the way, most devices cannot play x264 lossless either. TheChromecast+GTV can.

Hi jagabo, thanks for corroborating that, I noticed neither the Onn box,
Nor my LG TV natively were able to play it. I used CRF 1 and that worked fine

[Purple2112]

I wish I knew why my player won't play the converted 4:2:2 videos, it shows that 4:2:2 is supported.

4:2:2 does not play on any consumer hardware. It is a format used in the professional field and all consumer formats use 4:2:0 (or 4:1:1 in case of NTSC DV). You won't be able to play back 4:2:2 anywhere but on a computer.

The settings you show are the output settings of the HDMI connection. This is not related to the color space of the media to be played.

Thank you for clearing that up.

[Purple2112]

[QUOTE=Purple2112;2655144]

To get lossless encoding with x264 you need to switch to quality based encoding, not bitrate. Then set CRF or QP to 0.

And, by the way, most devices cannot play x264 lossless either. TheChromecast+GTV can.

If using VBR(Constant Quality) and CRF set to 100 is in fact H.264 lossless then I'm good, my Oppo player does play the videos that I've been converting this way. I'm just not sure why the default, auto, Max bitrate is only 8.2. and why there is even a bitrate setting at all when using this mode. You said to switch to quality based encoding, not bitrate. Here is that setting screen:

[Attachment 64578 - Click to enlarge]

I see that on other software the best CRF setting is 0, on mine it's 0 to 100, 100 being the best.

[Attachment 64579 - Click to enlarge]

[jagabo]

I don't know that software so I can't tell you what quality=100 means. Examine one of the encoded videos with MediaInfo (text mode) and look at the encoder settings. You will see what QP value was used.

Writing library                          : x264 core 157 r2969 d4099dd
Encoding settings                        : cabac=1 / ref=5 / deblock=1:0:0 / analyse=0x3:0x113 / me=hex / subme=8 / psy=0 / mixed_ref=1 / me_range=16 / chroma_me=1 / trellis=0 / 8x8dct=1 / cqm=0 / deadzone=21,11 / fast_pskip=0 / chroma_qp_offset=0 / threads=18 / lookahead_threads=3 / sliced_threads=0 / nr=0 / decimate=1 / interlaced=0 / bluray_compat=0 / stitchable=1 / constrained_intra=0 / bframes=0 / weightp=2 / keyint=50 / keyint_min=5 / scenecut=40 / intra_refresh=0 / rc=cqp / mbtree=0 / qp=0

Note that QP=0 is the actual lossless setting. But when you specify CRF=0 x264 automatically switches to QP encoding with QP set to 0.

And why do you need lossless x264? Even at CRF=10 you can hardly see any difference with a screen magnifier and A/B switching between still frames of the source and the encoded video. Around CRF=18 (and the slow preset, or anything slower) it's hard to see a difference at normal speed playback.

[Purple2112]

I don't know that software so I can't tell you what quality=100 means. Examine one of the encoded videos with MediaInfo (text mode) and look at the encoder settings. You will see what QP value was used.

It turns out that VBR(Constant Quality) with the value setting of 0-100 has nothing to do with lossless at all.. Apparently, this software does VBR(Constant Quantization) with the QP setting using VP8 or VP9 encoding only, no H.264. True lossless is VP9 encoding only.

[Attachment 64676 - Click to enlarge]

Note that QP=0 is the actual lossless setting. But when you specify CRF=0 x264 automatically switches to QP encoding with QP set to 0.

If I use VBR(Constant Quantization) the QP setting that you're talking about is available.

[Attachment 64679 - Click to enlarge]
There is also a true lossless without a QP setting, I guess options are different for different software.

[Attachment 64680 - Click to enlarge]

And why do you need lossless x264? Even at CRF=10 you can hardly see any difference with a screen magnifier and A/B switching between still frames of the source and the encoded video. Around CRF=18 (and the slow preset, or anything slower) it's hard to see a difference at normal speed playback.

Why wouldn't I want lossless, I started this thread asking for help getting the best possible quality regardless of bitrate and finished file size. Yes, I really would like to know how to use a screen magnifier and A/B switching between frames. That would help a lot because I am noticing differences even between the source and the lossless encoding. I just tried saving a frame picture from the software itself, but the size is only 655x480, way too small to see any detail.