Is MJPEG capture good enough for VHS archival?

[n1000]

One of the guides here suggests MJPEG or huffyuv. I know most people on these forums say to use huffyuv. However, I just do not have the space for long term storage in this format. I have 32 hours of video. I was looking for a smaller good enough alternative. So I chose Virtualdub with MJPEG (ffdshow, Quality - 100, Mode - 1 pass - Quality) . All my captures are 300GB total.

I can live with this, but what I really want to know is if this is considered acceptable to most people?

The quality looks good to me, but I don't know bad VHS material when I see it (unless it is really bad). I wonder if I have wasted my time with all this MJPEG capture (I spent a month capturing). Or perhaps the visual difference between MJPEG and huffyuv is only noticeable to the most skilled experts on video restoration?

I don't really want to start over my capture process, but I will if it is clear I am way off on quality because I chose MJPEG. I have read that I can go to huffyuv and then reincode to MPEG-2 at 15mbps. Will this be much better than MJPEG?

I am capturing old VHS home movies to share with family members (on standard DVD and h.264 for computer/mobile device viewing). I want reasonable size masters and high quality h.264 versions as the main final delivery format.

Any suggestions?

[msgohan]

I never use MJPEG, so I'm no expert on the required settings, but I can't get ffdshow to produce a YUY2 MJPEG file even if I choose the Force Input Colorspace option. Instead it converts my interlaced YUY2 to YV12 by assuming the input is progressive. This causes chroma blending.

[n1000]

When I check my MJPEG capture with mediainfo, it says my colorspace is "YUV". Is this incorrect?

I followed the guide here: http://www.digitalfaq.com/guides/vid...virtualdub.htm

However, it does not match exactly to version 1.9.11 I got from the forum here. It also does not cover the settings for huffyuv or mjpeg.

This post (http://www.digitalfaq.com/forum/vide...irtualdub.html) said there was a guide coming with the correct settings for configuring huffyuv in Virtualdub, but I cannot find it anywhere.

Perhaps all of this is covered in other posts, but I can't seem to find the answers.

If I do end up switching over to huffyuv, I do not want to make any mistakes on the capture settings. For example:

What should these settings be?

YUY2 compression method
RGB compression method should
Always suggest RBG format for output
Enable RGBA (RGB with alpha compression)
Swap fields on decompress

In Virtualdub->Video->Set Custom format, Do I make changes here? Do I set it to 720x480 YUY2? Is this where I am telling Virtualdub what color space to use?

In Video->Capture Pin there is also a place to select color space/Compression and Output size.

I am not sure if I need to set this as well?

Thanks

[sanlyn]

YV12 and YUY2 are YUV colorspaces. Colorspaces similar to YUY2 have twice the chroma information as colorspaces similar to YV12. MediaInfo gives further detail in the report's "color subsampling" line. YUY2-type systems are usually named as "4:2:2" subsampling to indicate higher color resolution than YV12 types, which are typically shown as 4:1:1 or 4:2:0. To simplify these formula names, the first "4" indicates luma data, the other numbers indicate bits of color data. NTSC videotape typically uses 4:2:2.

Basically, "YUV" is a name for video matrices that store brightness values and color values separately. Google is your friend: https://en.wikipedia.org/wiki/YUV.

Quote:

Originally Posted by n1000

What should these settings be?

YUY2 compression method: "Predict median"
RGB compression method: "Predict gradient"
Always suggest RBG format for output: disable
Enable RGBA (RGB with alpha compression): disable
Swap fields on decompress: disable

Quote:

Originally Posted by n1000

In Virtualdub->Video->Set Custom format, Do I make changes here? Do I set it to 720x480 YUY2? Is this where I am telling Virtualdub what color space to use?

Yes indeed.

Quote:

Originally Posted by n1000

In Video->Capture Pin there is also a place to select color space/Compression and Output size.

You shouldn't have to change anything here. The compressor you want is set under "Video" -> "Compression...".

Now all you have to tell VDub is whether you're capturing NTSC or PAL, and where to store the capture file.

[n1000]

Thanks for the great answers! I have a few other questions.

The main reason I chose MJPEG over Huffyuv is Huffyuv is not supported in my editor (Cyberlink PowerDirector 13). I would not be able to edit my video and author my DVDs with this codec.

MJPEG worked fine.

1) Is the difference in quality that different between the two? The guide ( http://www.digitalfaq.com/guides/vid...virtualdub.htm) says either one, but I can't find anyone actually using MJPEG on these forums.

2) What do people typically do to edit their videos encoded as Huffyuv? Is this just for long term storage? I've read people say use Avidemux for simple cuts, but I find this very lacking. I don't really want to capture in one format to immediately have to re-encode to another for editing/production.

[msgohan]

Besides cutting unwanted portions, what editing do you do?

[n1000]

Multiple precise cuts. This is the main thing I do. I also may merge clips from multilple captures into the same timeline. Sometimes I add transitions between clips, fade ins for video and audio. I also am used to having an integrated environment for fixing audio problems. I can do some of this in other tools but I need this capability for final format delivery.

[sanlyn]

Quote:

Originally Posted by n1000

The main reason I chose MJPEG over Huffyuv is Huffyuv is not supported in my editor (Cyberlink PowerDirector 13). I would not be able to edit my video and author my DVDs with this codec.

Since lossless media formats are the mainstay of professional restoration processing and editing, what does that tell you about Pinnacle?

Quote:

Originally Posted by n1000

1) Is the difference in quality that different between the two? The guide ( http://www.digitalfaq.com/guides/vid...virtualdub.htm) says either one, but I can't find anyone actually using MJPEG on these forums.

MJPEG fell from the mainstream for advanced hobbyists years ago. Nothing especially wrong with it and some advanced processing still use different implementations of MJPEG, just buggy in some respects as msgohan mentioned. Many advanced restoration and color correction scenarios require better chroma resolution such as YUY2 or RGB.

Quote:

Originally Posted by n1000

2) What do people typically do to edit their videos encoded as Huffyuv? Is this just for long term storage? I've read people say use Avidemux for simple cuts, but I find this very lacking. I don't really want to capture in one format to immediately have to re-encode to another for editing/production.

Not sure what you mean by that, unless you're saying that you prefer to never view your videos on anything except by playing them with a PC. If you want more widely shareable media like DVD/BluRay/AVCHD/mp4 or web posting, you can't use MJPEG. The "typical" user of huffyuv media doesn't do very much with them other than archive the more important material. More advanced users fix up a great deal more using tools like Avisynth and Virtualdub, more advanced NLE's with cleaner overall performance, and most often they'll use much better MPEG or h264/AVC encoders than are found in typical NLE's. But you're not doing any of that. And if, as you say, you wouldn't know the difference anyway, why would you want to change anything?

Rather than capture to a lossy format like MPEG, which is a final delivery format not designed for edits or modification without quality problems, why not continue with MJPEG but keep the archives you want as smaller files by encoding them to broadcast-quality 15mbps MPEG? BTW, your 32 hours of video (lucky guy, I started with over 1200 hours) will capture to losslessly compressed YUY2 720x480 huffyuv or UT Video at about 26 to 30 GB per hour, which will amount to a lot more than 300+ GB of capture until it's eventually encoded to final interframe format. So unless you think you'll put any effort into actual image or playback improvement instead of just the simple "edits" you've been doing, think it over. If you're looking for higher quality, lossless is the way to go, and you'll have to move beyond Pinnacle for many operations.

[n1000]

Quote:

Since lossless media formats are the mainstay of professional restoration processing and editing, what does that tell you about Pinnacle?

It does support lagarith. Just not huffyuv. That takes up more space as well compared to MJPEG.

Quote:

Not sure what you mean by that, unless you're saying that you prefer to never view your videos on anything except by playing them with a PC. If you want more widely shareable media like DVD/BluRay/AVCHD/mp4 or web posting, you can't use MJPEG.

That is not what I meant. I meant I want to use MJPEG as my "master" for editing or archival format. However, I only want to do this if it is very close in quality to huffyuv. I want to start with this master and then do things like:

resize/mask with Virtualdub to remove bad parts on the edge of captures.
I want to fix color issues (too much blue, or too much red, etc)
I want to white balance the image (PowerDirector has an easy way to do this)
I want to remove the grain (Virtualdub with temporal smoother or PowerDirector).
I need to deinterlace for h.264 copies (otherwise I never deinterlace).

I am sure all of this can be done using Virtualdub, AviSynth, or PowerDirector. All of this would occur with the MJPEG capture. Then I would produce DVD, MP4, BluRay, etc. I thought MJPEG was acceptable for this purpose (based on the guide), but recently, I am starting to question that (everyone is using huffyuv).

Quote:

And if, as you say, you wouldn't know the difference anyway, why would you want to change anything?

Well, I certainly don't. I am just making sure I am on the right track after all my initial MJPEG captures. I was hoping that even though MJPEG is not as good as huffyuv, that it was "pretty close with a huge savings on disk space". Something like that. But if everyone here who has experience thinks I have really made a bad choice with my initial capture codec, I will start over. After all I am only doing this once. If only the most elite experts in this field can tell the difference between huffyuv and MJPEG (and I can get 99% the same quality results in the end), then I will stick with what I have. I am very much in the early stages of learning and trying to make the best decisions (and minimize rework).

Quote:

...why not continue with MJPEG but keep the archives you want as smaller files by encoding them to broadcast-quality 15mbps MPEG?

I have seen this before. Does broadcast quality just mean you encode as mpeg with 15mbps? Do I need to install a special codec/tool to do this? So, you are also saying I can reduce the hard drive space required by converting my mjpeg to mpeg at 15mbps and retain the same quality? And then I can delete the mjpegs? I want to come back in 10 years and convert to the next latest format. mpeg at 15mbps is a good choice for this (assuming I don't start over with huffyuv)?

Last question: Is logarith as respected as huffyuv for archival, restoration, etc?

[msgohan]

Quote:

Originally Posted by n1000

I meant I want to use MJPEG as my "master" for editing or archival format. However, I only want to do this if it is very close in quality to huffyuv. I want to start with this master and then do things like:

resize/mask with Virtualdub to remove bad parts on the edge of captures.
I want to fix color issues (too much blue, or too much red, etc)
I want to white balance the image (PowerDirector has an easy way to do this)
I want to remove the grain (Virtualdub with temporal smoother or PowerDirector).
I need to deinterlace for h.264 copies (otherwise I never deinterlace).

Take one of your MJPEGs and do this whole process, saving the settings for each step along the way. Recapture the same tape using Lagarith, and re-apply the whole process. Compare the outputs and see what you think. This way you don't have to rely on our guess at what level of loss you find noticeable/acceptable.