Fast motion on VHS leads to blurry DVD

[quickpick]

Hello...

I am currently transferring VHS tapes to DVD, and thanks to some terrific help from these forums, I've settled on a JVC HR-S9911U for my SVHS, a DataVideo TBC-1000 and a JVC MH30 DVD recorder.

I am pleased with the output, except when there is fast motion involved. Considering my goal is transferring over horse races, this has become a problem. I have a 5MB sample clip here:

http://home.si.rr.com/slewogold/video.mpg

I'm sure there is a technical name for what I call a blur effect, but in the race you can hopefully see what I am talking about. The part with the man talking after the race is sharp. If only I can get the races to look that good.

Both clips are from the same tape, which in this instance happens to be a professionally produced video.

I'm not sure if there is a solution or an explanation, but any insight is highly appreciated.

Thanks!

[guns1inger]

Your problem appears to be interlacing, which isn't really a problem after all. Interlacing is a part of broadcast video whereby each frame of video is built up from two fields, each offset by a scan-line form the other. When displayed on a TV, persistence of vision allows you to see a single frame, even though two fields have been shown in quick succession to make this happen.

The problem is that your TV is not the same as you PC. Your PC is a progressive display device, which shows frames instead of fields. Because the fields are slightly offset (especially during fast moving video), this looks very off when seen on a progressive device.

Progressive televisions address this by deinterlacing on the fly. You can use software to deinterlace, however it generally results in a throwing away information or blurring it much worse than what you are seeing.

If you leave it interlaced, you should find it looks fine on your TV. If you want to put your mind at ease, create a small sample disk on a DVD-RW and watch it.

It is a non-issue. But it does come as a shock the first time you see it up close in your own project.

[devlynh]

A couple of things;
First your source is VHS, bad at best, awful at worst. The digitizing will only be as good as the worst item in the chain. Even professionally produced and recorded VHS is 240 lines at best.
Second, although your bit rate is high the DVD recorder is not likely to be able to deal with fast motion as well as a single pass or two pass software codec. The DVD recorder has a finite amount of time per frame (1/30 sec) to analyze and compress the frame, a software codec will slow down to allow better looking frames or allow a second pass to optimize the compression factor (Q). I believe you would get better results be digitizing as DV into a computer (with the TBC in the chain) and then use software to compress to MPEG for DVD.
Other than that I have seen worse stuff with better equipment.

TTFN

[lordsmurf]

a software codec will slow down to allow better looking frames or allow a second pass to optimize the compression factor (Q).

That's not really true. You won't really gain anything from a software method. A DVD recorder has a decent buffer, and you'll notice that it tends to be a full second offset from input to output, meaning it gather at least 30 frames at all times, possibly even more. The MPEG hardware we have right now does really well.

Just be sure you're smarter enough to allow it to happen. If you have a tape with fast motion, use XP 1-hour mode, or FR180 3-hour mode. That way the machine is allowed to use good bitrates for the framesizes.

As far as the video clip, it looks great. I did not see any noise or macroblocking. A horses feet move too fast be recorded precisely by all but high-speed film. Your tv or VHS or whatever will have a slight blur on feet. Always.

[FulciLives]

I watched the clip and it looked fine to me ... can't expect such fast motion to look crystal clear.

Could try like LordSmurf said ... do XP mode (1 hour mode) for such things. Quality DVD discs are cheap these days.

- John "FulciLives" Coleman

[trhouse]

I did a frame grab with virtualdubmod and than grabbed one horse out of that frame below,



The horizontal lines that look shifted right or left a bit is the interlace issue mentioned above. The horses front and rear, left legs being blurred is not caused by the recorder and is not an encoding problem.

If you look very carefully, such as below the horses neck and nose, you will see faint vertical lines uniformly spaced. Those are signs of macroblocking which is caused by the encoder. This issue may be improved with a two pass software encode or a higher bit rate.

[dvd3500]

Hmm... might be a field order problem?

[lordsmurf]

There is no field order problem here.

[quickpick]

Wow, thank you all for the terrific insight. I feel better that at least it doesn't seem like I'm going to do any better. Just one final question from something lordsmurf said:

Just be sure you're smarter enough to allow it to happen. If you have a tape with fast motion, use XP 1-hour mode, or FR180 3-hour mode. That way the machine is allowed to use good bitrates for the framesizes.

I always record with the best possible mode. One of the nice things with the JVC MH30 is the 5 minute increments you can adjust the speed...the clip I had happened to be recorded at FR65. Is there anything that can cause a problem using these "odd" recording modes?

Thank you again for all of the help.

[edDV]

For motion analysis you want to keep interlace 60 field per second integrity and examine fields.

Once you deinterlace to 29.97 frames per second, you blur or artificially interpolate the motion that exists between the odd and even fields. Alternative is to "line double" at 59.94 frames per second but this adds artificial interpolated lines that weren't in the original captured video.

When viewing the material, you need to stop view to original 1/60 sec fields if you want to examine motion detail.

Bottom line. Capture interlaced, encode interlaced and view the DVD on an interlaced TV monitor (Not a plasma or LCD unless field view modes are available).

[trhouse]

There are occasional reports that recorders have some issue with recording modes. For example this recent comment from another thread,

Both machines let you record at a setting that adjusts rates for available disc space. That’s fine, but the format used can’t be read by my PC (PowerDVD had trouble with it), and my SD-4800 can play the disc but can’t see chapters (?!). The D-KR2 lets me set the highest bitrate. and in most instances the Dolby or PCM audio rate, for a tape that runs 1hr+8min. With the ES20, those extra 8 minutes force me to record at 2-hr SP if I want a disc that’s readable anywhere.

The author is referring to a Toshiba and a Panasonic. Testing is the best way to determine if you will have any issues with FR65.

[drstew]

[Just be sure you're smarter enough to allow it to happen. If you have a tape with fast motion, use XP 1-hour mode, or FR180 3-hour mode. That way the machine is allowed to use good bitrates for the framesizes.]

Wouldn't FR90 1.5-hour mode produce a very good result (higher bit rate than SP 2-hour mode, more resolution than FR180 3-hour mode)?[/quote]

[dvd3500]

There is no field order problem here.

Help me understand how you can be sure of this. Thanks!

[lordsmurf]

I always record with the best possible mode. One of the nice things with the JVC MH30 is the 5 minute increments you can adjust the speed...the clip I had happened to be recorded at FR65. Is there anything that can cause a problem using these "odd" recording modes?

That's perfectly fine. FR65 is a good mode.

Wouldn't FR90 1.5-hour mode produce a very good result (higher bit rate than SP 2-hour mode, more resolution than FR180 3-hour mode)?

Yes, FR90 would be good. Better than FR120/SP.

Help me understand how you can be sure of this. Thanks!

Because I looked at the test clip and didn't see one.

[dvd3500]

I can't download it for some reason. Am I right that a field orde rproblem would cause the picture to "jump"?

[guns1inger]

A field order problem usually produces a fairly extreme sort of effect, especially with fast motion. The slight blurring seen int he sample clip would suggest very strongly that the field order is correct.

It is pretty simple to test and see the difference for yourself. Load up some interlaced DV footage and output a short segment, then change the field order and output it again. You will soon pick which is right and which is not.

[dvd3500]

gotcha tanks!