What am I doing wrong? HDV to DVD

jimboy · Post by **jimboy** » Tue Sep 19, 2006 1:31 am

I noticed something today.

I presently have two computers that are nearly identical. Both are 2.4Ghz Intel P4 with Asus P4PE motherboards. The only difference is the video card installed. The PC I use with Ulead has an ATI 9800 256M and the other has an ATI all in wonder 7500 card.

When playing back the captured HDV clips on my PC with the 7500 all in wonder (PC I use for DVD burning) the clips show no interlace distortions using Windows Media Player. When using the PC with the ATI 9800 card in it the clips do show the interlace distortion like in the above picture.
I can "print screen" the video to the clipboard with the 9800 card but I can't do this with the 7500 all in wonder. When I "print screen" (and doing a paste into windows paint program) just black shows up where the video should be.
The 7500 card will output composite NTSC, the 9800 will not. I assume I'm seeing the resultant video card output on my PC screen with the 7500 card and therefore it cannot be print screened. Right?

And Jerry I'm currently working on sending you some clips.

jchunter · Post by **jchunter** » Tue Sep 19, 2006 2:34 am

Jim,
Well, this could turn out to be a playback problem, after all. High definition video has caused quite a number of playback devices and software to exhibit problems. For example, I had found Media Player Classic to be the best HD player until I did some tests for this thread, where I discovered that MPC can't express all the resolution that Virtual DubMpeg can (even though the latter can't playback HD without screwing up the aspect ratio). Go figure...

For the record, I have an ATI 9800Pro 128MB video card, with the 6.14.10.6396 version of the driver. But, in the PC, I can clearly see interlace distortions, with some players, that disappear when I deinterlace.

BTW, I always make my final decisions, based on the HDTV output, rather than the PC because that is the only thing that counts when you are showing your video to friends.

jimboy · Post by **jimboy** » Wed Sep 20, 2006 2:50 am

I retested the HDV to DVD process by taking the original captured HDV video
and created a new mpeg file via Share. I selected frame based, two pass encode, VBR 9000, and Dolby audio at 384kbps. I must say there isn't a whole lot of loss with this transfer, but, the swish pan still looks like the picture in the first post of this thread.

I'm working on sending Jerry Jones parts of this video so he can play with it.
Meanwhile I'm looking at other software. Avid & Sorensen Squeeze.

Jerry Jones · Post by **Jerry Jones** » Fri Sep 29, 2006 4:39 am

Well, I received a couple of DVDs from Jim.

(Thanks, Jim!)

At first, I got the impression that some of the HDV files had been rendered out with an incorrect field order setting.

But that impression was incorrect; Jim explained they were all raw captures that had not been rendered.

The bottom line is that Jim's HDV source files are fine.

My computer's S-Video output is connected to an external TV monitor.

The S-Video port apparently isn't designed to display live HDV clip playback; it's instead designed to display standard definition video playback.

Consequently, certain HDV clips can appear to display the kind of vibration and flicker one typically associates with reversed fields.

Once I got that figured out, I down-converted the clips using the following settings:

MPEG files
24 bits, 720 x 480, 29.97 fps
Upper Field First
(DVD-NTSC), 16:9
Video data rate: 9200 kbps
Audio data rate: 384 kbps
MPEG audio layer 2, 48 KHz, Stereo

This worked well.

The down-converted clips that earlier seemed to display incorrectly in HDV format were now being played back in NTSC-DVD format and they looked great.

The MPEG encoder smooths out the super sharp edges.

Still, the NTSC-DVD material does retain incredible sharpness after being down-converted from HDV clips.

However, there seems to be a price to pay for all of that resolution.

There is considerable motion artifacting -- combing -- on anything that moves horizontally.

In my conversations with others who've worked with HDV, I'm told this is a normal motion artifact that results from the down-conversion of such high resolution clips.

Apparently, this is also a price we pay for interlaced imaging.

I'm told that 720/60p (60 progressive frames per second) HDV eliminates this type of motion artifacting when the HD is viewed on progressive high definition displays.

Yes -- one can DEINTERLACE 1080i material to reduce the motion artifacting, but then one loses half of one's vertical resolution, which would seem to defeat the purpose of acquiring in high definition in the first place.

So it's up to you.

By the way, had I been working with 720p clips instead of 1080i clips, I would have instead chosen the FRAME-BASED setting for down-conversion to create a 480p DVD instead of a 480i DVD.

Hope this helps,

Jerry Jones
http://www.jonesgroup.net

Jerry Jones · Post by **Jerry Jones** » Sat Sep 30, 2006 4:03 am

I forgot to mention one other interesting observation.

If I encoded the Jim's HDV clips to either DV .avi or to uncompressed .avi, the motion artifacting no longer appeared during playback.

But when I then converted to .avi clips to DVD-compliant MPEG-2, the artifacting appeared again.

So this would seem to be an "MPEG" thing.

And your mileage may also vary in accordance with the specific video display adapter you use to send the video display to a monitor.

Jerry Jones
http://www.jonesgroup.net

jchunter · Post by **jchunter** » Sat Sep 30, 2006 3:14 pm

Jerry,
I am very happy to see someone else doing real video testing. I hope this will help to restore a spirit of science in this forum, rather than hearsay and speculation.

Your observations are similar to mine. The original captured video appears slightly jerky when panning, as if there was a field order problem. Many others have reported this effect on the Sony user forums at http://www.sonyhdvinfo.com. To date, no one has come up with an explanation.

The first question is: What is the origin of this jerky playback?

The second question is: What can we do about it?

Jerry Jones · Post by **Jerry Jones** » Sat Sep 30, 2006 8:14 pm

Yes, you are quite correct.

The problem is being reported on the Adobe, Canopus, DV, and many other online forums.

I think it may be the "nasty little secret" of the HDV and -- possibly -- the forthcoming AVCHD HD formats.

My current theories:

1. One's PC processor, in some cases, might not be fast enough to play HDV steams with reliable smoothness;

2. The equipment one uses to *display* the playback of high definition HDV clips may have a bearing on how they will appear on a specific type of monitor (monitors can be progressive or interlaced and the capabilities of computer video display adapters attached to those monitor may vary);

3. INTERLACE as a technique for acquiring/displaying horizontal motion in video may not be an optimal solution compared to PROGRESSIVE scanning at high frame rates (e.g. 720/60p). ABC and ESPN have chosen 720p as their format for fast-action sports broadcasting, for example.

Dr. Alvy Ray Smith is a strident advocate of 720/60p as a better solution than 1080i.

He claims 1080i should -- in reality -- be descibed as "540i."

Here are his arguments:

http://tinyurl.com/75spo

As for real solutions, I'm still searching for those myself.

I keep hoping to find an article on the Web that gives a step-by-step approach that reduces the motion artifacting of down-converted HDV without sacrificing vertical resolution through DEINTERLACING.

I've scoured the Web on this topic for months.

But, unfortunately, I've not found a method that overcomes the limitations of my equipment here.

I would be interested to experiment with true high definition DVD players.

I would like to find out if the motion artifacting is still apparent when playing an HDV clip burned to a true high definition DVD disc (Blu-ray Disc or HD DVD) using an HDMI connected high definition DVD player and a *tube* HDTV (with an interlaced screen).

If the motion artifacting is still apparent, then I would be inclined to believe that HDV may not be living up to the hype.

In addition, if somebody were to ask me if I think it's a good idea to spend money on HDV camcorders as a means of getting the best down-converted, standard definition DVD images, I would be inclined to say "stick with 3CCD MiniDV instead."

At least, that's how I see it at this point in time.

Jerry Jones
http://www.jonesgroup.net

maddrummer3301 · Post by **maddrummer3301** » Sat Sep 30, 2006 8:57 pm

Jerry Jones · Post by **Jerry Jones** » Sat Sep 30, 2006 9:59 pm

Well, I doubt the hardware decoder would do any better than a software decoder.

Why?

Because a hardware decoder is a piece of hardware with a software encoder embedded within.

There's really no magic there.

This HDV MPEG motion artifacting issue is well-documented on other sites.

For example, here's an article called "HDV Often Looks Great, But Not Always":

http://tinyurl.com/eqnzu

Motion, especially fast-moving and detailed motion as in sports footage, presents quite a challenge for the real-time MPEG encoders in standard HDV cameras. It’s a challenge HDV can often meet, but not always. (Go to www.studiomonthly.com to view examples of both a good and bad HDV clip and you’ll see what I’m talking about.) This is the downside of HDV’s impressive MPEG compression efficiency.

Jerry Jones
http://www.jonesgroup.net

Jerry Jones · Post by **Jerry Jones** » Sat Sep 30, 2006 10:05 pm

maddrummer3301 wrote:My converts are down to 720x480 Upper field first at 59.94 FPS for the first step, which is technically 720x480p. Then to dvd I down convert again to 29.97 Upper Field First. The 2nd converstion is very fast.
Seems to keep the motion and everything OK that way.

This is interesting.

I haven't tried this yet, but I plan to do so.

It would be amazing if this really did work.

Jerry Jones
http://www.jonesgroup.net

maddrummer3301 · Post by **maddrummer3301** » Sun Oct 01, 2006 12:21 am

Jerry Jones · Post by **Jerry Jones** » Sun Oct 01, 2006 1:07 am

Just tried it and... unfortunately... no joy.

Thanks for the suggestion, however.

Jerry Jones
http://www.jonesgroup.net

jchunter · Post by **jchunter** » Sun Oct 01, 2006 1:07 am

I come at this from the standpoint that semiconductor technology was not yet at the state where it could economically support progressive 1080 line video in a camcorder. So the first consumer HD camcorders were 1080 interlaced. Consider this a boundary condition and forget about whether or not 720p would have been better - its off the point.

So, when the subject matter is moving, the interlaced FRAME looks like the OP's first post on this thread because its two FIELDS were exposed 1/60 sec. (16.7ms.) apart in time. The interlaced Frame is not time-coherent. Panned frames look like double exposures.

Now, if you want to correct this, think about displaying the interlaced FIELDS at 60/sec rather than FRAMES at 30/sec. Then you are seeing a time-coherent 540 line field every 16.7ms.

If you were to display Field1, as it was shot, in the even numbered lines and Field2 in the odd lines, the image would appear to hop up and down, by one line, 60 times per second. This would be clearly annoying, so it would be better to shift the Field2 up by one line, displaying both fields on the same even numbered lines. You would be displaying time-coherent 540 line video in an acceptable manner, with no interlace artifacts on moving subjects.

Does this cut the 1080 line vertical resolution in half? Yes, for static images, but its the best we can do with today's technology. However, moving images display much greater resolution because the interlace artifacts are gone.

After all, is video for displaying static images or moving images?

Is 1080i vertical resolution twice as good as standard def video (480i)? Yes, because 480 line interlaced video would be displayed with 240 time-coherent fields. The HD image has 2.25 times more vertical resolution than SD and I have confirmed this with resolution chart tests.

This is what I expect ideal deinterlacing would accomplish.

Why is the original interlaced playback jittery? I still can't see any reason why it should be.

Despite the defects, the HD video from the HC1 and HC3 blows the socks off any SD video that I have ever seen.

Jerry Jones · Post by **Jerry Jones** » Sun Oct 01, 2006 3:28 am

Users of Sony HC1 and HC3 camcorders are also reporting a "rolling shutter" effect:

http://tinyurl.com/r4qbb

A lot of users are really unhappy about it, too.

Jerry Jones
http://www.jonesgroup.net

jchunter · Post by **jchunter** » Sun Oct 01, 2006 5:59 pm

Time-coherent fields continued

With CRT monitors, this is the way interlaced video is displayed anyway – at 60 time-coherent fields per second and the picture looks good because the phosphors have enough persistence to smooth out the transitions from one field to the other.

I am wondering if the fast response time LCD monitors are a possible cause of the jittery playback of interlaced video. My monitor has a response time of 15ms. and I have noticed that many LCD monitors on the market today have response time of 5 – 10ms. Video playback software may be unable to simulate (digitally) the phosphor persistence effect...

Just out of curiosity, Jim, what is your monitor's response time spec?