Hey guys, two quick questions.

1: Is there a way to remove the MPEG compression artifacts with any kind of filter. It's making keying almost impossible.

2. How good does SD footage look from the Z1U. I don't hear alot about this particular aspect of the camera.

I doubt if those artifacts can be removed, but maybe someone else will have some ideas.

Actually there has been quite a bit of discussion regarding the Z1 in DV mode and also downconversion in-camera. I haven't been working directly with HDV yet, but have done a lot of SD work with my Z1 in both PAL and NTSC mode. I'm really happy with the results. I think you get superior results by shooting HDV 1080i and downconverting the firewire output as opposed to shooting in DV mode. However you will completely avoid the MPEG artifacts if you just shoot DV mode.

Another thing you might look at is shooting in HDV 1080i and setting component downconvert to 480p. This seems to give really sharp results - on my 1280x768 Sony 17" LCD monitor I could hardly tell it from HD. You can't send this progressive output over firewire though, just through component so you would need a capture card or firewire converter box.

Boyd: the firewire converter box, will it take the component out of the Z1 and convert it to firewire so it can be digitized. If so, I would assume it will also have an audio input along with the component inputs. And if so, who makes it, and how much? Or am I totally wrong?

You can't send this progressive output over firewire though, just through component so you would need a capture card or firewire converter box.

As far as I know, all boxes that can manage component in, can also manage audio input. I've yet to see one that can't.
Convergent Design's HD Connect box, HD Connect LE, will do this, so will the Decklink Multi-Bridge, except it doesn't have firewire anything on it. (I just bought one last week, very impressed)
Miranda's does have firewire, and component in, plus audio, of course.

I haven't used the ADS box, but here are the specs:

http://www.adstech.com/products/API-555/specifications/api-555_spec.asp?pid=API-555

Audio capture in 48, 44.1 and 32 kHz (12 and 16 bit)

Well, I want to shot in HD, and downconvert to 853x480. I thought that would eliminate the MPEG compression. But after doing this with footage online, the MPEG compression artifacts are STILL there.

I was out earlier today and passed by CompUSA.... I couldn't resist the urge to stop by and pick up one of those ADS Pyro capture boxes. I'm going to try some tests capturing footage from the component outputs set for 480p and will post some examples soon.

Daniel, why would you resize to 853x480, is this something you're going to stream on the web maybe? That isn't really a valid size for DV footage. 16:9 anamorphic DV is 720x480. On playback your monitor stretches it out to the equivalent of 853x480.

Regardless, I would expect different results from capturing the Z1's progressive component output than you got by playing around with online footage.

Yes, I know it's not a valid size. However, I will be adding lots of visual effects before hand and I need to work on a native 16:9 shot.

One thing I don't understand is that the other day when I ripped some footage from a DVD, I converted the VOB to AVI using VirtualDub. When I looked at the color space, it looked like it was 4:4:4. I thought it would have all sorts of horrible artifacts since DVD footage is stored as 4:2:0

OK, I just hooked up the ADS Pyro capture box to my z1 via component and set the camera to downconvert to 480p. Long story short: it doesn't work. The Pyro can't lock onto the 480p signal and I get a variety of changing weird things, like 4 little messed up images, or a larger one that's tearing apart.

Now if I switch the z1's component output to 480i then the box works as expected. So I captured the same short clip with the Pyro box, and directly from the z1's i.Link port using HDV > DV downconversion. There's a big quality difference between the two. The 480i component video captured by the Pyro is much, much noisier than the z1's downconverted DV.

Clearly the problem is coming from the noisy capture circuits in the Pyro box; if I connect the z1's 480i component output directly to my 22" Samsung 1280x720 LCD screen it looks far better than it does when I play the captured component video on the same screen (using firewire > component transcoding on my DVD recorder).

So my hope for a cheap way to get 480p into the computer was just wishful thinking I guess :-) Too bad, if I play the 480p component output to that same 22" LCD screen it really looks much better than 480i. I can hardly tell it from the 1080i component video.

Oh well, the Pyro goes back to CompUSA tomorrow...

Boyd, if you won't mind, could you post a few frame grabs of the 480i signal that you captured. I'd like to inpect the frames myself. If you could post a tif or targa, that would be great.

Here you go. These are unaltered frame grabs from FCP, exported as TIFF with best depth. The source footage was a slow tripod mounted pan.

1080i HDV output as 480i component and captured with the ADC Pyro AVlink: http://greenmist.com/hdv/component.tif

1080i HDV output with i.Link downconversion to DV and captured via firewire: http://greenmist.com/hdv/firewire.tif

The noise is more apparent in the moving image. But even if the component capture were of equal quality, there wouldn't be any point to capturing 480i component video. 480p would be worth the effort. And of course when you capture through a device like this there is no timecode so you can't use batch capturing.

Thank you very much. I will be inspecting these grabs, as well as some footage online. I will let you know how my work came out.

Thank you anyways for trying. After seeing the prohibitive costs (at least for me) of the Miranda and other converters, it would have been a blessing if that would have worked for you. I would have bought one myself also. Thanks anyways.



So my hope for a cheap way to get 480p into the computer was just wishful thinking I guess :-) Too bad, if I play the 480p component output to that same 22" LCD screen it really looks much better than 480i. I can hardly tell it from the 1080i component video.

Oh well, the Pyro goes back to CompUSA tomorrow...

Alright guys, I ran extensive tests last night involving HDV footage for keying.

Here are the results:


1. HDV is fine for keying low motion shots, such as interviews or whatnot. The MPEG blocking is very minimal in these shots, and they key well.

2. HDV is almost unacceptable at keying high motion scenes, such as shooting a fight scene at a standard shutter speed. MPEG blocking around punches or kicks is WAAAY to high. Also, weird artifacts surface around actors motion blurred body parts when shot on greenscreen. A grey halo appears.


3. I have not tested fight scene footage against bluescreen. It may not exibit the same artifacts.

4. But overall, HDV does not key well, even worse than DV it seems. Perhaps at a higher bit rate, this would not be a problem. Say at 50mb or 100mb.

overall, HDV does not key well, even worse than DV it seems.

If you downres it to DV how do the keys compare? The in-camera downconversion on my Z1 looks quite good to me, better than my VX-2000 or PDX-10 images. Have not looked at high motion sequences carefully though.

Unfortunately, the MPEG artifacts are still present when I downconvert to SD. I don't know why though. I figured it would smooth them out, but it doesn't.

Has anyone tried bring the HDV into Vegas and then using Vegas to create an image sequence, which can be used for compositing in various apps. It seems that one could retain the HDV quality with this approach. It is much slower though.

Vincent Burnett

I think that Daniel is saying that the "HDV quality" is the problem when it comes to keying. If that's the case then it doesn't matter how you process it.... garbage in = garbage out.

I haven't tried myself, so I just don't know, but a number of other people have made the same observation that HDV doesn't work so well for this kind of application.

Yes, that is correct.

The reason I started this investigation is that we have a Z1U at school. A friend of mine and I are going to be shooting a lightsaber duel for a school project. The whole thing will take place in photorealistic virtual sets. I have to be able to get perfect keys for this. It has to be seemless.

Now, I have found work arounds for DV, and I can get very clean keys with it. But with HDV, I just can't do anything to get decent keys.

HDV color quality should be atleast as good as DV.

You would think. Now, HDV does look beautiful as native footage. But once you try to key it, it presents all sorts of problems. However, I have been able to get professional keys from DV. Take a look...

http://img.photobucket.com/albums/v58/PixelMagic/DVsuperbrawlComp.jpg

When Adam Wilt reviewed the Z1 in DV magazine he said
Single-frame or slow-mo playback reveals blocking or quilting errors, scattered and localized mosquito noise, offset spatial details, and posterization.

I wouldn't recommend HDV for compositing, slow-mo, or still frames, but overall the compression losses were less than I see in over-the-air or satellite HDTV broadcasts.

http://www.dv.com/reviews

You would think. Now, HDV does look beautiful as native footage. But once you try to key it, it presents all sorts of problems. However, I have been able to get professional keys from DV. Take a look...

http://img.photobucket.com/albums/v58/PixelMagic/DVsuperbrawlComp.jpg

Have you tried working with the video uncompressed when you are adding effects? I think perhaps you may be compressing it twice to get your final output. You can tweak your compression settings or go uncompressed. While the video file will be massive, if it is uncompressed, it will save you the headache of trying to get your compression settings right to avoid artifacts.

No, I use uncompressed footage.