Am I correct in thinking that for a feature documentary, the kind of results from HDV, because of its restricted color space, with Chromakey would not be of high enough quality?

Thanks
Greg

The only way to shoot chromakey on HDV is to deliver at standard definition.

To expand... Shoot HDV, ingest into AIC or other high quality codec. Use high end keyer such as KeyLight in After Effects, or DVmatte Pro or zMatte and matte in your background.

Take the quasi-1960x1080i footage, deinterlace, scale it down to SD (The 4:2:0 of HDV becomes close to 4:2:2 through scaling) and output to uncompressed 8 or 10 bit Standard Def (although DV50 and PhotJPEG at 95% are pretty good too) and incorporate into your final edit.

It's not perfect, but it's better than DV. A LOT better. However, there's noise in the background and HDV is fragile with exposure.

If that's not good enough, and you're strapped for cash, borrow a HVX-200. With a bit more cash, rent an XDCAM-HD.

OTOH, there's talk about using the HDMI output from a Sony V1 into Final Cut Pro via a $250 HDMI card, which delivers very good HD chromakey for pocket money.

You could try HDV footage at HD res...
I would shoot in progressive mode, interlacing makes
keying harder.

Take a look at this, all shot with the HVX.

http://www.mediamax.com/mzhd2/Links/94F7950751

4:2:0 color space is the same type of color sampling used by PAL DV and has the same amount of color as 4:1:1 DV, just the chromatic pattern is arranged differently. I think the 4:2:0 color sampling lends itself better to keying then 4:1:1. If you can color correct DV, or PAL video you shouldn't have any problems with the color sampling affecting your keying in HDV.

However, HDV is more compressed then DV and you have a higher potentional for compression artifacts which can affect your keying. Interlaced video does not compress as well progressive scan- so if you can shoot HDV progressive scan, with say the JVC HD100, it will be easier to key than if you shot interlaced to begin with.

De-interlacing the video should help (assuming you shot it interlaced to begin with). Also if you can apply a chroma smooth filter, that may help too. Also, the less movement you have in your chroma key shots the better.

It all depends on the camera used. There is no magic standard how cameras are judged based on how well they key. Cameras however are a lot like film stocks. Certain film stocks are better for keying then others.

It is correct that interlaced HDV is hard to deal with. This is because there are actually two different forms of 4:2:0 sampling in mpeg2. Interlaced encoding and progressive encoding have two different forms of chroma sub sampling. Progressive uses a form of chroma that is similar to how jpeg images compress chroma. Interlaced on the other hand has to alternate the chroma every other line. This is so the second field doesn't always get interpolated chorma.

Progressive 4:2:0 is much better then 4:1:1 to key because all you are really dealing with is chroma pixel blocks that are 2x2 pixels in size. With 4:1:1 the chroma blocks are 4 pixels wide which can leave some pretty ugly wholes and jagged edges.

Interlaced 4:2:0 is actually worse then 4:1:1 because not only does it give you wholes but the chroma samples are off every other line which gives you alternating wholes in your key.

With that said however a lot of people have pulled off some pretty amazing results with 4:1:1 DV and progressive 4:2:0 is even better yet so you should be able to get very good results.

Cameras that record true 100% frame encoded mpeg2 frames are the JVC series of cameras and the Canon series shooting with the F modes.

The other thing you need to think about as well is that besides chroma resolution there are a lot of other factors that determine how good a camera is at keying. Electronic edge enhancements are 100 times more damaging to good keys then chroma detail will ever be. You need to use a HDV camera where all edge enhancements can be turned off while still giving you a high detailed image.

Finally just like with software mpeg2 encoders not all camera encoder chips are of equal quality. This is also true of the DSP in the camera. Keying needs as clean of a image as you can get. This means trying to get as few mpeg2 compression artifacts as you can. A true 24p recording camera makes a little bit better use of the bits because there are slightly less frames to encode in each GOP.

Try to find some examples by searching these forums of projects people have done with all the HDV cameras.