Hd100

B

brianluce

Carbonite Member
I've had trouble greenscreening with DV. Will the hd100 be superior than the typical pd170. dvx, xl1 etc in terms of suitability for keying? where will this cam fit in for greenscreening, the HVX should be good with its 4.2.2 sampling. how about 4.2.0?
 
what NLE or program are you using to key.

My limit experience with the Pinnacle product had decent keying.
 
Depends on your flow. If your downconverting HDV to say, DVCPRO 50 for SD final product, you'll get an awesome key.

If you shoot HDV and try to stay in the HDV world for keying, you'll have a lot of the same problems that DV has because of the color resolution issues. Note that the problems are worked around if you get a product that simultaneously pulls a luma key to help the chroma.
 
an intermediate HD codec, such as cineform, perhaps?

Anyhow, if it is for SD (DVD for instance), it should be no problem whatsoever: shoot on HD, do the keying, and then render to SD. Small imperfections would be taken care of by the downsizing to SD. That would be better then first downconverting it to DVCPRO 50 - or any DV50 4:2:2 or DV100 codec.

It can be harder to make a very good key in HD, but you just need to be aware that the downconverting will lend a helping hand in smoothing the composition...
 
just wondering how you can shoot something in 4:2:0 and then "downconvert" it to 4:2:2? How can you exceed the native color sampling in post?
 
sorry to harp, but if software can convert 4.2.0 hd to 4.2.2 sd, then what good is the dvx200's 422 capability? it can be apparently accomplished in post with the more form friendly jvc which also features interchangeable lenses.
 
Brian, it's the process of downres-ing that improves the color sampling.

Imagine you shoot HDV, and you have your 1280x720 sized pixels. The color samples are effectively 2 pixels wide and 2 pixels tall, for every single "luma" pixel.

Now what would happen if you shrink everything down (downconverting)? The luma pixels get interpolated together while shrinking, and so do the color samples. a shrunken 2x2 pixel color sample above suddenly now only takes up 1x1 pixels in standard definition.

Follow? My description is not as clear as I'd like it, but if you read it a few times it may make sense. I've tested this theory and it really does work...remember though you have to downconvert to a codec that can carry 4:2:2 color like DVCPRO 50 or uncompressed. Downconvert to DV and you create the same problem all over again.
 
I concur with Nate in that if you down convert to DVCPro50 or uncompressed from HDV you'll pick up luma in the downconvert. Still, this does not take the place of good lighting for your chromakey or a quality keyer like Primatte (Commotion) or Serious Magic or Boris Red (and compariable). I've had really good luck even with 411 while using Boris Red or Primatte. The trick is to use the composite choker to 'de-alias' your key.
 
this is encouraging foks. now the hvx200 can do 422 natively, but it seems that with the hd 100, an extra step in post (the downversion from hd to sd) negates the hvx's advantage? am i missing something?

second question does the hd to sd downconversion take extreme render times?
 
Hmm.. I really would like to know an answer to this. If this is indeed correct, than in therms of sd, the HD100 could be on par with the HVX. However, in HD, the advantage is clearly on the HVX side, which, for independent filmmakers, is still what matters, no?
 
Well, there won't be a lot of difference in SD (if any)

Specifically, if you would later recompress in DV25, there would be no difference...
 
Well, without both cameras being out, there's no way for a valid test to be performed. There's a lot more to the imaging chain than just the compression format!

But without the HVX being available, we can't do a direct comparison, so the only thing we can even try to compare is the compression format. And so I tried, but I can't come up with a valid test circumstance that would really demonstrate it.

Here's what I came up with:

(this picture is the OneRiverMedia codec test page, standard-def, and rendered at 4:2:2 through DVCPRO50).
http://www.icexpo.com/HD100/OneRiver-DVCPRO50.png

(this picture is the OneRiverMedia codec test page, in high-def 720p. I rendered it out to HDV, re-imported the HDV file and resized to standard-def while rendering out to 4:2:2 DVCPRO50).
http://www.icexpo.com/HD100/OneRiver-HDV-to-DVCPRO50.png

Not much of value that can be told from this, since they're not the same source picture. But hey, it's all I could think of at the time...
 
Sure. Nothing's changed, as far as I know, so nothing would be different.

However, Nate Weaver has demonstrated over on DVInfo.net that some very, very nice results are possible with the stock lens. In the clips he posted, there was only one that I saw any significant aberration, which was on a building in the far upper-right corner. The rest looked pretty darn good and very clean. I think the key will be to know the tool and avoid doing things that it can't do well. Which means, stick to wider angles whenever possible, and watch carefully when objects are on the edges of the frame (especially high-contrast objects). Nate's clips look absolutely usable and I was very impressed by them. The difference between this camera and the HD1 is night and day...
 
You must log in or register to reply here.
Back
Top