Bluescreen - HDV vs DV

Feroxfilm

Feroxfilm

Member
Hi all.
This is a question for all you blue/greenscreen people.
I`m doing this short, and most of the time I`m using blue/green.

Now - I read that film and HD is the best solution, due to color resolution. DV isn`t any good. DigiBeta is great. But I don`t have an Beta deck, and I would prefer to do the editing at my homeoffice.

So - with DigiBeta out of the question - what do I use.
The DVX100a with 24p (or 25p sins I live in PAL land) and all the other good stuff that`s comes with an DVX.

Or - use the FX. Since it is HD in a way.

I can edit with Vegas 6 (PC dude) and use a good chroma software.
Is there any good chroma softwares for HDV?!

DV or HDV on a bluescreen gig?!
 
I personally find it near impossible to pull a broadcast standard key from DV because of its limited 4:2:0 colour space. HDV is the same, however has greater colour resolution because of its overall greater resolution.

Does any one know of a filter that converts HDV into 4:2:2 SD? That would be ideal in this situation. Maybe try converting the HDV steam into uncompressed SD. I guess that would have the same effect, but I haven't tried it.
 
Does any one know of a filter that converts HDV into 4:2:2 SD? That would be ideal in this situation.

I agree :)

But if I have to do on DV, and use a chromafilter, like Ultra 2. THEN add some grains. Maybe from Cinelook. Or make the finished composition a little more "gritty" or "fuzzy" or whatever. Subtile of course. has anybody tried that? I know the guy behind "Sky Captain did it, and look *#¤% amazing. Considering it was DV and AE version 1.1 ;)
 
Feroxfilm said:
I know the guy behind "Sky Captain did it, and look *#¤% amazing. Considering it was DV and AE version 1.1 ;)
Click to expand...

I believe that he only made six minutes of the film that way, without Jude Law etc. He then used that to pitch the idea to the studio, who funded the whole film and a far more expensive effects process.

I don't think any of his original DV footage ever got to the cinemas and was completely re-shot
 
It has to be good enough to be included in a DVD as a spesial feature.
If maybe the finished film is a reshoot with better cams, etc (just dreaming here)
 
From what I hear Vegas will convert HDV to 4.2.2 SD very nicely, render to the Sony YUV codec and it should key well using the CK FX in Vegas. Even for SD the key can be improved by adding chroma blur prior to the key FX
 
Wouldn´t any software capable of resizing images be able to convert HD to SD at 4.2.2? I dont think it is a problem of capability but just pure logic. Wouldn´t a 360x288 image resized from HDV be a 4:4:4?

COrrect me if a i am wrong, please.
 
macgregor said:
Wouldn´t any software capable of resizing images be able to convert HD to SD at 4.2.2? I dont think it is a problem of capability but just pure logic. Wouldn´t a 360x288 image resized from HDV be a 4:4:4?

COrrect me if a i am wrong, please.
Click to expand...

But if you downsize to, say... the DV codec, it will only be in 4:1:1 or 4:2:0 (PAL) again, which doesn't help.
 
If i downsize HDV footage wouldn´t be to DV codec, but uncompressed or MJPEG in the case of using some fx or preserving quality.
 
macgregor said:
If i downsize HDV footage wouldn´t be to DV codec, but uncompressed or MJPEG in the case of using some fx or preserving quality.
Click to expand...

Exactly... if you downsize it and use the DV codec you will have wasted you time.

I am not sure if a straight downsize of uncompressed HD (uncompressed from 4:2:0 HDV) footage would just downsize the chroma too though. Surely the software would need to be told to downsize the luma and conserve the chroma resolution.
 
Not really. I dont think the software should do anything special.

For example:

Clipboard01.jpg


this picture would emulate a color channel in 4.1.1 because each pixel of the image becomes 4 pixels in the screen.


But once it is resampled to half (in reality 4 times smaller), you see the image perfectly without pixelization, meaning you have 4.4.4.

Clipboard02.jpg


I am telling all this without being an engineer. But i think it is correct.
 
Firstly, DV chromakey is possible - with results that are kissing cousins to digibeta. Use a plugin like zMatte or DVmattePro, which reconstitute the half-rez chroma info with stuff from the luminance channel. I do a lot of chromakey with DV from a wide range of sources, with good results.

Secondly, if you shoot HDV chromakey, then downsample to a codec like DVCPRO50 (4:2:2), you're going to get great SD chromakey. You'll probably want to deinterlace if from Sony's HDV codec, but is interlacing part of HD's future? I think not.

Finally, check out stuff like ChromaFlex. A great solution to shooting chromakey in confined non-optimum circumstances. The fact that it works better when the subject is closer to the background is great for corporate shoots, and whilst you need some lighting to fill in shadows to avoid the green cast from the LEDs, the luminance levels are so low that I've been using Tungsten Halogen desk lamps to fill.
 
See, I have had quite the opposite results. I'm a visual effects student in college, and I downsampled some HDV footage to uncompressed SD using VirtualDub, and instead of 4:4:4 SD, I just got crappy MPEG blocking artifacts on the SD image as well.
 
Pixel - was the footage you worked with shot interlaced or in one of the Cineframe modes? I have a theory that Cineframe25 or Cineframe30 footage will give much better keys (in particular upon downsampling) than the original interlaced source. Due largely to how 4:2:0 works with interlaced imagery.

-Spiff
 
All I know is that when you downsize HDV to SD (uncompressed or lightly compressed, eg MJPEG) you should end up with 4:4:4 colour info, thats how I make my DVD files, its by far the best quality i've been able to get so far.
 
He's not kidding about the 4:4:4 SD. While "uncompressed" is naive, here's how it goes down. First define "H" as the horizontal resolution obtained from a 1/2-pixel shifted 3 CCD system. In theory, "H" is probably somewhere between 1440 and 960, but not less than 960).

60i modes: (effective resolution "H"x1080)*
Y: 1440x1080
Cr: 720x540
Cb: 720x540
Note that the Cr and Cb information are temporally offset by 1 field.

CF30/25 (effective information "H"x540)
Y:1440x1080
Cr: 720x540
Cb: 720x540
Note in this case the Cr and Cb come from the same temporal sample. Note also that the vertical resolution available for the CF modes is essentially the same as the effective resolution of the pseudo-progressive frame. In a sense you have "H"x540 4:2:2 going into a 1440x1080 4:2:0 container - which is enough to hold it (if not a little wasteful - there are twice as many luma samples stored as is necessary).

If we're generous and say "H" is in fact 1440, then in a CF image you have 1440x540 4:2:2. Scale that horizontally by a factor of 2, and you have 720x540 4:4:4. Do a little vertical interpolation and you've got 720x480 4:4:4. You certainly do much better than 4:2:2.

And seriously - if you do this down conversion properly and watch the uncompressed SD running of your hard drive - you'll be absolutely blown away by how good it looks compared to DV.

-Spiff
 
You must log in or register to reply here.
Back
Top