Better Choice for Keying: MiniDV or SonyIMX ?

[bartrobinson]

I know very little about SonyIMX also, but it being a professional format and miniDV being a consumer format, I would think SonyIMX would be the way to go. They are both compressed from what I know. If they can adjust the compression on the IMX, go for the least possible obviously. Make sure any detail enhancement settings are turned off. Don't use any diffusion/softening filters on the lens.....

[RichardB]

Hi Manuel

I'm not an expert on IMX but I'll tell you what I know. Feel free, anyone, to jump in if I go off the rails :)

The SonyIMX will give you a better key, not so much because of the reduced compression, but because of the better colour sampling.

MiniDV compresses at 5:1, which is not so bad as the CoDec is quite sharp, but suffers from colour sampling 4:2:0 (or 4:1:1 in NTSC land) which means the colour information is sampled at 1/4 the value of the luma. This makes for a poor colour key, and explains why luma samples often work suprisingly well!

The IMX - as I understand- has better compression (up to 3.3:1) and better sampling (4.2.2), so colour is sampled at 1/2 the value of the luma: it's fair to say the IMX is exactly twice the quailty of the miniDV.
The downside is the MPEG format- you need specialised recorders, etc, you'll have to check your NLE and compositor can take MPEG, etc etc. It's a new format and not a user friendly one. I couldn't even tell you what the interface is.

Two other thoughts: consider DVPRO50 which has the same spec as the IMX but is transfered over the much more common SDI or SDTI interfaces (it's basically two miniDV cameras combined)

OR

use the DV camera as a camera only, put the image through a cheap vision mixer on the day and output chroma green onto the recording deck (or even straight into your computer!) for a much cleaner DV chroma signal.

Hth
RichardB

[Blur1]

Regarding IMX, it may well have an SDI variant that you can connect to an uncompressed capture card, which will effectively allow an "uncompressed" capture. What this really means is that you will be going from the native mpeg compression to an uncompressed Quicktime movie or similar. This is a normal way of working, for example if you capture HDCam, most systems except flame will not allow ingesting the footage using the original codec, due to licensing issues.

Also, as stated DVCPro50 is a good option if you are capturing through Final Cut Pro because it can capture natively over Firewire using a built-in Quicktime DVCPro50 codec. So you get a lossless transfer to your hard drive from the original tape stream, much the same way as mini DV. For transferring to your computer, you need to use one of the DVCPro50 Decks with a Firewire output. What this means though is that the file size is much smaller than capturing compressed over SDI, and the quality is the same.

As far as keying Mini DV, I have done a music video where I keyed about 5 minutes worth. It was a pain, but doable. Here are my tips.
1. Don't compromise on lighting the greenscreen, meaning make sure it has even illumination, because the odds are against you already and you don't want to add to your problems.
2. Use a greenscreen, unless the subject has luminance or chrominance values that match green and can't be garbage matted. A good example of what not to key against green is blonde hair. There is hardly any luma variation and the chroma info will not yield anything you can really use. If you're keying blonde hair I would use a blue screen because it has better contrast.
3. Try to avoid flowing blonde hair, or any long hair blowing around for that matter!
4. The luminance channel of DV footage is fully sampled, the equivalent of a digital still camera, and is very useful when creating a DV key. If you are keying with After Effects, check out the DVMattePro plug-in at dvgarage.com, it does a good job of utilizing everything possible to aid the key.
5. Use a degrainer such as the excellent Remove Grain effect in After Effects 6.5, to do a pass on your footage that removes noise and compression artifacts, then key the footage and pipe the matte back into your original footage.

Good luck

Michael

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.