[tig] More Viper Stuff

Digital Praxis digital.praxis
Tue Jul 1 19:33:00 BST 2003


Sorry if this message is disjointed from others but I'm bouncing around the world again and can't always keep on top of all TIG messages. Therefore I may repeat info but I have done a number of 'film' projects via Viper, including 3 x 20min shorts and various comparisons & evaluations...

Firstly I don't like the term 2K being applied to HD material, although I know this is becoming a common use, as per da Vinci documentation. 2K to me is 2048x1556/1536.

Viper's biggest problem is the lack of pre-balance (white and black) which causes an offset in the three colour channels. This causes colour 'drift' In RGB as contrast/brightness are changed within the colour grading process. It's always best to filter the lens to correct for the 'green' bias when filming daylight (treat the Viper as a tungsten stock).

The second problem is that the sensitivity to exposure of the RGB channels is different with different amounts of detail being recorded in lowlight and highlight levels so that when performing excessive grades there can be impossible colour biases at the extremes.

The ability to map the CCD to various aspect ratios is interesting but not enough of a benifit as most work will probably be via sperical lens setup with any conversion done in post.

The log mapping of the 12bit linear data is also not matched to Cineon or any other known format so things can get a bit awkward there, especially if going back to film (Quantel's iQ has inbuilt conversion tables to help with this, as I think does Spectre?).

In all the tests I have performed the dynamic range has never bettered 8.5 stops, although I know other people have seen 9. Its also true that the actual DR captured by any CCD is set during manufacture and most 'HD' cameras can capture such wide ranges if you set them to a low contrast output. They just don't map to a log output but that can be done in post anyway.

Of interest the Dalsa camera captures 13 bit linear and can be set to output true .CIN data so things are changing. What I wonder will the Sony F950 be capable of outputting?

In the last Viper I saw the RGB channels were also not aligning correctly with about a pixel of random offset betweem them. This can be an issue when using the data for blue/green screen keying, but that may have been a problem with that particular camera?

The natural green bias can also make green screen keying a problem, so once again an optical pre-filter is a real key requirement.

There are a couple of Viper images on the TIG Twiki pages (DigitalCinematography link) which come from one of the short films we did.

If anyone has any other questions feel free to post on the TIG or contact me directly.

The evolution of digital capture is growing very quickly and with TI's new True 2K digital micro mirror projector chip capable of 1800-1 contrast ratios things are hotting up nicely. I recently saw a 2K RGB 4:4:4 projection via an iQ (using its modified HSDL to provide true 2K realtime 24fps images) and the images were stunning. I have never seen a film projection like this, except via IMAX.

We live in interesting times!

Steve


Steve Shaw
Digital Praxis
+44 (0)7765 400 908:m
digital.praxis at virgin.net





More information about the Tig mailing list