10bit signals

[Jim Ioannidis]

Hi guys,

I started reading Scott Billups book "Digital Moviemaking" and I have a question that I just need to get answered because either I'm slow (quite possibly) or this is not making sense.

In the early parts of the book Scott talks about 8bit, 16bit and 10bit signals.

He says that a 8bit signal can generate 256 levels of brightness and color, that sounds right.
a 16bit signal can produce over 65,000 levels of color or brightness, this also is right.

But he says that a 10bit signal can produceover 5,000 levels of color or brightness. Am I doing my binary coversions wrong, How can you get 5,000 from 10bit. The most you should be able to get from 10bit I believe is 1024.

He doesn't seem to explain where the 5000 comes from and i just gotta know.

ok, thanx guys.

[Jeff Donald]

Here's a good article about 8-bit, 10-bit and 16-bit color and DV. The product mentioned is for a Mac, so you PC guys can disregard the ads. The theory is still the same, so enjoy.

http://www.digitalvoodoo.net/products/10bit/

Jeff

[Jim Ioannidis]

thanx for the link Jeff.

As far as I can tell that article supports what I was saying.
The article says that 10bit has 4 times more levels that 8bit.
8bit is 256 and 4 times that is 1024.

hmmm, might have to head down to Barnes n Noble and check what other books say on this.

[Adrian Douglas]

Remember, it's only true if it's in color. B&W is just a fantasy world. That's why Hollywood went to color because nobody beleived B&W. Color movies never lie and you can believe everything you see in them...it's true.

[Adrian van der Park]

A lot of times 10 bit is 10 bit logrithmic. I don't have Scott Billups book (but I've heard it's good and a new edition is on its way in the new year) so I can't say if he was or wasn't refering to log space.

I know that when film gets scanned at work and I have to do vfx for it, that I have 10 bit log cineons to deal with. I run the plates through digital fusion and drop them to 8 bit linear (video) to work with since they load quicker. But when I render my cgi, from my 3d package, I have to render out 16 bit lin (in tif format) when going to the fusion station or if the renders are going to Discreet Inferno bay, then I render out 10bit log cineons. There the high bit renders are combined with high bit background plate photography and are composited by a dedicated artist.

So that's a background why film (after it's scanned) has more colour range than video, and thus a richer look. Ideally the compositor should not touch the bg plate, and just muck with the cgi to sit in. That way when it's put back to film, the photography should not have changed at all.

Generally, when colour timing video (I use that term in it's loosest sense) I bump it into 16bit colour space, that way glows and blurs etc. tend to have less banding and can smooth out some tricky spots, then bump it back down to 8 bit if it's going to tape, or 10bit log if it's going to film.

Adrian