32 bit color and Rebelitude

Love the book, gospel, word, amen, blah blah blah... (no, really) but I have some initial questions:

I note that there's not much discussion of the 32 bit workflow in the Rebel's Guide, presumably because we're not dealing with HDR or film transfers, yes? But still, if I'm using Rebel CC (or Colorista!) would it not be better to be in 32 bit so that there's no degradation between plug-ins? Or is 16 bit sufficient for our purposes?

Also, are you familiar with Microcosm, the lossless 16 bit Quicktime codec from Digital Anarchy? Would that make a good format for DI or Answer Print? It's compression ratio seems pretty good to me, is there an advantage to using still sequences rather than a 'cosm file?

Thanks for the book!

_________________
bart

Hey Bart, thanks for the question—this is exactly the kind of discussion I'd envisioned for this forum.

For DI work in AE, I use both 16 and 32bpc. There are cases where you need the overbright protection, and cases where you don't.

Both Rebel CC and Colorista work in 32bpc, so the nice thing is you can work interactively in 8 or 16, and then switch to 32. If you see an improvement, then render in 32.

But you should always use at least 16, for the reasons I describe on pp. 248–250.

I am familiar with Microcosm, but I do prefer image sequences for final Digital Masters. 16bpc PNGs are great, but TIFFs are good too. And as I mention in the book, it's OK to use an 8-bit file for your master. Far more important to process at high bit depths than to store.

-Stu

And your thoughts about CineForm ,Sheer,MPEG Streamclip etc?

BTW None 16 is FREE!

The DV Rebel approach does not require any third-party codecs.

-Stu

_________________
The DV Rebel's Guide

	Quote:
	it's OK to use an 8-bit file for your master. Far more important to process at high bit depths than to store.

Why is this? Wouldn't you lose all the highs when it writes the 8bit file?

are 16bpc PNG's the ideal format? Why is a PNG better than TIFF (if you save at the same bit depth)?

Just curious about these things, and feel like I've finally found the place to ask! Smile

sean

_________________
if it's not fun, what's the point?
www.mackdadd.com

My tutorials, if you're into that sort of thing.
http://www.mackdadd.com/html/visualfx.html

	hype wrote:
	Why is this? Wouldn't you lose all the highs when it writes the 8bit file?

Yes, just as you would with a 16-bit int file. You have to say goodbye to them sometime, since you're creating a video master, and video isn't HDR.

Check out this shot from Black Hawk Down:

When this was filmed, I'm sure that there was plenty of detail in the sky behind Tom Sizemore. But in the process of creating the high-contrast, punchy look of the film, that detail was knowingly thrown away. It's very important that it was there in the first place, but not at all important to hold onto it once the mastering is over. I wrote about this a bit on ProLost here.

	hype wrote:
	are 16bpc PNG's the ideal format? Why is a PNG better than TIFF (if you save at the same bit depth)?

The lossless compression in PNG is a little better (YMMV of course) and PNG is a less abused format. There are as many varieties of TIFF as there are programs that save TIFFs.

-Stu

_________________
The DV Rebel's Guide

Following-up on this thread, is there any disadvantage to working in 32-bit mode? Will it simply take longer to process/render?

Also, is it acceptable to do much of your correction in 8-bit mode, so you can view familiar 0-255 values, and then switch to 16bpc or 32bpc to take a final look and render? RGB values in the thousands are foreign to me.

And also Stu,

I am hearing/reading podcasts/articles that you've written, and it looks like (to me) that you say go 32 bit instead of 8 or 16. But in the Guide, you say go 16 bit. Could you clarify?

The question of whether 32bpc mode is worth the extra render time is a good one. You discover the answer in a different way than you do for the difference between 8 and 16bpc.

16bpc is just 8bpc but way, way better quality. Anything you do at 8bpc will a) still make sense and b) look better (maybe noticeably, maybe not) in 16bpc.

But 32bpc is not just another step up. It's a different animal. In 32bpc, colors don't clip. This means that some effects will look much different, especially those involving very bright colors and multiple color corrections. This difference might be desirable (removing clipping caused by an exposure effect) or undesirable (a weird color cast to the highlights after a gamma correction).

So while it's a no-brainer to bump up to 16bpc for rendering, it's not so simple for 32bpc. With 32bpc, you should try it. If you like the results better, then use it. If you don't see any difference, or don't like the difference you see, then stick with 16bpc.

That's my advice for onlining. For effects work, start 32bpc and stay there, you won't be sorry.

-Stu

_________________
The DV Rebel's Guide

Awesome info, thanks.

	Stu wrote:
	That's my advice for onlining. For effects work, start 32bpc and stay there, you won't be sorry. -Stu

and by this you mean compositing (mate painting ,2d, 3d and video) or AE "generated" FX like energy balls, beams, lightning etc?

All of the above.

-Stu

_________________
The DV Rebel's Guide

well seems like this video I'm working on now will be all 32bpc... as it is fully made in Ae's 3d space and rally have on effects.. thanx for that tip..

and yes.. what mode would you recommend then for green screen footage?

I'll give you one guess. Smile

-Stu

_________________
The DV Rebel's Guide