Three whys eh? :)

For the record:

1) Linear Fades
2) Coherent Preview Behavior
3) Non-mangled histograms (and preserved midtone contrast) any time you/grade alter levels
and a special bonus
4) Color Curves plugin actually acts like a curves plugin, rather than an abomination

To be honest I use 8-bit mode almost exclusively because it's so much more stable and less wonky, but that's the only reason I use it. Otherwise, I'd happily use FR for everything.

Your reply is interesting and I am certain it raises a lot of questions about your workflow desired- I am speaking on behalf of the larger audience here. It would be educating as to what you use for source video, what your delivery outcome is expected to be, and very importantly how did you arrive at these findings.

I have only worked demo material in full range coming from RAW, and going to something like exr, dpx, or xavc. I will say that I feel like the stability is much better in VP13. I also think there are a great many factors that go into stability, and most are probably end user issues with choice of hardware and configuration.

Can you not download a trial of VP13 and judge stability for yourself?

I unfortunately already downloaded the demo, so I'm past the 30 days. But thank you for your thoughts on VP13's stability.

I think Vegas' problems with 8 bit processing, and truncation, are pretty well fleshed out by now. My understanding is the general consensus is do your (critical) color work elsewhere, right? And let Vegas do what Vegas does well--timeline editing and compositing.

Also, since 8-bit often forces us to do a lot of levels conversions (esp if you're mixing sources), you do get a more jagged histogram, which probably falls into the category of maybe you see it, maybe you don't. Personally I'm convinced Vegas softens midtone contrast in 8-bit mode if you drag the levels (and maybe even if you don't).

Someone once said Vegas really needs a 10 or 12-bit mode, and I wouldn't disagree.

Previewing in full range is WYSIWYG which I do find refreshing.

The linear gamma fades do have a pretty look, I think, especially between richly colored stills.

Unless you do front and back levels conversions, the color curves plugin doesn't adapt properly to 8-bit mode (it still acts on the full 0-255 range), which forces us to do workarounds if you want, for example, to smoothly enhance contrast w/o clipping.

I appreciate all your trying to say - but unfortunately it does not add up in the real world.

VegasPro edits appropriately with 10bit digitial intermediate codecs such as Cineform, Avid, Sony MXF, and BlackMagic Design. For the most part you do not have access to anything but 8 bit sources and your result is going to be 8bit outcomes no matter what. Now given that there is a place, a very limited place for 32bit mode video levels and 2.2 gamma. I have covered that earlier.
As for linear gamma, you just !! would not be using that on 8bit sources - industry recommends that you leave that issue for 16bpc or 32bpc sources. And that is just not the common video scenario - in fact something more akin to editing still photos or shooting HDR images.

In 8bit here are some of the workarounds:

Color curves:
http://www.glennchan.info/articles/vegas/color-correction/tutorial.htm

Linear light (you can it linear gamma) transitions:
http://www.scomo.co.uk/

In an ideal world, Vegas would handle all the levels conversions for you. Then you can edit and do a first pass of all your effects in 8-bit mode with fast performance. Then switch to 32-bit and render if you have any problems with banding or bit depth.

As for linear gamma, you just !! would not be using that on 8bit sources - industry recommends that you leave that issue for 16bpc or 32bpc sources. And that is just not the common video scenario - in fact something more akin to editing still photos or shooting HDR images.
Check out the SMLuminance plug-in.

It receives 8-bit values from Vegas, internally performs calculations at 32-bit float (or an approximation of it???), and you can see the difference.

Glen - your mention of the SML plugin is interesting - but I note this development cycle near the end of 2011 pointed out the plug-in created memory leaks in the system _AND it would be NOT advisable to download this on the net from a black-hole.

Some compare it's function to Sony built-in gradient wipe - obviously that might be a stretch - but cannot support and recommend this mention..

Glen you're saying imposing linear gamma on 8-bit sources is not advised (which, as a general rule, makes sense to me) but that linear fades can done in 8-bit via workarounds if it suits your taste, yes?

I'm probably conflating the two 32 bit modes here a bit, so I apologize for that.

Yes, obviously an 8-bit source to an 8-bit codec yields...an 8-bit result. But, the math used internally to get there makes a difference. If it stays in 8-bit, as Vegas does in 8-bit mode, you get some truly ugly looking histograms if you apply levels/color processing en route to your 8-bit output.

Does that matter? That's not an argument I'm looking start.

But you can bump up to 32-bit processing, which makes for much prettier histograms, which means less quantizing/truncating is taking place, even for 8-bit source to 8-bit codec.

To make things more complicated, you can then take your (32-bit processed) 8-bit result, stick it back into an 8-bit project, edit it without any grading, and it will still look like the 32-bit processed file when you render it out, using 8-bit math, to your 8-bit codec.

Does it matter? Presumably people care more about how things look on screen than how smooth histograms are, so maybe not.

**And if your source footage is well-shot, so that minimal levels, color, and contrast work is needed, 8-bit math seems to work fine.

But I will say whenever I'm pushing on 8-bit source files, I find Vegas (in 8-bit mode) the very hardest of all the video editors I use to get a nice, bright, punchy result. That's probably due to a lot of factors, including the preview issues, but that's another topic...

Thanks again for the stability report!

Yes, obviously an 8-bit source to an 8-bit codec yields...an 8-bit result.
The actual answer is complicated.

Whenever you go from 8bit RGB <--> 8bit Y'CbCr, there will be rounding error.

You can get rounding errors when going from 8-bit Y'CbCr to 8-bit RGB back to 8-bit Y'CbCr. This is very common in Vegas. Most of the time you won't see problems from this.

The 32-bit floating point mode in Vegas may or may not fix that.
Sometimes it may not fix things because when you render, Vegas might convert everything to 8-bit RGB and then pass that onto the codec.

Glen you're saying imposing linear gamma on 8-bit sources is not advised (which, as a general rule, makes sense to me)
I didn't say that?

Linear light compositing on 8-bit sources does make sense. The difference is dramatic. Clearly the aesthetics are different; arguably better. From a technical standpoint it isn't really better or worse.

----
I find that the 32-bit floating point mode in Vegas is unintuitive for beginners (bad design). So normally I would recommend that people stay away from it.

Switching between the 1.000 mode and 2.222 video gamma mode changes the behaviour of filters. This is pretty unintuitive. Changing how images are composited should not change the behaviour of filters (IMO), but that's what Vegas does.

And then changing between video and studio levels changes the behaviour of some but not all of the codecs... again that is unintuitive.

But if it works for you then go ahead and use it.

Glenn I misread your earlier post--sorry. You were quoting someone else.

I think the 32FR mode does give footage a distinctive look. Subtle, sometimes, obvious others, kind of maddening to pin down. But it's a mess to work with, for many reasons.

Here's a mildly-related question: if I'm working in 8-bit in Vegas, and I do any grading on a film clip, I'm getting rounding errors from that levels correction. That has to be compounded with more rounding errors when Vegas performs the studio-to-computer conversion for preview, right?

So is there any way to preview in Vegas in 8-bit mode that isn't twice compromised? Or am I misinterpreting how Vegas does its math?

While the rounding errors exist, they usually aren't relevant because you won't see them. If you use footage from a camera, then there will be noise that hides all of the problems.

If you use computer-generated footage then maybe the rounding errors would be a problem.

In general, I wouldn't worry about them.

So is there any way to preview in Vegas in 8-bit mode that isn't twice compromised?
You want your preview to be similar to what your final product will look like. So, you probably want to preview with correct levels. There's no point in looking at a preview with the wrong levels.

If the end viewer would see any problems with the levels conversions, you want to see them too. The levels conversion to make the preview correct may introduce problems with banding; this is desirable.

---
The simple answer:
If you can't see the rounding errors (and usually you can't), they aren't a problem. So just ignore them.