Thanks bob

Your comments are always worth reading here

Well, Bob, ( I know you already know this), DV's 4:1:1, 8bit colorspace is a very small world compared to 4:4:4 10 bit colorspace to 4:4:4 32bit colorspace. The large screen differently reveals the weakness of the DV format.

While this is indeed only too true I was trying to move a bit beyond that. DV's 8bits can do a great job for final output, where the differences really show up is when you shoot DV!

Same goes for film too, 35mm neg is around 14 bit depth, 35mm print stock is around 10 bit, that's quite a big difference. It's done for economic reasons, release prints would costs the earth if they had the same depth as neg but they don't really need it as they can scrunch what's on the neg quite nicely onto the print.

Same goes for shooting 10 bit video, you get a lot more room to play aorund with it before your render out. Rendering it out to 8 bit DV you can get gorgeous results that are more than adequate for most things.

BTW, I've projected PAL DV in a cinema on several occassions, cut with PAL DV off 35mm. It looked pretty much OK. I'm certain if the 35mm had been off film we would have been in trouble. One thing though, I did nothing to the DV originated footage, well I lie, I fixed up the sound quite a bit. Poor images in a cinema are one thing, bad audio is fatal to the experience.

Bob.

1- If you hack things in Vegas, a lot of the things you can do on a Da Vinci you can do in Vegas. Albiet many times slower.

"power windows" you can do in Vegas using bezier masks. In the filter FX window, be sure to click the triangle things so the filters apply before the mask.
In Vegas, your bezier masks can be any shape (unlike Da Vinci). On software-based grading systems like Quantel iQ and Lustre, they have motion tracking to automate the movement of the mask (which doesn't exist in Vegas).

If you want to pull clean secondary color corrections in Vegas:
Use the eyedropper and drag-select an area of color you want to isolate.
Hit show mask- if it looks clean then you're good. If there's blockiness to it, try:
Chroma blur, horizontal blur = 2, vert = 0. On some other material, horizontal = 3 and vert = 1 works well too. Chroma blur is really slow unfortunately.
Mike Crash's dynamic noise reduction would also work well if the crud in the isolation is from image noise (i.e. video noise in DV, or grain from film). It may have negative effects on image sharpness and may create artifacts on motion (where the motion kind of sticks in place).
You can also use other noise reduction filters in Virtual dub or whatever, and bring it into vegas as a take.

Anyways they graded this thing on a DaVinci and guess what, they found Secondary Color Correction just wouldn't work for them even on such a high end system. Reason being the amount of color compression in DV. In other words what might have been two slightly different colors in the shot can end up being identical colors on the tape. This means the software is unable to differentiate the two as they have the same numerical value.
I don't think DV compression has much to do with two "slightly" different colors ending up being identical on tape.

As far as I can tell, the things that suck about DV compression is:
4:1:1 colorspace - For television, the 4:1:1 compression is not a big deal because we don't have very good color resolution (compared to luminance). However, in some cases the color information can look "blocky". This solved by smoothing out or interpolating the chroma, although there isn't that much underlying color resolution. So it might be a problem if you want to use color information to affect luminance/brightness values.
Adds noise???

I don't have American Cinematographer, so there may have been something lost in translation.

What's possible in Vegas:
Here is some of my own footage that I color corrected in Vegas.
The following file is a 30s sequence I did for school... 11MB in size.
Using just the tools in Vegas, you can make a soft, pastel-ly/rich look.
http://www.glennchan.info/Proofs/dvinfo/EFP1-CC-comparison.mov

Of course 35mm is a better format than DV25, but I think you can achieve a lot with DV25. Generally the most impact from color grading comes from primary color corrections, and that isn't a problem with DV. Magic Bullet Editors for example gives nice results and all it does is primary color correction. Secondary color correction is sketchier with DV- the main problems (probably in order of importance as follows) being video noise, low color resolution, and possible artifacts from DV compression. But you can still do it.

It is possible to up-convert the chroma sampling. Graeme Nattress has a plugin for FCP that does this. Vegas has a Chroma blur.

Of course the footage will still be in 8-bit colour to begin with.

With DV you need to shoot to take account of the format. Reds are a no-no, greens can be good. DV likes neutral colours. In documentary it is hard to control the colours, but for a fictional production the ideal would be to shoot to a colour scheme.

I think that clip shows up what I'm talking about. Some of the daylight still looks pretty blue and the daylight coming through the overhead windows is clipped to hard white. If you'd set the camera so that didn't happen then the talent would have been way too dark and you'd never been able to pull them up.
Under the conditions at the time it's very well shot and the results are excellent, to get it any better you'd either need to gel all the windows or throw so much light at the talent you'd cook them.
I'm not in any shape or form trying to put a downer on DV, without it I'd be an aweful lot poorer.
What I'm getting at is I've seen plenty of before and after shots of footage just like yours where all those problems have been corrected. What we don't see in those before and after frame grabs is the difference in depth behind them. The frame grabs were from 14 bit DIs, the before images looked way worse than yours and yet they could be pulled into line to give a stunning image, I'd bet if I could get the same file into Vegas it'd be able to perform the same magic and I could render it out as great looking DV.
But put a DV camera on the same set with the same lighting and just for a laugh underexpose it 2 stops and I'd get a frame grab that looked just the same as the 14 bit DI does on the printed page, but there's no way I'll ever get it to look like their 'after' shot.
Again this isn't a sleight on DV, it's about the limits of what we can see on a printed page or on our monitors.
Bob.

One thing you are overlooking here are the camera's CCD's. The trend is toward smaller and smaller ones. High end cameras have three 1/2 CCD's, or three 2/3 CCD's, etc while many of us are shooting with three 1/3 CCD's or smaller. The light gathering ability of the camera makes a huge difference in terms of what you are discussing.

John

farss:
If you shoot DV in a "smart" way, maybe you could avoid the problems you mentioned?

In my case:
The footage was shot with a low-end DVCPRO camera (200-series, forget exact model). It's a few thousand dollars, so in the same price range as Sony PD150, DVX100, etc. If that footage was shot with the DVX100, you could adjust the camera's gamma settings to give maximum exposure latitude. Apparently the DVX100 gets 8 stops according to an article I saw on dvxuser.com. One thing to watch out for though is that there is sometimes color shifting as colors approach clipping (turns towards cyan, magenta, or yellow)... in some DVX100 footage, you can see this. Cheaper cameras exhibit it a lot more.
That effect can happen to film too... if you watch 24, the highlights on the ground of CTU headquarters have this cyan fringe around them. Although on film, I suspect you get a lot more exposure latitude before that happens.

Exposure was set so zebras appeared on a quarter on a piece of white foamcore. For that footage, I think it would've been possible to expose a little darker, and then brighten everything up in post. Having a monitoring tool like DVRack or a waveform monitor would help out a lot. This sometimes brings up noise. It's important to shoot with as little noise as possible (it may be that video has less noise than film grain?).

Lighting:
I used 2 X 250W Lowel lights (I think open face) and gelled them with CTBs, but not full strength ones (which I should have).

That's how I would try to make the most out of DV. While it may not compare to film, I think it still gives you a pretty good range to work with. Of course, not everyone shoots with a NTSC monitor, camera with adjustable gamma, waveform monitor or DVrack, powerful lights, etc. (I had none of those things.)
Exposure latitude: Broadcast TV has a range of about 3-4 stops. You want to shoot with a little more than that, but 8 stops from a DVX100 should be good enough? Instant monitoring allows you to control lighting conditions so you can get things right when you shoot.

Just so no one thinks this is an attack on DV (well DV25 to be more specific) another of the Sundance entrants was shot on HDCAM and posted with FCP. Same sort of problem arose. HDCAM is 8 bit 4:2:2 from what I know.
HDCAM has less compression than DV25, it has more spatial resolution but you've still got the same depth, 8 bits. I'm certain that HDCAM cameras let you tweak things more than the average DV25 camera does too.
What I'm on about is what you can do to an image once it's shot, of course how it was shot will have a huge bearing on how your final image looks, of course one needs to understand the limitations of what you're shooting on, be it film or VHS! Yes bigger CCDs help no end too but what happens to the analogue data from the CCDs has a very big impact on what you can do with the digital data.
High end digital still cameras let you record RAW data, the RAW image can look pretty off but you can do way more with that file than if you let the camera scrunch it down to 8 bit jpegs.

Bob.

So, where does that leave HDV with its 4:2:0 color scheme and long form GOP MPEG compression?

glennchan, out of curiosity what did you do to the footage?

http://www.dvinfo.net/conf/showthread.php?t=43873&page=1&pp=15