Vegas Pro 8 Workflow

[Piotr Wozniacki]

Originally Posted by Randy Strome

Also, if you are going for web output, the 32 bit final conversion will produce a huge file in comparison to a 8 bit Computer RGB start point.

Wrong - the Vegas 32bit floating point is the calculation precision setting, and will only increase render time, not the output file size.

[Michael H. Stevens]

Originally Posted by Randy Strome

Hi Michael,

What is your intended viewing device for your project? If you are producing for computer viewing (and are talking about studio 8 bit in your question) then that should be fine, but the downside is that you will have the colorspace change which will entirely change the look of your video at the end, so you will need to edit with "incorrect" colors until you make the final change.

Also, if you are going for web output, the 32 bit final conversion will produce a huge file in comparison to a 8 bit Computer RGB start point.

For now, not yet having a Blu-ray burner or Blu-ray DVD software, I am rendering to m2t data disk for playing from PS3 (or computer HDD) to common HD LCD/Plasma TV. The 8-bit allows me to see the histogram correctly as it displays similar to the EX1 histogram and when I change to 32 bit for render I see no color changes on my monitor, and my monitor looks almost exactly the same as the final result on a TV screen.

[Randy Strome]

Originally Posted by Michael H. Stevens

The 8-bit allows me to see the histogram correctly as it displays similar to the EX1 histogram and when I change to 32 bit for render I see no color changes on my monitor, and my monitor looks almost exactly the same as the final result on a TV screen.

Just to clarify, which 8 bit are you referring to (studio or computer)? I would think it odd if Studio matched your EX1 histogram, and if conversion to 32 bit did not dramtically change the histogram and the colors. Also, when you mention your monitor, are you refing to a computer display or an external monitor?

So many things to clarify :)

[Bill Ravens]

u won't see "significant" differences with 32-bit FP, unless you didn't set the levels properly. If the levels are set improperly you will see significant loss of contrast or crushed blacks, depending on whether you clipped or compressed.

If the levels are set properly, what u WILL see is improved detail in the shadows and hilites. If your monitor is not calibrated, the nuances of 32 bit FP will go un-noticed because the monitor is clipping the data.

[Randy Strome]

Originally Posted by Bill Ravens

u won't see "significant" differences with 32-bit FP, unless you didn't set the levels properly. If the levels are set improperly you will see significant loss of contrast or crushed blacks, depending on whether you clipped or compressed.

If the levels are set properly, what u WILL see is improved detail in the shadows and hilites. If your monitor is not calibrated, the nuances of 32 bit FP will go un-noticed because the monitor is clipping the data.

What Sami posted in post #13 for 8 bit and 32 bit (top 2 examples without any studio to computer RGB) is what I am seeing. I guess I do consider that a significant change, and I would not classify his shot as highly clipped for the subject. In any event, I would not want to make alterations based on his 8 bit and then convert to 32.

[Bill Ravens]

It's abundantly clear from his histograms, that his levels were NOT set right! I'm getting so tired of this arguement. Go back and look at his histogram. The highs are out of legal limit and the lows are way above the legal limit. So, in an NTSC display, of course everything looks washed out. Everything must fit in the 16-235 range, not 10-255, not 24-255.

People are BLINDLY applying the studio to computer RGB and computer to studio RGB presets without thinking about what they're doing, nor are they crosschecking the results against the histogram. You can't be unconscious when you start pushing buttons without understanding what you're doing.

[Michael H. Stevens]

Originally Posted by Randy Strome

Just to clarify, which 8 bit are you referring to (studio or computer)? I would think it odd if Studio matched your EX1 histogram, and if conversion to 32 bit did not dramtically change the histogram and the colors. Also, when you mention your monitor, are you refing to a computer display or an external monitor?

So many things to clarify :)

I "believe" I am in Computer RGB always. The mxf files from the camera (directs via Clip-Browser -no Cineform) appear to be in Computer RGB because if you do a Computer to Studio conversion you see a change but if you do a Studio to Computer conversion on the mxf files you do not see a change. At present, until I learn a better way, I am making NO levels adjustments and NO color space conversion at any time. I edit in 8-bit so the scopes make sense, but editing in 32 bit by eye seems to me to be just as good. I'm not concerned with legal levels for broadcast - just final result on LCD/Plasma HD TVs played from disk.

There is so much talk about these conversion it seems they need be used at some time I do not know when. Maybe I will find out when I use my Cineform HD but have not run it yet.

[Bill Ravens]

Originally Posted by Michael H. Stevens

I'm not concerned with legal levels for broadcast - just final result on LCD/Plasma HD TVs played from disk.

You will, eventually end up with clipping from the monitor that will look like it's the NLE's fault. I hope you aren't charging people money for that.

[Michael H. Stevens]

Originally Posted by Bill Ravens

You will, eventually end up with clipping from the monitor that will look like it's the NLE's fault. I hope you aren't charging people money for that.

Bill: Don't understand what you mean by "clipping from the monitor". Please explain what you mean and more importantly, tell me what I "should" be doing. That would be very much appreciated and I might even go away.

[Bill Ravens]

Mike..

;o) sorry, forgive my rudeness. It's very frustrating that Sony makes this so hard. (OK, it's not all Sony's fault, but, I want to blame them)

DVD players and TV monitors will display colors in the RGB 16-235 range. This is what Sony euphemistically calls "studio RGB", but really, it's "video RGB". The reason broadcast companies have these limits aren't because it's some kind of law. It's because TV sets won't display luma values below 16 or above 100. So, if you deliver a DVD with luma values outside this range, it will clip because the display player and monitor just can't deliver at that value. In fact, some TV sets will go berserk if you feed black levels below RGB16 because that space carries a bunch of sync and timing signals for the TV raster. Now, none of this applies if all you ever do is display on a computer monitor since it CAN and will display 0-254. But, if the LCD display isn't adjusted properly, it will clip anyway, since you've left it no headroom for misadjustment.

OK, what to do. It's really quite simple. If you want to stick to 8 bit, only, that's fine. But, be aware that some codecs like Quicktime, Windows Media(from within Vegas), and all still images come into Vegas in Computer RGB. If you're limiting yourself to working in 8-bit,they MUST be corrected to 8-bit, or you'll suffer image losses.

Rather than keep all this "sometimes I do and sometimes I don't" in your head, it's so much simpler to just look at the histogram for all your clips. If the left end isn't 16, adjust it with the LEVEL FX slider, don't worry about using the preset. If the right end isn't 254, adjust it with the FX slider, don't worry about using the preset.

Can't get much simpler than that unless you want to use the "Broadcast Colors FX", but, it's not recommended because it fixes out of range values by soft clipping.

Understand, please, that we're talking various levels of finesse. In most cases, your average home videographer won't ever see the differences. But, anyone who knows what they're watching, will see it in a moment.

[Glenn Chan]

1) I have an HD camcorder that limits output to 100IRE.

Did I say that all camera records superwhites? I simply didn't say that.

2) Superwhites are not illegal in all editing programs.

By definition, they are illegal. Y' > 235 is not in legal range. Illegal values are illegal... I don't know how else to put it.

With FCP you simply check the box if you want Superwhite handling.

It (the setting under the video processing tab anyways) changes how still images are decoded... turning that setting on will decode still images so that values can end up in the superwhite range (to 'match' other Y'CbCr sources). It's not useful/helpful in my opinion, but that's what it does.

Superwhites are still illegal.

The NLEs with the largest installed base do NOT use GPUs so it's a stretch to claim that GPUs explain "volumes."

I never said that GPUs explain volumes.

The "volumes" you talk about must exist in the data values themselves -- not in a FP unit. I think you are confusing the range of bit-values in data with the computation of colorspace during conversions.

Ok the point is this.

1- Visually/graphically, you can plot out the colorspaces. e.g. you can take 8-bit computer RGB, 8-bit studio RGB, and 8-bit Y'CbCr and plot it out on a 3-D graph. You can pick Y', Cb, and Cr as your axes (or RGB; doesn't really matter). You would see that Y'CbCr has the largest volume, then studio RGB, then computer RGB.

That's what I mean by volume. If you don't have such a graph, then it's not necessarily a helpful way of thinking about things / visualizing things.

2- In the end, it means the same thing. Y'CbCr gamut is larger than studio RGB, which in turn is larger than computer RGB.

So suppose you start you have these two sets of values:
Y' = 254, neutral chroma
Y' = 253, neutral chroma

If you convert those values to studio RGB, you get 253 RGB and 254 RGB. No clipping will happen.

If you convert those values to computer RGB, you get:
(254 - 16) * 255 / 219 --> ~277.12...
(253 - 16) * 255 / 219 --> ~275.96...

Now if you use 8-bit (unsigned int) to store computer RGB, then clipping will happen because 255 is the largest value that a 8-bit unsigned int can represent.

So this is how studio RGB represents a larger range of numbers than computer RGB.

2b- If you use floating point numbers, no clipping will happen because floating point numbers can represent a huge range of numbers. You can look this stuff up in computer science literature.
(*Many floating point implementations use a range of 0.0f - 1.0f; but you more or less get the same results).

For examples, do folks always check Superwhites in FCP?

Yes you have to if you are making a broadcast master. You either use the broadcast safe filter, or adjust the signal (e.g. the VTR's proc amp) while ingesting, or run material through a video legalizer.

[Glenn Chan]

I "believe" I am in Computer RGB always.

No, it's not true. Yeah it would make sense but it's not true. :(
32-bit Vegas project: HDV decodes to computer RGB levels.
8-bit Vegas project: HDV decodes to studio RGB levels.
Material converted to the Cineform codec decodes to studio RGB levels, regardless of whether the Vegas project is in 8-bit mode or 32-bit mode.

It's pretty easy to tell because you can flip between 8-bit and 32-bit modes, and undo/redo that. You'll either see an obvious color shift or no color shift.

The mxf files from the camera (directs via Clip-Browser -no Cineform) appear to be in Computer RGB because if you do a Computer to Studio conversion you see a change but if you do a Studio to Computer conversion on the mxf files you do not see a change.

You should be seeing a change... :/

There is so much talk about these conversion it seems they need be used at some time I do not know when. Maybe I will find out when I use my Cineform HD but have not run it yet.

I've already spelled out the answer. Bill means well but he is not giving you the right answer.

1- Pick a working color space (studio or computer RGB). Suppose you pick studio RGB.

2- Convert all your sources to that working color space. So any computer RGB sources (e.g. still images) need to be converted to studio RGB... on those clips, apply the "computer RGB to studio RGB" color corrector preset.

3- When you make your deliverable, check what color space that codec expects. Suppose you are rendering to Windows Media (which expects computer RGB).
Nest your .veg into a new project, apply a studio to computer RGB conversion to that, and render that out.

[Randy Strome]

Originally Posted by Bill Ravens

It's abundantly clear from his histograms, that his levels were NOT set right! I'm getting so tired of this arguement. Go back and look at his histogram. The highs are out of legal limit and the lows are way above the legal limit. So, in an NTSC display, of course everything looks washed out. Everything must fit in the 16-235 range, not 10-255, not 24-255.

Hi Bill,

Thanks for the response. If the end product will be QT, WVM etc, however, it would seem a good way to shoot. Keep the brights just below the clipping threshold (granted, the example appears to be just above) and back them back down in post. Open in 32 bit or convert to Computer RGB and adjust for a nice full 0-255 histogram. Leave a bit of whites headroom if you wish for viewers who may have poorly (or un) calibrated monitors.

[Glenn Chan]

DVD players and TV monitors will display colors in the RGB 16-235 range.

Bill... that's not really true.

DVDs store values as Y'CbCr. They do not store values as RGB.

From a practical level, you need to make sure the Y'CbCr values have the right levels. So in Vegas, you need to make sure your R'G'B' values are converted to Y'CbCr correctly. This is tricky because the RGB-->YCbCr conversion depends on what the codec is doing. In Vegas, Vegas' MPEG2 codec will behave differently depending on whether the project is 8-bit or 32-bit.

If you want the color spaces converted correctly, some of my previous posts have that information.

[Daniel Alexander]

It's such a shame that an amazing program like vegas would leave it's users in such confusion. I personaly have decided that i don't have time for wrangling colour space, i really do expect that to be sorted by the NLE, maybe im asking too much but either way im leaving vegas as i have done my research and these types of discussions on other NLE's are only about 2 or 3 posts long with clear and logical solutions as opposed to a vegas discussion which is 5-6 pages deep with what seems to be a multitude of different routes to an uncertain result.

I never understood why alot of people working mainly for broadcast stayed clear from Vegas, it's not because it's an uncapable program but i think its asking too much time and too much experimenting for something that should be straightforward, and it doesnt help when new codecs and new recording formats are introduced to add to the "ok so what colour space should i use with this" question. Think i'm going to boot edius up until my new mac arrives.