Vegas Pro 8 Workflow

[Randy Strome]

If I open an 8 bit project, then I have a very washed out looking preview. Adding the Studio to Computer FX creates great looking video. Alternatively, changing to 32 bit makes for very similar looking and pleasing video. If i go the 32 bit route, it does not seem to matter if I start with a 32 bit project or switch to 32 bits after the files are already on the timeline (both work well).

(Side note: 32 bit and computer RGB applied together are way overdone and look poor.)

Now, if I reconvert to studio or 8 bits at any time, I go right back to washed out video.

So I am thinking that 2 workflows may work for me.

1. Work the entire project in 8 bit with Computer RGB FX and render without switching to studio.

2. Work in 32 bit, edit and adjust, and then switch to 8 bit, and then add the Computer RGB FX prior to render.

Your thoughts and comments are greatly appreciated.

[Steve Mullen]

Originally Posted by Piotr Wozniacki

the two conversions are available (i.e. Computer RGB to Studio RGB.

I believe these settings are in Vegas because Vegas doesn't work in YUV as does FCP. YUV has a range of 16 to 234 -- although going up to 108IRE likely pushes the range up to 254. (Never 255.) This is the range you want to STAY in.

I would guess that Computer RGB is good old fashioned 0 to 255.

Studio RGB sounds like a clever way to say "video" RGB. If Sony had used the term "video" they would have had to admit they use RGB not YUV. Which is a good reason to avoid both Premiere and Vegas, IMHO. (RGB should, IMHO, only be used when importing graphics where it is a one-way conversion to YUV.)

What I assume is Studio RGB, is RGB confined to a range from 16 to 234. If I'm correct, you can see how the EX1 VIDEO could be clipped during conversion.

Going from YUV to Computer RGB expands a narrow range of data to a wider range which certainly could wash-out colors.

In other words, either CONVERSION could cause a problem. Worse, because RGB and YUV color-spaces are different -- chroma errors creep in.

All this was well understood a decade ago when the Media 100 provided better images than all other products. The key was YUV in and YUV through and YUV out.

[Bob Grant]

The original workflow will have an issue with clips split across cards.
This is a real pain if using the card reader. You need to have both parts of the clip available for the clip browser to stitch them back together again.
Best way I can find to get this to work is to offload the clips into the one folder prior to doing the mxf rewrap.
If you don't do this the exporter will complain. You'll get both halves of the clip however Vegas will not stitch them back together again without loosing a few frames.

[Randy Strome]

Originally Posted by Raymond Schlogel

In other words, is the added contrast that I see when I change it to 32 bit because that's the way it's "supposed" to look or is it making it that way?

Hi Ray,
I am unsure on this as well. The broader histogram I am seeing in 32 bit or with the 8 bit conversion to Computer RGB appears closer to what I recall seeing in my EX 1 histogram, and if I had begun by opening a 32 bit Vegas project, I would never have asked this question, because my video looks perfect (for what it is) in 32 bit. I was using 8 bit because the manual makes it sound like that is best for my needs.

I am beginning to consider the conversion to computer RGB function to be "emulate 32 bit in 8 bit".

Sony is suggesting switching to 32 bit for color correction (which is essentially what the Studio to Computer RGB conversion is), but I do not see any noticable benefit to doing so.



From Sony:

"Pixel format
Choose a setting from the drop-down list to indicate whether you want to perform video processing (compositing, scaling, previewing, rendering, and most video plug-ins) using 8-bit or 32-bit, floating-point arithmetic.

The 32-bit floating point setting allows greater precision for processing video, but requires significantly more processing power than working with 8-bit video.

Tips:

32-bit floating point is recommended when working with 10-bit YUV input/output or when using xvYCC/x.v.Color media.

When using 8-bit input/output, the 32-bit floating point setting can prevent banding from compositing that contains fades, feathered edges, or gradients.

Video plug-ins and media generators that do not support floating-point processing are indicated by a blue icon in the Plug-In Manager and Plug-In Chooser with this icon in the Video FX and Media Generators windows.

If you're creating a 32-bit project, you can increase performance during by using the 8-bit setting during editing and switching to 32-bit floating point before performing color correction, compositing adjustments, or rendering.

Compositing gamma
When you choose 32-bit floating point from the Pixel format drop-down list, you can choose a compositing gamma value.

1.000 (Linear)
The default setting when you choose 32-bit floating point from the Pixel format drop-down list.

2.200 (Video)
Processing in 8-bit video is always performed using a setting of 2.200. Choose this setting when you want to ensure maximum compatibility with projects created in previous version of Vegas."

[Randy Strome]

Originally Posted by Steve Mullen

I believe these settings are in Vegas because
Going from YUV to Computer RGB expands a narrow range of data to a wider range which certainly could wash-out colors.

...but the opposite is occuring. Conversion to 32 bit or Computer RGB is increasing contast color saturation etc. It is the 8 bit (default) Studio setting that looks washed out.

The reason Studio to Computer RGB is handled as an FX is that it is simply a grouping of color tweaks which add contast, saturation, offset, gain, levels adjustments, etc. Each is user customizable, and the "conversion" can be applied from differnt preset groups (including CC, CC Secondary, and Levels) with many different options from each.

This group of options seems to be designed to quickly optimize video for output aimed at final viewing on a computer monitor.

This thread has been awesome for me, and personally, after 3 days, I am already digging Vegas more than any other editing system I have used.

[Seth Bloombaum]

Glenn Chan is a frequent contributor on the Vegas forum, and has written a series of articles on Vegas. In particular, you might want to have a look at the "Colorspace" article, as well as "V8 Colorspace" and 8-bit vs. 32-bit.

http://www.glennchan.info/articles/articles.html

Without comment on Mullen's pejorative views and guesses (puh-leaze!), he is correct about the use of the studio vs. computer conversions, they are meant to give an easy preset to-from 0-255 and 16-234, that is, video and pc ire range.

Searching "color space" over in the vegas forum will turn up lots of additional info, again Glenn is most helpful. For example...

[Michael H. Stevens]

HERE is part of what I posted on the Vegas forum:

"After more testing I have found that I can get the Vegas scopes to agree with the EX1 histogram if I use:

** An 8-bit project
** And apply the Computer to studio RGB conversion.

But now I get the washing out that Piotr reported, and to correct it needs the good old "S" curve in CC..... Do I forget the scopes and use 32-bit and computer RGB or do I go 8-bit and use computer RGB or do I go 8 bit Studio RGB and then have to color correct every image from the camera?"

AND so I ask in addition to what was asked above, does this mean the EX1 was made to produce a computer RGB color space and not a studio one, and if so, why?

[Bill Ravens]

First of all, both the JVC HD110 and the Sony EX1 use a REC709 color space that includes the equivalent of RGB0-255. This is the spec for YPbPr.....that's what they use, all of the new HD(V) cams.

As Steve so pointedly, if not a bit afrontingly, said, Vegas uses the RGB color space. The 32 bit float option in Vegas is SCS's bandaid attempt to bring Vegas up to date. I've had very little trouble using vegas with either the EX1 or the HD110. It's a question of applying the Levels FX in the appropriate way, if you want to use 32 bit float. If you do a conversion beforehand with Cineform, for example, the CFHD files import into Vegas in the RGB color splace and you don't have this conundrum of what to do. I'm not using the CFHD intermediate, because I'd rather wrangle the colorspaces.

I'm beginning to think that the right course of action is to dump Vegas and move to Edius Broadcast. As Dr. Mullen says, it's in the right native colorspace to begin with.

[Bob Grant]

Bill,
what you're saying is interesting however let's not forget that a camera 'sees' RGB and all display devices display RGB. More cameras are coming along that record RAW sensor data. I'd suggest that Y'CbCr and it's problems may one day be a thing of the past. Perhaps the problem with Vegas is that it's way ahead of its time. It'll be interesting to see how the various NLE's around cope with Sony's xvColor.

[Bill Ravens]

Bob,
I could be wrong, but, I don't think the cameras see in RGB. They see in YUV. Actually, the sensor, per se, sees in RGB, but the data it sees is written in YUV....a somewhat lossier version of RGB. Vegas has to transpose this format back to to RGB for its use. I submit this is a problem, no matter how you slice and dice it. FCP and Edius both work in native YUV 4:2:2, Sony is left in the dust of legacy RGB.

[Steve Mullen]

All cameras capture as RGB which is immediately converted to YUV. YUV is video. Graphics are RGB.

I'm sorry for the negatives on Vegas. I just feel an $8000-$10000 camcorder demands spending $1000 for software. As I said, this is only my OPINION.

Several years ago I asked the Adobe folks how Premiere handled peaks above 100IRE. They said YUV 235 was mapped to RGB 255 and so the upper 8 to 10-IRE was lost. Since loss of highlights is already a problem with 1/3-inch CCDs -- this seemed too big a compromise. (I've used Premiere from day one and still like it.)

To me -- clipping and color-space conversion issues are things I don't want to worry about.

(These links try hard to explain Vegas: http://glennchan.info/articles/vegas...lorspaces.html and http://glennchan.info/articles/vegas...r/v8color.html )

PS: EX1 and 4.6

4.6. reads EX1 mp4 native files. What a dream! There are 2 workflows, depending how quickly you need to edit.
FASTEST WAY
1. Insert SxS card into Express slot.
2. Open Sony clip browser 1.1, open mp4 files on card.
3. Drag them to Edius 4.6 bin.
4. Done. You're reading off the SxS card.

BEST WAY
1. Insert SxS card into Express slot.
2. Copy the card's entire BPAV folder to your hard drive. (This will take a couple of minutes)
3. Open Sony clip browser 1.1, open mp4 files in downloaded folder
4. Drag them to Edius 4.6 bin
5. Edit!

http://www.canopus.com/products/EDIUSPro/index.php

http://www.canopus.com/canopus/press/whatshot.php

[Glenn Chan]

I believe these settings are in Vegas because Vegas doesn't work in YUV as does FCP. YUV has a range of 16 to 234

I believe FCP can work in both YUV and RGB. In RGB you can get clipping happening. (By RGB, I mean the values passed between filters; internally, the filters can work in whatever color space they want.)

Legal range for Y' is 16-235 (not 234).

If Sony had used the term "video" they would have had to admit they use RGB not YUV.

The phrase "studio RGB" occurs frequently in Vegas and strongly suggests that Vegas uses RGB.

Which is a good reason to avoid both Premiere and Vegas, IMHO. (RGB should, IMHO, only be used when importing graphics where it is a one-way conversion to YUV.)

Pretty much the majority of image processing is done in R'G'B' color space. If your pipeline is 8-bit, then it can be advantageous to handle Y'CbCr (mistakenly called YUV). Premiere and FCP can do this I believe.

If you are working with 24/32-bit float, then it doesn't really matter quality-wise whether you go with Y'CbCr or R'G'B' (though for other reasons, you would choose R'G'B'). The range and precision of 24/32 floating point numbers is pretty much good enough that you incur negligible loss when converting between color spaces (and you usually have to convert color spaces anyways).

What I assume is Studio RGB, is RGB confined to a range from 16 to 234. If I'm correct, you can see how the EX1 VIDEO could be clipped during conversion.

It's actually the opposite that is true.

For 8-bit color spaces:
8-bit Y'CbCr has the largest "volume". Because the volume is large, the bits are spread out thinly (so theoretically you'd have greater banding issues; in practice you rarely ever see this).

8-bit studio RGB has the second largest volume. Some weird illegal values will get clipped (any value with extremely high Y' and non-neutral chroma). Some cameras generate those values.

8-bit computer RGB has the least volume.

So when you convert between these colorspaces, clipping can occur (when going from large volume to smaller volume). Loss of precision will occur.

32-bit color spaces:
You don't really get clipping because 32-bit floating point numbers have a huge range.

Going from YUV to Computer RGB expands a narrow range of data to a wider range which certainly could wash-out colors.

Computer RGB is a smaller range.
Neither computer RGB nor studio RGB results in washed-out colors if you handle the color space conversions correctly. (Granted, in Vegas, this is a manual process.)

Worse, because RGB and YUV color-spaces are different -- chroma errors creep in.

Y'CbCr ultimately is converted to R'G'B' for display. There are errors in the sense that rounding errors do occur.

All this was well understood a decade ago when the Media 100 provided better images than all other products. The key was YUV in and YUV through and YUV out. Not to start a platform war, but frankly, given the low cost and high-performance of today's Intel Macs -- I can't understand why folks are still playing around with XP/Vista and IMHO non-pro NLEs. (Or, at least switch to EDIUS which was designed by folks who understand video.)

From a technical standpoint, you'll get the highest quality from taking all your sources and working with 32-bit float (or 24/64/80 if you want to get academic; higher is better). Those values can be R'G'B' or Y'CbCr (for design reasons, no one is going to choose Y'CbCr).

Any GPU-based system nowadays is going to be working with 24 or 32-bit floating point R'G'B'. (I think that includes Edius.)

1b- Most NLEs actually handle chroma subsampling incorrectly. FCP, Vegas, Premiere are ones I've tested (all handle it incorrectly / not according to the letter of Rec. 709 or 601).

2- From a real world standpoint, you can watch TV (or heck, even the broadcast master) and not figure out which NLE was used to finish. The only exceptions are certain trademark effects/filters or if you are using an old Avid (and you pull that material into a NLE and see all the compression artifacts).

The bit depth stuff mostly doesn't matter, unless you need to do linear light operations.

Several years ago I asked the Adobe folks how Premiere handled peaks above 100IRE.

They now handle processing in Y'CbCr too.

[Steve Mullen]

Originally Posted by Glenn Chan

Pretty much the majority of image processing is done in R'G'B' color space.

If you are working with 24/32-bit float, then it doesn't really matter quality-wise whether you go with Y'CbCr or R'G'B'.

Other than color-space conversion errors -- there's no problem with RGB if your YUV video is limited to 235. With some cameras it is. With the EX1 it isn't. You -- I assume -- do not want the upper 8 IRE clipped-off. Thus, if you use Vegas, you need to do whatever it is you need to do to prevent this.

Something seems odd about your use of the word "volume." We know YUV is 16-235 with option to go up to 254. You say Studio RGB has LESS volume. You also say Computer RGB has even less "volume" than Studio RGB -- yet it has a range of 0 to 255. So what range is less than 0-235(254) yet more than 0-255?

Working with FP is potentially MUCH MUCH slower option. Using it as a solution to the lack of YUV seems a slow way to go.

By the way -- I'd love to see a statement from Adobe that they have switched the internals to YUV. I doubt it.

I'd also love to see statements that describe which and when NLE's use GPUs and that FP is used. Liquid does, but for only very limited FX. Media Composer does not.

[Randy Strome]

Wow Glenn,

That was a very interesting read, even though I must admit that a good deal of it passed over my head.

Some here are saying that conversion (in 8 bit) to Computer RGB is muting their colors. I am seeing just the opposite. I get a wider (more correct to what I noted on camera) histogram, and richer colors (when viewing the vegas preview window on my dual calibrated computer monitors). This appears to emulate 32 bit, which also produces a pleasing result and a similar histogram to the EX1's. Only 8 bit studio RGB looks washed out, with a narrow histogram. Am I missing a setting?

[Piotr Wozniacki]

NO, you are not missing anything Randy. Let's straighten this up:

1. the colours are richer, and scopes wider, in 32bit/2.22 gamma, than in 8bit project settings

2. the colours are richer, and scopes wider, after applying the Studio to Computer RGB conversion preset - even in an 8bit project.

If somebody really can see the opposite happenning, please tell us more about their other settings cause this sounds wrong!