[I]" I typically use the RGB waveforms as my only reference - is that a good thing to be doing or am I missing the boat here?"[/I]

I assume you mean "RGB Parade"?

If so then they are a tool I have used from time to time.

I wish more cameras had them too as they're a great tool for checking if a channel is being clipped. I do see this kind of scope in a couple of EVFs that are now on the market e.g. the Cineroid (I think, cannot check, both of ours are away for repair )

Bob.

I depends on both, and also RGB Parade to check the Black and White's consistency.

For human skin in histogram, usually you will have more Red amount than Green and Blue amount.

Set

Lovely post.
For levels (not necessarily everything) I generally use 1.
The "relative distribution" has never been an issue for me, as I work knee-to-shoulder, just as forty years ago.

Bob, you lay out some very important questions. One of the most intrusive issues on these question within the VegasPro context is the variation between versions.

I merely have access to Vegas7. 0d, Vegas 8.0b, Vegas 9.0e, and VegasPro 13 Build373.. yet I can tell these versions do play differently in the values and the user interface for these tools.

So then the question number 1) becomes, has the use and definition of these tools improved with maturation, or just merely changed?

2) Has the use of these tools been affected by underlying issues of codecs, compositing, or what other factors?

There is apparently and frustratingly an amount of room for a great deal of variability. I would also like to point out as has another forum user, that using a specific plug-in such as AVColorLab, versus the specific Vegas version (?) in built controls have also a very different outcome in managing video that you might be reviewing in the scopes.

"Not only did I soon realise the Histogram can tell fibs seeing gaps in it freaked me out needlessly when trying to adjust an image."

Any time you increase the contrast, the histogram lines/bars move further apart. There is no information to allocate levels to the pixels that you might expect to appear between these lines, so you see gaps, even if the actual source had a continuous probability density function (PDF).

A histogram is only an approximation to the PDF.

The gaps are added air; we call it banding in the image.

FWIW, my approach is to use a histogram on the camera to aid in correct exposure of the camera, and my 7" viewfinder has this feature built in. I use the principle of Expose To The Right (ETTR), by manually setting the optimum iris setting for best picture performance then adjust the exposure using a variable density filter.

During colour grading, I use the waveform monitor in Resolve to adjust many aspects of the image, black levels, colour, exposure and overall levels, but no other tools.

My aim is to arrive in Vegas with all those parameters nailed down and fixed, so I do not end up using the Histogram or Waveform monitors in Vegas at all.

Geoff

[I]"One of the most intrusive issues on these question within the VegasPro context is the variation between versions."[/I]

Indeed, I have noticed this as well going from V9 to V13. I suspect V13 is right and we've all been getting it wrong for over a decade. Using V13 what I've shot looks much the same on my external monitor as it did in my camera's viewfinder and on an external monitor connected to the camera by HD SDI to a good monitor. I used to think it was me.

Bob.

I use both.....but adjust by the histogram...peek at the waveform...and of course use my own opinion of what I see in my monitors.....and do test renders of sections and plop it in other programs if needing some more scopes and may drop the raw there also in some cases...

"

That would be the changes in Vegas' decoders, probably not the scopes behavior.

Looking at the images in the first post (they don't show on my tablet), I need to add that in order to achieve the tailing effect seen there, a sharpening filter was deliberately added after the levels filter, rather than before as recommended.

If one looks at the effects of the high-pass (sharpen) filter without the levels filter, the effect is a radial "blooming" of all spectrums, including luminance. If one looks at the luminance and composite waveform displays, the slight x-y expansion is visible there too, although one can rightfully argue it is more evident on the histogram display (as well as the vectorscope).

This effect may actually be useful in the production effort. If one is sure to encode their program to comply with the strictest of digital standards (PBS, as one example), it is useful to be notified of chroma slop, in addition to any luminance breaches, well ahead of any delivery deadlines. In that case, the tailing seen on the histogram in the first post would serve as an instant warning, rather than relying on an analysis of the relatively weaker impression of the same noise on either of the waveform displays.

Now, if one actually positions the sharpening filter correctly in the chain, the luminance clamping is visibly evident on both the waveform and histograms, establishing the practical value of leveling last, not first, in any chosen production workflow.

[I]"Now, if one actually positions the sharpening filter correctly in the chain, the luminance clamping is visibly evident on both the waveform and histograms, establishing the practical value of leveling last, not first, in any chosen production workflow."[/I]


1) There's no clamping, only scaling. The Sony Broadcast Colors FX does clamp, the Levels FX does not clamp.

2) What If a user has mixed media e.g. a few stills and compliant video? Wouldn't one apply a levels FX to the stills to conform them to the video?


One thing I haven't mentioned because it only just dawned on me is this:

The Histograms in Vegas are BROKEN

They are not by any definition histograms. I opened one of the same image in IrfanView, found the Histogram thing and got this:



Vegas's Histograms do not any "binning" which is why they produce confusing results. Oddly enough the waveform monitor does which is why it doesn't produce misleading results.

Bob.

ps: sorry to shout but I'm still recovering from a retina examination and my eyes are producing vision with seriously clipped highlights thanks to the medication, oh the irony :)

I put FullRange Shirley (from musicvd10's thread) in IrfanView and Corel Photo-Paint and got almost the same results (why not identical?).

Vegas 13's histograms are similar, but the y-axis is not linear (log perhaps, but how would you present 0?)

The spikes you see in Bob's histogram presumably arise from the overlaid histograms in the copy of FullRange Shirley in the other thread, and which were not present when musicvid10 produced his histograms.

Hi Peter,
If I change Vegas's Histogram to Luminance only it starts to look a bit more like the IrfanView one but there's still a very significant difference. What IrfanView's histogram is showing makes sense looking at the image, Vegas's one doesn't. Oddly enough though they report around the same average value, weird.

Bob.

The important thing is, do the peaks preserve their relative heights to one another and occur in the same positions on the x-axis?

If so, then it is likely that the y-axis in Vegas is some sort of transform of the y-axis we expect (known as the "frequency" in statistics, but not to be confused with Hz frequency). The "count" is another way of describing it.

The transformed version would not change any conclusions you might draw from studying the histogram.

Another interesting thing i noticed was that when I put the image in Vegas and increased the contrast, the R, G and B histograms split up with gaps between the lines, but the luminance didn't until the contrast was very high. I suppose that is because the luminance is a real (non-integer) number and can have values in between the R, G and B integer values.

I don't recall ever needing to use the histo as a precision tool for measuring actual distribution levels, only the presence of pixels at a given value, something at which the display is quite effective. Indeed, there is no scale factor on the y-axis, and the window itself is scalable on two axes, lending itself to all manner of shaping. For precision leveling such as my example above, it is not necessary to know how many pixels are in a given range, only that they are there.

Of course, individual mixed events need to be conformed relative to each other. Stills and generated media need particular attention. In a few instances, a levels filter might be used at a track or event level, lacking a need for something more sophisticated. But this is all being done contextually, that being native RGB edit / preview space. Once all is said and done, one places the final Studio RGB filter on the Output, dead last in the chain. I don't recall ever having said anything else. Regardless of what one calls it, when used correctly the result is a 100% effective levels clamping, that I don't think is duplicated even in the Broadcast plugin, which is really another holdover from analog transmission days.

One thing I didn't mention about preemptive placement of a Computer RGB filter before editing is that everything outside 16-235 is then hidden from view, and one has to lower the levels just to bring the tonal values back into vision. As a lab color corrector from the Woodstock era, i find this duplication in the earliest stages of editing repugnant.

We always cling to our earlier toys the most. I remember clearly how hard it was to give up thinking in VU terms when I began working in digital audio. The thing both digital video and audio have in common is a hard ceiling, something that is not represented in IRE Waveform or VU metering displays. Better in my thinking to let a sleeping dog lie, and learn to think in terms of digital hard limits and to work within them.

As I continue to explore using Vegas Pro, reading this discussion has led me to a conclusion given the last few comments about Vegas Pro's waveforms/histogram feature being broken... It appears, to me at least, that one should not do any color correction/grading in Vegas Pro instead doing that part in Resolve or another similar program.

Am I correct in this or am I missing something in the process of this discussion? Does this conclusion apply to those delivering to the web or not? I've never run into this kind of a situation while editing in Premiere Pro so it only compounds my confusion on using Vegas as a primary NLE.

Well, there's no longer any need for puzzlement over the y-scaling of the Vegas histogram. Yes, it looks different than an image histogram in Irfanview or Photoshop.
No, it is not "broken," and does exactly what it should do in a video environment.

I made a linear gradient wedge in Photoshop against a transparent background.
You will need to match the media settings in Vegas (1818x980).
You can download it here:
http://dl.dropboxusercontent.com/u/20519276/LevelsWrong/Untitled-2.psd

Notice that that the Irfanview histogram is linear. Note also that it measures Luminosity (Y). That's as expected -- no gamma weighting.
The histo in Photoshop looks just the same - linear..


[EDITED] Here's a clearer-eyed explanation: Vegas histo could be just a Common log(base10) histogram. For certain, it's not linear.
http://en.wikipedia.org/wiki/Common_logarithm



The guys at Sonic Foundry who came up with this stuff were not slouches. I use the histo in Vegas for a few things. Leveling is one of them. The waveform and vectorscope have their place in my visual arsenal as well. The RGB Parade is cute.

Thanks Bob, for the opportunity to learn something new from your inquiry. Perhaps I'll have an opportunity to return the favor, and one of my threads will provide an incentive for you to do the same.

@musicvid10 - Again, I'm left with a sense of confusion around all of this - I think it's mainly because of how Vegas Pro manages levels from the initial onset, but maybe if I explain my workflow, you can provide me with a simple answer. This workflow applies for both Vegas and PPro for me

1) I shoot NOTHING but DSLR footage using a custom Cine profile by Vision Color. I record at 24p on my Canon DSLR's.

2) I record dual syc audio at 96khz 24bit for highest quality master audio

3) I copy all files over and rename them accordingly with Bulk Rename Utility

4) I either leave as native h264 footage or transcode to Cineform AVI's with GoPro Studio

5) I import all media into Vegas (or Ppro) and lay on timeline

6) Run PluralEyes to sync scratch audio from DSLR footage with clean audio from audio recorder

7) Save that project with audio sync clips as the base line. Then I edit the project until an initial edit is done, then save that as a new project file. Each revision gets it's own project file after the initial first edit.

8) Given your statement about scopes/Levels/etc NOT being broken as I initially perceived it, this is where I lose the rendering process for Vegas: Once ready to render - what do I apply to properly render levels/values for the Web? For Blu-Ray distribution? For Broadcast? This is the $64,000 question I can't seem to get my thick head to understand - or am I making this complicated when in reality it's very simple?

Your DSLR footage is full range 0-255. It's the nature of the beast. I know nothing about your camera profiles.

For any of the delivery options you mention, as well as almost all others, your footage should be leveled and rendered to 16-235 YUV, to avoid any unwanted clipping in the highlights and shadows.

The method you choose to accomplish this is up to you. I have described my workflow in detail. Other methods also described in the threads you have been reading are equally valid, to the extent that the results are valid. Each method has its advantages and disadvantages. You brought up another valid consideration in one of the other threads -- simply remembering to do that last critical pre-rendering step to conform levels to the delivery colorspace, After you've had a chance to try the methods suggested, do let us know which one you setlle on, and why.

Cliff , I shoot a limited amount of DSLR for second unit to bring into a camera master project. For particular reasons of color space, pulldown, and ease of leveling - to match what is going to happen on the timeline - I always transcode DSLR shot at 24P thru native Cineform to make the necessary adjusts outside of VegasPro. Never ever bring direct DSLR raw to the timeline - we have talked about this much before.

Now as musicvid10 suggests - you will be within the adjustments minor that they are to balance everything carefully in VegasPro rendering. My goal for output is Blu-ray professional - never anything else that I take a lot of time to fuss with.