Help: Sony Color Curves Endpoint Workaround?

fausseplanete wrote on 9/27/2010, 9:46 PM

I use Levels. Agree about rounding errors but that's what's practical.

I use nested projects. Rather than "fight the system", the immediate edit works in the intersection of Vegas and typical-camera space, namely 16-255 (oh yes!). Then I nest that project in one or more "packaging projects" (my term) for each kind of target, e.g. DVD (16..235) or WMV (0..255). It's in these projects, at master level, where I do the Levels-mapping and, for 16-235 deliverables, BroadcastColors plus an underlying track of (16,16,16) color block (since BroadcastColors bizarrely and worryingly doesn't fix transparents or mask featherings to (0,0,0).

Incidentally for grading I use Levels gamma (slider) in conjunction with color curves, since their combined use gives better subjective results where extreme correction is needed (less tendency to make light flesh have pizza tone). Much experimentation led to this method.

Real shame there's not yet an option to set project properties to 0/255 or 16/235.

farss wrote on 9/28/2010, 12:25 AM

You can move the start and end points to any value you desire. To do this first click in any of the white space then you can move them.

Bob.

Serena wrote on 9/28/2010, 12:50 AM

There is a Glenn Chan plug-in for setting the 16-235 limits in colour curves.

farss wrote on 9/28/2010, 1:06 AM

There is?
All I know of is a sample .veg file that contains curves that are setup to the limits. What I described above is covered in his article here, about 1/3 of the way down the page.

Bob.

Serena wrote on 9/28/2010, 1:18 AM

Bob, now I have to say "I believe there is one". I've had a quick search of various VASST tutorial DVDs and haven't found it where I thought it would be. Obviously I use the method you described. Maybe in the VASST library of plug-ins.

Andy_L wrote on 9/28/2010, 10:32 AM

fausseplanete,

With nested projects aren't you losing the ability to correct clip-by-clip? That seems like kind of a big disadvantage. ??

KSmith wrote on 9/30/2010, 7:03 PM

More of a question than an answer --and from a perpetual newbie:

Taking Bob's suggestion, I put a black to white gradient on the timeline and using video scopes tuned the curves endpoints to 16-235 and saved as a curves preset. Would this do what you want (at least as a limiter)?

Keith

musicvid10 wrote on 9/30/2010, 7:08 PM

"You can move the start and end points to any value you desire."

The question was answered in Bob's one-sentence response.
A fundamental understanding of what Curves does that is both similar and different from Levels is necessary.

Andy_L wrote on 10/1/2010, 7:48 AM

You can't actually move the endpoints to any value you desire. You can only move the endpoints along the y-axis, which automatically alters the image.

Glen Chan's curves preset comes the closest to what I'm looking for -- two extra nodes that act as endpoints at the ends of the video levels color space.

Basically, you know you're on the right track if the preset doesn't change the image in any way until you start reshaping the curve. I haven't figured out a way to get nodes at exactly the right spot, but using the gradients media is a good tip. I'll keep playing around with that.

Regardless, it's important to realize that in video levels mode, the curves tool is not acting as you'd expect unless you create new endpoints.

farss wrote on 10/1/2010, 8:08 AM

"Regardless, it's important to realize that in video levels mode, the curves tool is not acting as you'd expect unless you create new endpoints. "

I don't know what you expect but it works exactly as I expect it to.
The Curves are absolutely correct in how they work. The 'curve' plots an output value (Y axis) for every possible input value (X axis). For that very reason you cannot move the end points in the X axis, if you could you'd have input values that did not map to any output value, what should happen then?

By comparison the Levels FX is a bit misleading. The Input Start and Input End values aren't actually that, all values below Input Start are mapped the same and all values above Input End are mapped the same but that is only by implication, you've kind of got to know that. The Curves require you to explicitly map that, as you said by adding another node or two.

Bob.

Andy_L wrote on 10/1/2010, 8:59 AM

I would expect a curves tool's endpoints to represent black & white, and for the curve itself to operate on the legal range of the image. Input values outside the legal range of the video levels color space would map unchanged.

In my opinion, curves is not the tool to use to bring values into or out of legal range. It's for shaping the response of the visible image. I can see why Sony's color curves is built the way it is, given how Sony deals with superblacks and superwhites in video levels mode.

I just don't think it makes for a very good curves tool, because most people are going to assume that when they drag a point on the curve, they're compressing or expanding shadow/highlight detail. But unless you create your own endpoints, you're actually pulling or pushing values in and out of legal (video levels) range.

farss wrote on 10/1/2010, 3:34 PM

"I would expect a curves tool's endpoints to represent black & white, and for the curve itself to operate on the legal range of the image. Input values outside the legal range of the video levels color space would map unchanged."

Nothing inside Vegas works this way. One could argue that everything Vegas does is wrong if you wish but at least it is consistently wrong.

" I can see why Sony's color curves is built the way it is, given how Sony deals with superblacks and superwhites in video levels mode. "

The mode doesn't change how the FXs work as far as I can see. The mode just changes how Y'CbCr values are mapped to RGB.

"I just don't think it makes for a very good curves tool, because most people are going to assume that when they drag a point on the curve, they're compressing or expanding detail. But unless you create your own endpoints, you're actually pulling or pushing values in and out of legal (video levels) range."

Most people would probably make the same assumption about the Levels FX as well. The only safe way to make any level adjustments with Vegas is with an eagle eye on the scopes and with an understanding of how to read them.

Bob.

Andy_L wrote on 10/1/2010, 5:07 PM

Yeah, I see your point. But the levels tool seems more coherent to me. People are naturally going to turn to it to set black and white points, and eventually they'll figure out where that's supposed to be in the Vegas system (hopefully).

On the other hand, people turn to Curves (typically) for contrast adjustments, which if you don't go through some hoops ends up acting a lot like the levels control.

farss wrote on 10/1/2010, 5:22 PM

The power of Curves over Levels is you can make complex gamma adjustments that to the best of my knowledge you cannot do with any other tool in Vegas. If you're really persistant you can even selectively adjust saturation with the Curves.
I certainly agree the Curves FX is not an easy tool to work with. The inability to dial in precise values for the nodes is quite a limitation and the tangents seem to obey some law more akin to chaos theory.

Bob.

KSmith wrote on 10/3/2010, 3:10 PM

Thanks musicvid for the succinct reply. It did get me thinking a bit harder which is always the best result of poking through this forum --I find the Vegas documentation a bit thin sometimes.

I am curious as to why the 'Computer RGB to Studio RGB' preset in the
Color Corrector plugin didn't even rate a mention in this thread.

This topic is of particular interest to me at the moment because I've been asked to supply some YouTube-ready clips, and I ran across some other posts (yours included) about YouTube and Vimeo applying a 16-235 to 0-255 expansion to all video, needed or not.

Is there a reason the Color Corrector wasn't mentioned(?) and Bill, would it need the 'eagle-eye' on the scopes as much as Levels and Curves? (begs the question, Is the mechanism the same?)

It seems to me this is a luminance problem, so the Color Corrector seems a less obvious place to find it than Curves or Levels, but I guess it depends on the point where the two can't be considered separately.

Thanks! This is the best, most pleasant/flameless forum I've run across.

Keith

Andy_L wrote on 10/3/2010, 3:40 PM

Keith, to the best of my knowledge the levels plugin computer-to-studio and studio-to-computer presets do exactly the same thing as the corresponding secondary corrector plugin presets. So it was mentioned already. :)

musicvid10 wrote on 10/3/2010, 3:51 PM

Thanks for the kind words, and I regard the Sony Color Corrector as the most powerful and versatile tool we have, with also probably the steepest learning curve. With forty years of experience doing color correction, I still find it just as easy to mess things up as get it right.

the Studio<->Computer RGB feature is also available in the Levels tool (also Secondary CC), so when the more powerful features of Color Corrector are not needed, it is just as handy to apply it as a Levels adjustment. Really, being able to adjust levels and gamma in one handy tool is often all I need, assuming lighting and WB was anywhere within reason.

The colorspace conversions being fixed-point calculations, experienced users will use them as a starting point, with the scopes and the editor's eyes (as I've so often pointed out) determining the final outcome. Reading a book by the real pioneer of luminance "levels," Ansel Adams, never did anyone a bit of harm afaiac.

One of the nicest features about Vegas, as Glenn Chan is quick to point out, is that the Y'CbCr->RGB->Y'CbCr work is all done under the hood in Vegas, and comes out right 99% of the time. All we have to do as end users is get the "RGB" part right for the format and our artistic purpose.

"Thanks! This is the best, most pleasant/flameless forum I've run across."
I hope you are still comfortable with that statement around the holidays. Recent history has demonstrated otherwise . . .

farss wrote on 10/3/2010, 4:36 PM

Just my two bobs worth.

The problem with just doing a simple cRGB<>sRGB conversion is most / many modern cameras don't play by the rules. Certainly with say stills from a DSC that's all you should need do. On the other hand some video cameras do record Y' = 16 to 255 which conforms to neither cRGB or sRGB.
As such I'm loathe to recommend any simple approach. Knowledge is power especially in this game. If you understand what is going on and are prepared to do a bit of work/experimenting with the advanced tools there's a weath of goodness that can be extracted from images that you might miss just sticking to a boilerplate approach. NLE's are non destructive so no harm done having a go to learn.

Bob.

musicvid10 wrote on 10/3/2010, 4:48 PM

Absolutely spot-on, Bob. And two bobs are better than one ;?)

Not only do most cameras (and their operators) not obey a strict set of rules, but what happens as a result of applying a stock correction to everything misses important artistic opportunities, in almost every case except the proverbial family picnic.

For various moods, settings, scenes, it may be actually desirable to deepen the shadows or burn the highlights a bit. One might adjust the gamma in a backlit scene, for instance. Or compress for all the detail possible in a contrasty snow scene (I know you don't have that problem). Or add contrast to that funny Halloween shoot.

Once the rules are known and understood, playing by them is only for the "PhD" ("Push here, Dummy") types .

KSmith wrote on 10/3/2010, 10:00 PM

Thanks again, for so much, and so fast. Your answers make perfect sense. Also, my apologies to Andy_L since it appears I may have hi-jacked his thread.

I'm not coming at this from laziness and thankfully none of you have accused me of it. I've been messing with still cameras for more years than I'd like to remember, and perhaps my preference for Picture Window Pro will illustrate that I'm not really a preset user at heart. I'm deeply ashamed if it appears that way :-)

I'm also fairly well armed on the audio end, having been a freelance musician for (eek!) over 40 years, and have done a fair bit of digital recording over the last eight, or so.

That said, I must admit that getting a handle on video technicalities seems to be the hardest thing I've ever attempted by at least one order of magnitude. Just in compositing alone, the lack of consistent terminology from book to book is a source of grief, and then the variety of standards, the order in which they developed, what problems they solved, and as you mention, the varying levels of compliance seem to make it an absolute minefield. Being a 'first principles' kind of guy, it's beginning to seem that the best primer would be a really good book on the history of television. Any suggestions?

Getting back to the question at hand, I realize the trap I was falling into was thinking in stills.
I'll spend hours tweaking one still image, and then hope to find a setting for a video that answers for a thousand frames --pretty silly.

This is clearly a remapping job and scaling job. Just watching the histograms over a variety of frames/clips with the 16-235 preset makes it clear that most of the images are ruined by being well below full scale in either space. I have played with the knee function on the EX-1 (terrible white background in one of the clubs I shoot in) and have learned how dangerous luminance compression can be with images. Obviously, (or not) it comes down to scopes, a custom re-map and keyframes.

This makes me all the more curious to see the Glenn Chan plug-in. Can it really work? It seems to have disappeared from Vasst. I guess I'll have to go without.

As usual, by the time you've thought it through well enough to ask the question sensibly, most of the answer is already in front of you.

Thanks again!
Keith

PS: You 're probably wondering what this noob is doing with an EX-1. It belongs to another musician friend of mine. He likes shooting, hates editing and is a bit of a technophobe. It lives at my place most of the time. I do feel lucky. Hopefully, I can help him pay for it!
k

KSmith wrote on 10/3/2010, 10:15 PM

Musicvid:
"I hope you are still comfortable with that statement around the holidays. Recent history has demonstrated otherwise . . ."

Heh. Oh, yes! I've seen them. But compared to the hi-jinks on on some of the audio forums, it's been pretty mild and the grammar is better here --even in mid-flame!

Also, your point about Ansel is well taken. I have the whole series and his portrait is on my wall. He was absolutely one of the finest human beings to ever grace this earth.

Keith

farss wrote on 10/3/2010, 10:20 PM

"You 're probably wondering what this noob is doing with an EX-1"

I'm a noob and own an EX-1. Great camera however between what you can do in the camera and what you can do in post I agree it is all a bit overwhelming or not...maybe :)

Think light in....light out. Helps me coming from an instrumentation background.

Bob.

KSmith wrote on 10/3/2010, 10:49 PM

"overwhelming or not...maybe"

No maybe there!

Light in-light out appeals and is very much in line my stills approach.
I think what undermines my confidence, and hence my thinking is statements like 'something is done automatically'. OK, but when? Why? So, what am I looking at? It's especially confusing when I don't understand in sufficient detail why this automatic process is necessary.

Maybe I'm getting old. I often find the 'friendliest stuff' is the hardest to use. I need to understand the process or it just frustrates me. Bad consumer!

Bob, it's been my great good fortune to run into really extraordinary people more often than I have probably deserved. I've read quite a few of your posts and I hold you in that regard.The instrumentation background clarifies the "Who is this guy?" mystery to a great degree.

I've spent more time lurking here than you might think. Your posts, along with musicvid's and several others (you know who they are) are always helpful and make forums like this the best part of my web life.

Keith

musicvid10 wrote on 10/4/2010, 6:56 AM

"This is clearly a remapping job and scaling job. Just watching the histograms over a variety of frames/clips with the 16-235 preset makes it clear that most of the images are ruined by being well below full scale in either space. I have played with the knee function on the EX-1 (terrible white background in one of the clubs I shoot in) and have learned how dangerous luminance compression can be with images. Obviously, (or not) it comes down to scopes, a custom re-map and keyframes."

Ever wonder why there is relatively little discussion of lighting equipment, setup, and technique on these forums? Go over to any serious still photography forum and see how much space is devoted to lighting discussions. It seems the young HD generation is largely content to spend a fortune on camera gear and interchangeable lenses, throw lighting decisions to the muses or pretend that good lighting doesn't exist, and then try to "fix" everything in post.

During my photo lab days (actually close to twenty years) we knew who were the pros and who were the wannabees right when the film came out of the soup, without even looking at the name on the work order. And it was always the latter who demanded custom souping, contrast masks, unthinkable color and exposure corrections, custom burning (lots of it), and had plenty of blame to go around for the lab personnel for messing up their precious "art."

"Maybe I'm getting old. I often find the 'friendliest stuff' is the hardest to use. I need to understand the process or it just frustrates me."
Very typical of gifted learners. I tell new math students to memorize and get comfortable using the quadratic formula, then we'll back up and figure out how it was derived from completing the square, a principle most of them can grasp visually. The minefield you alluded to can never be successfully navigated, because no roadmap has ever been created, and one would be impossible to extrapolate from all the different standards, conventions, and practices that have emerged collaterally in the last seventy-five years. That so much of it has been distilled in the Vegas engine is a credit, and should not be a source of frustration.

"Getting back to the question at hand, I realize the trap I was falling into was thinking in stills."
Not really. If you think in terms of "stills multiplied," you are miles ahead of GenXYZ types who don't have a clue about lighting or how to get full range on "film." Hit your angles, scene and lighting changes in post, make a decision (the decision not to do anything always being the default choice), and move on. Tackling the continuity issue then becomes much easier as one gains experience.

There are some notable exceptions, of course. Bob (farss) comes to mind, as do several others who have not yet checked in on this discussion. And some of the younger talents like Patryck are very sensitive to lighting and can create wonderful moods with something as simple as a bare bulb. When the controls in Vegas are free to be used creatively, rather than out of necessity, mediocrity ends and art at least has a chance at succeeding.

I read a few of Adams' books before I was out of high school. With his acclaimed trilogy at $49.95 on Amazon, it may be time for me to find out how much I forgot, or didn't grasp the first time around. And don't apologize for hijacking this thread; it's gone in a good direction.