Lee Varis' introduction to color management

[digitaldog]

Actually that isn't too far from the truth, according to this: http://www.tftcentral.co.uk/articles/pointers_gamut.htm

Look, stating that 99.9% of images fall within Adobe RGB or 90% sRGB is rubbish! Where's the critical thinking?
In order to state 99% of images are anything but images, someone, somewhere has to have access to and analyze all images. I can assure you Lee nor TFT has any of my 10's of thousands of images. I can assure you that drum scans made from my Velvia images exceed the gamut of Adobe RGB in some areas of color space.
Further, this doesn't speak to the gamut potential of raw images!

100% of images encoded or converted to sRGB fall without the gamut of sRGB. 


The  potential gamut of a raw file,(the scene gamut, the raw processor and of course, the controls the user applies to the raw data) plays a huge role as we both (all?) know. In my video on color gamut, I had no difficulty rendering images from raw that exceeded Adobe RGB (1998) gamut for some colors. It's not at all difficult. IF I wanted to show more (or was in Lee's camp, limit them), a few sliders like Vibrance, HSL or Saturation would easily skew whatever results I wanted in terms of the rendered gamut from the raw!


Anyone who states that 99% of images are this or that needs to prove that point when called out.

[D Fosse]

Yes, I know that. The percentage isn't important here and that wasn't my point. But it's a good illustration for putting it in context.

Of course it's no problem producing images with colors way outside Adobe RGB. I get that all the time, and sometimes you do need ProPhoto to deal with it properly.

But a lot of these colors - that you get out of ACR/Lightroom - are sensor/processing artifacts, not real and not realistic at all. Usually the finished image benefits greatly from taming them and bringing them in line with the rest. Go outside in spring and shoot some fresh foliage. I bet you get dark saturated yellows that clip the blue channel completely in anything but ProPhoto. Go outside and look again - that clipped yellow is not what you see at all. So why keep it? It's an artifact.

And thus remapped, it's my experience that most finished images, master files, fit nicely within Adobe RGB. I'm a pragmatist.

But the most important point I'm trying to make is that good color has nothing to do with total gamut volume. Good color, and even more so realistic color, is about relationships. Yes, gamut clipping is real. What I'm saying is it's not a problem, just an interesting challenge. Consider dynamic range. Who can reproduce the full dynamic range of a sunset in a photograph? Can't be done, and nobody complains. What we do is remap. In fact good dynamic range remapping has always been considered a large part of the art of photography.

Over time I've come to question the dogma of ProPhoto. I'm not interested in saturated color as such, and certainly not in gamut volume percentages. I'm interested in good, realistic color.

And please don't put me in there with these rockwells and fongs and whatshisnames. I'm way beyond that.

[digitaldog]

And please don't put me in there with these rockwells and fongs and whatshisnames. I'm way beyond that.

I would NEVER do that! I'm not even putting Lee in that camp. I just think some critical peer review is in order.


Rather than just tell people what to do, I prefer to let them test the waters themselves with their own equipment and based on their own findings. It's why I don't tell people you have to use ProPhoto RGB, I provide them a test file and let them see how it all pans out. I attempt to let them know what's going on with the Adobe raw processing engine and how it's based on a ProPhoto RGB gamut. If they want to encode every and all image into sRGB, fine with me. I've said often that an sRGB workflow isn't going to produce poor quality images or output! I've attempted to provide some facts and methods where people can test these workflow items on their own. Once they do, I'm not at all interested in convincing them to change. What I refuse to do is tell people BS like 90% of 'real world' images are in sRGB. No more than I tell people 90% of all images are shot on DSLR's with 35mm sized sensors and those sensors are too small, and they should use a 4x5 view camera with a scanning back. I think it's a disservice to an audience to suggest that high bit editing is a waste, you can't see the difference (it depends on a lot of factors). Or that all graphic cards are 8-bit anyway. That's simply untrue. But I guarantee you people will post on the net that there's no reason for high bit data and all cards are 8-bit anyway because they saw it on a video. Or that 90% of images fall into sRGB without really thinking about how silly and impossible a statement that is. The actual percentage isn't important. What's important is recognizing some people are attempting to educate others by making up 'facts'. It's not necessary unless the presenter is either unfit to provide the content or has an agenda.

[jrp]

OK but he did show some pictures that you might think had much wider gamut than sRGB / Adobe RGB and that turned out not to be the case.  He also pointed out that using a color space where you have finer grained control over the colors that you are actually using is an advantage (that has to be set against the disadvantage the that you point to, that some of the colors may be clipped in scenes that have particular characteristics).

[digitaldog]

OK but he did show some pictures that you might think had much wider gamut than sRGB / Adobe RGB and that turned out not to be the case.  He also pointed out that using a color space where you have finer grained control over the colors that you are actually using is an advantage (that has to be set against the disadvantage the that you point to, that some of the colors may be clipped in scenes that have particular characteristics).

Correct and the video does have some merit mixed with some statements that don't. It's not black and white.

[Peter_DL]

OK but he did show some pictures that you might think had much wider gamut than sRGB / Adobe RGB and that turned out not to be the case.

… and even when it happens that the scene gamut and captured gamut are wider than sRGB, we have to acknowledge that:

>> … all the camera makers are using highly tweaked methods of getting the camera color into either sRGB or ARGB. They are using non-standard color transforms that massage out of gamut color into the JPEGs. Ironically, neither ACR nor LR can match the hand tuned color conversions from camera color to sRGB or ARGB in the same way using standard color space transforms because normal RGB>RGB color spaces are all locked into RelCol rendering-unless you get the beta V4 ICC profiles for color spaces from color.org.

…both Thomas Knoll and Eric Chan (ACR engineers) have tested and proven that the camera color (raw) to on-board JPEG in sRGB and ARGB are "non-standard" renderings with hand tuned tweaks to "make gamut clipping better". Neither Canon nor Nikon will admit to this of course, it's all "secret sauce".
It's pretty easy to test, just shoot a raw + JPEG, and try to take the raw and convert to sRGB or ARGB in ACR/LR...neither ACR nor LR are capable of taming out of gamut colors to match the OOC JPEGs. It's actually something Thomas and Eric are "looking at" (meaning researching with the eye towards improving).<<.

http://forum.luminous-landscape.com/index.php?topic=92767.msg755676#msg755676

--

[D Fosse]

I've attempted to provide some facts and methods where people can test these workflow items on their own.

Indeed you have and your points are well made and duly taken.

Actually my issue isn't with anything you have said specifically, it just sort of came up in this context. What I dislike is the prevailing wisdom out there that PP is the only color space you should work in, ever. That's a myth in its own right and it isn't true, PP has its drawbacks too. I use it when I need it, and don't when I don't.

Ever heard "ProPhoto is the only color space real men use"? I'm sure that started as a joke, but a lot of people take that dead seriously.

People throwing out blanket statements as gospel without understanding, is something I suspect you dislike as much as I do. So it was probably a bit unfair to say what I said, as a direct response to your post. I just felt it needed to be said somewhere. Anyway, I made my point and that's all I wanted.

[digitaldog]

Ever heard "ProPhoto is the only color space real men use"? I'm sure that started as a joke, but a lot of people take that dead seriously.

No, that's a new one. Whoever stated is is obviously compensating! 
Indeed, there are pitfalls with ProPhoto RGB. I believe I stated many, many years ago that if there were one perfect RGB working space, we'd all be using it. In fact I said there are no prefect RGB working spaces through an Adobe white paper way back in 2006!
http://www.adobe.com/digitalimag/pdfs/phscs2ip_colspace.pdf
Page 7

[Schewe]

Ever heard "ProPhoto is the only color space real men use"? I'm sure that started as a joke, but a lot of people take that dead seriously.

Actually, I'm pretty sure I coined that phrase at a Photoshop World session :~)

It's also a misguide goal to use a "more efficient" working space for several reasons 1: ProPhoto RGB is the ONLY color space that can complete contain the camera color space and the current fine art ink jet printers and 2: Nobody has ever PROVED that there's a benefit to using a smaller more efficient color space. In the years (since Photoshop 5) when Bruce tested PP RGB https://en.wikipedia.org/wiki/ProPhoto_RGB_color_space (also called ROMM RGB), I've never encounter a problem use PP RGB. Early on Bruce was worried that PP RGB was too big and created his own color space called Bruce RGB (similar to Adobe RGB but with a better Cyan cromaticity) but ended up forgoing Bruce RGB for PP RGB.

[D Fosse]

Nobody has ever PROVED that there's a benefit to using a smaller more efficient color space.

There are in fact several good reasons for using a smaller color space.

• As long as you don't have any clipping, being able to actually see all the information in your file (on a wide gamut display).
• ProPhoto is highly compressed in the shadows, concealing valuable diagnostic information in the histogram. Below is the same file in ARGB and PPRGB. Note the red channel in the shadows, this shows up as an unpleasant cyan color cast in the shadows. One glance at the ARGB histo, and you spot the problem immediately.
• The long-standing Photoshop bug that causes severe shadow color banding in ProPhoto files, with GPU at "normal" or "advanced". The reason is probably, again, the compressed shadows, causing small errors to balloon.

[D Fosse]

ProPhoto RGB is the ONLY color space that can complete contain the camera color space and the current fine art ink jet printers

Oh, and more thing: As I said above, a lot of the highly saturated colors you get out of Lightroom are sensor/processing artifacts. One very common problem area is the blue channel, which I think is a basic inherent problem in the Bayer filtering with current sensors. Just take a random landscape shot and look at the blue channel. There's usually very poor shadow separation and/or muddy highlights - and these two in combination tend to produce some "runaway" colors that go out of gamut quickly. As well as producing some sickly yellowish cast to shadow areas.

I've seen this phenomenon on every Nikon DSLR I've ever owned. And a while ago the was a thread on the Adobe Lightroom forum, by a Canon user who was bothered by the same thing. And it's not Lightroom as such - Capture One shows the same thing.

And yes, there are colors that fall outside Adobe RGB that good inkjet printers can reproduce. But does that make your prints "better"? Seriously. I repeat once again, with emphasis: Good color is about relationships and has nothing to do with fractions of total gamut volume. More saturation isn't always "better", but often just more garish and unpleasant to the eye.

[earlybird]

Good color is about relationships

Josef Albers could not have said it any better. 

[D Fosse]

Yeah, I come from an art school background, so maybe that's why 

[fdisilvestro]

There are in fact several good reasons for using a smaller color space.

• ProPhoto is highly compressed in the shadows, concealing valuable diagnostic information in the histogram. Below is the same file in ARGB and PPRGB. Note the red channel in the shadows, this shows up as an unpleasant cyan color cast in the shadows. One glance at the ARGB histo, and you spot the problem immediately.
• The long-standing Photoshop bug that causes severe shadow color banding in ProPhoto files, with GPU at "normal" or "advanced". The reason is probably, again, the compressed shadows, causing small errors to balloon.

I don't completely agree with these statements

1.- Low values of red or any color in a histogram do not necessarily translate into shadows as well as any high or clipped value in the histogram do not translate necessarily into highlights, since you cannot prove that the channels have those values simultaneously. This is a common mistake some people make and try to solve clipped channels by underexposing when the cause might be saturation outside the color space. In addition, even if they were related to shadows, you don't have certainty of where in the image are those shadows.

I see this more an issue of the lame histogram available in most applications (why can't they give us something like RawDigger?) than a problem of the color space itself.

For example: the cyan patch in the color checker 24 (last one in the third row) is outside of the gamut of sRGB and the red will be zero and the histogram shows a peak also at zero, as shown in the following image, and it has nothing to do with a shadow (The cursor was over the cyan patch, but it was not captured in the screen shot, ACR configured for sRGB)



I find reading out values in the info palette in Lab mode, a more reliable way to identify color casts in deep shadows or extreme highlights.

2.- The photoshop bug might be due to the compressed shadows, but as I see it, that is a problem of the existing tools and not of the color space per se

Good color is about relationships and has nothing to do with fractions of total gamut volume. More saturation isn't always "better", but often just more garish and unpleasant to the eye.

Absolutely, I could not agree more.

[D Fosse]

Re 1. - I know that and you're absolutely right. But that's not the case here, no saturated colors. What you see in the histogram is just the shadow compression.

In any case I didn't take the time to look for the ideal example, I just took a random shot that I happened to have handy. This is something I have observed over many years.

Below is another totally random example showing the shadow compression.

[digitaldog]

Good color is about relationships and has nothing to do with fractions of total gamut volume. More saturation isn't always "better", but often just more garish and unpleasant to the eye.

Agreed but...I'll state this again because I've seen this on actual prints, I've seen it using ColorThink on images: Many of these colors fall outside Adobe RGB (1998) and when we encode into such a color space or smaller gamut, we can clip the colors to the degree that smooth gradations become solid blobs in print, again due to the dissimilar shapes and differences in how the two spaces relate to luminance. So the advantage of ProPhoto isn't only about retaining all those out-of-gamut colors it's also about maintaining the dissimilarities between them, so that you can map them into a printable color space as gradations rather than ending up as blobs.

[D Fosse]

True and I don't dispute that. I hate clipping as much as the next guy, and severe clipping is never a pretty sight.

What I'm saying is we need to deal with it - or I need to, I should say - at an earlier stage, before it becomes a problem. There's no need to take highly saturated colors at face value, as if the more saturated, the more important to keep at all costs.

Anyway. I'm not going to turn this into "for-or-against ProPhoto". I'm neither, it has its use in the toolbox. But there's no free lunch either, there's always a compromise.

I really don't have much more to say on the matter. I suppose Jeff has some additional viewpoints.

[digitaldog]

Anyway. I'm not going to turn this into "for-or-against ProPhoto".

The point in the context of this discussion is Lee isn't doing that. But that's not the biggest issue some of us have with the video.
I'm somewhat surprised the fallout about his ideas on high bit editing isn't being discussed either...

[Schewe]

I'm somewhat surprised the fallout about his ideas on high bit editing isn't being discussed either...

Hum, I guess I'll need to look into what Lee said and get back to him :~) What did he say? That you don't need to edit in 16 bit?

[digitaldog]

What did he say? That you don't need to edit in 16 bit?

In a nutshell, yup. Last part of the video.