Colour of light matters

[Tim Lookingbill]

Tim Lookingbill has them because he (Tim) was claiming that it (digital capture) was "color metrically accurate"... shall be so (to accurately capture "pleasing" painting)

No, I didn't say the digital capture was color metrically accurate. I said notice the differences (the alternating hues) in the color renderings of shadow detail in the painting compared to the digital capture.

It's the differences that point to an understanding about optics (in the painting) established by the masters on how to render color that would appear to have depth and three dimensionality that is not in the original scene=colormetrically accurate.

[Vladimirovich]

It's the differences that point to an understanding about optics (in the painting) established by the masters on how to render color that would appear to have depth and three dimensionality that is not in the original scene=colormetrically accurate.

I am sorry - paining looks very flat to me...

[Iliah]

> I had to over simplify Kodak scientist's studies because

you do not know those studies.

[Tim Lookingbill]

Below shows on the left a colormetrically accurate=(exactly what my eyes saw) version of food items I shot under 100watt GE Soft White tungsten bulb using my Pentax K100D's Tungsten WB preset and viewed with ACR's defaults.

I couldn't believe how ugly the image looked because when I kept looking back at the scene next to my display the food looked fresher and more vibrant. That's because my eyes already adapted to the amber color cast of the 2800K tungsten lights.

To support the adaptive effect as the cause for the mismatch I turned away and stared into a dark corner under the table not lit by the tungsten bulb and waited for my eyes to adapt and then looked back at the scene and sure enough I got a match between the ugly ACR preview shown in the sample below and the actual scene.

The scene changed color so how the heck is DeltaE numbers going to help me if my eyes don't see what a spectro does.


Now I had to make the image look "pleasing" (version on the right) but didn't know where to start because the actual scene now looks like crap. I had to use my memory of the colors I saw of greens, reds and tans from lettuce, tomato hamburger buns I saw in old '90's Dairy Queen commercials of flying hamburgers and DQ ice cream cones and manipulated the HSL tables and WB to trick the eye into seeing fresher looking food. Spectros can't do that so DeltaE's are useless.

My point behind this is that ACR's tools are simple and act on the color in such a way that deliver quicker "pleasing" results when relying on edits similar to this in order to fight the adaptive effect. Other Raw converters I've used do things to the color with regard to hue/sat and WB during edits that makes this a slower process for me.

[Iliah]

So you have some dE values to prove that or this is your subjective interpretation?

Red channel on the palm blown out prematurely.

[digitaldog]

Red channel on the palm blown out prematurely.

What? From what image, the one downloaded from the web page? Don't see it. I see all three channels blown out in the sky area (so what)?

[Iliah]

> From what image, the one downloaded from the web page? Don't see it

That explains it all.

[Tim Lookingbill]

> From what image, the one downloaded from the web page? Don't see it

That explains it all.

Nah!...My last post explains this subject better than all your posts combined.

[Iliah]

I am sorry - paining looks very flat to me...

Well, the purpose of the thing was a little different http://www.janblencowe.com/blog/maine-plein-air-landscape-painting-a-meritorious-entry/

[Iliah]

Nah!...My last post explains this subject better than all your posts combined.

Yes, "a colormetrically accurate=(exactly what my eyes saw)" made my day.

[digitaldog]

> From what image, the one downloaded from the web page? Don't see it

That explains it all.

You make lots of assumptions and pass em on as fact (like the primary issue you have conveniently ignored all day, this "problem" with ACR). That may fly when you post on DP Review, but doesn't wash with the LuLa audience as well. Anyway, you wrote:

Unfortunately the digital part of it looks exposed and processed in a wrong way.

You have Tim's raw? If not, you once again made a big assumption and passed it off as a fact. You could have no way to know anything about the exposure. It could have a stop or more latitude. Hence my question to you (From what image, the one downloaded from the web page?) which you conveniently haven't answered, like most of the questions presented to you. Again, you have the raw? You're making points about exposure and development based on Tim's post to the web? Is this yet another digression (like the one on the Fraser piece) to ignore your lack of any facts presented here in terms of this invisible ACR problem?

[Tim Lookingbill]

Here's another demonstration of how some Raw converters follow colormetrically accurate constructs of color appearances that shows up in their edits I see quite a bit and is the reason I like ACR's color engine which bases its color appearances with regard to contrast and densities from a more color purity construct.

Below are links to two versions of subtractive color mixing samples of cyan, magenta and yellow which were first established and implemented into color film processes and dyes.

The first is an actual photo of these purities in the form of individual filters viewed on a light table...

http://www.cis.rit.edu/fairchild/WhyIsColor/images/SubtractiveColorDisks.jpg

The link below is a synthetic RGB representation of the same filtering process and color mixing with special attention to densities and how more richer they appear than the ones above...

http://www.jiscdigitalmedia.ac.uk/images/cmy_01.gif

If you had a choice between the two colorants and their palettes in order to create a painting, which of the two would you choose? I pick the synthetic version for its more dense and rich colors in the shadows and vibrance in highlights which mimics the behavior of very expensive dyes on paper including inkjets.

If you can't figure this out from this, then you're just too dense to teach to.

[Iliah]

> You have Tim's raw? If not, you once again made a big assumption

I do not need the raw to know exposure is off on this one (if there even was a raw), and I also know that perfect exposure was never the goal for this shot. I know this shot was measured "evaluative", and it can't be perfect. By the features of the image I can say I doubt the camera can record raw, as it must be a rather compact one; and being compact it should not allow a lot of DR to get the whole scene DR into the shot without help of filters, especially given the fairly high sensor density.

I did not download the web image, I see no need for it. But you can, and if you will - select R255 and see. For me the colour gradation on the palm is enough to say red is saturated above correct, and some red must be at full 255. And also, if you will download - check EXIF.

[Tim Lookingbill]

Yes, "a colormetrically accurate=(exactly what my eyes saw)" made my day.

Yes as defined by lining up the Lab numbers to the CCchart pictured below which is colormetrically accurate by the numbers as well as visually accurate. That's what I mean when I make the connection when a machine's numbers actually match up to what I see which is rare because my eyes keep adapting to every scene I photograph which are not CC charts.

Did that make your day?

[Vladimirovich]

Here's another demonstration of how some Raw converters follow colormetrically accurate constructs of color appearances that shows up in their edits I see quite a bit and is the reason I like ACR's color engine which bases its color appearances with regard to contrast and densities from a more color purity construct.

again - there are a lot of nice images from ACR, but a lot of shitty ones ("I see quite a bit")... says nothing about ACR, more about the person behind it... the same it true about other raw converters.

[Vladimirovich]

Yes as defined by lining up the Lab numbers to the CCchart pictured below which is colormetrically accurate by the numbers as well as visually accurate. That's what I mean when I make the connection when a machine's numbers actually match up to what I see which is rare because my eyes keep adapting to every scene I photograph which are not CC charts.

Did that make your day?

we are getting close... now an actual raw file shot under a regular domestic (like some average Joe has hanging somewhere) fluo coupled with conversion parameters (ACR) & Adobe Standard (from OEM distribution) will certainly put a nail in Iliah's coffin.

[Tim Lookingbill]

we are getting close... now an actual raw file shot under a regular domestic (like some average Joe has hanging somewhere) fluo coupled with conversion parameters (ACR) & Adobe Standard (from OEM distribution) will certainly put a nail in Iliah's coffin.

Getting close to what? I don't understand what you're getting at.

[Stas Wilf]

Boiling this quarrel down to real problems, the only one I can see is that ACR engine determines which profile to use from the color temperature, specified at development stage. It means that:
- it would use the profile built for daylight for shots made under LEDs or cold fluorescent lights;
- it would use the profile built for daylight for shots made in twilight or in open shade, where CCT is much higher than 6500K and color shifts are already noticeable;
- it would use some mixture of two profiles for shots made under warm flourescent lights, which do not resemble neither tungsten nor daylight;
- it would use some mixture of two profiles depending on the color cast that its user finds pleasing at the moment and introduces using WB slider.
Worse, this way of using profiles is embedded into DNG specification.

On the other side, the engine allows us to use any single-illuminant profile we want to use. So this seems to be a problem only for uneducated users who, as usual, are left to enjoy the results of the process they do not completely understand.

[digitaldog]

I do not need the raw to know exposure is off on this one

Of course you don't. You just assume and speculate and pass it off as a fact. You can't have any idea about the camera original, raw or JPEG without having the camera original. But why go down another of your digression designed rabbit holes? In the course of one day, you've provided at least three examples of making things up and passing them off as facts, the most egregious being the original flat earth theory you continue to ignore: the invisible 'problem' with the ACR engine that we can't see. We'll skip the awful mistakes you had to correct numerous times in Bruce's piece (being friends of late Bruce (G-d bless him) never corrected technical inaccuracies in his CameraRaw article? I pointed those out many times) when what he wrote isn't incorrect. Now we are to believe that simply viewing a web image, you know all about the capture, exposure and processing of an image without even looking at it's histogram. Basically the bread crumb trail you've built in writhing here is easily dismissed as you are unwilling to back up your assumptions with fact. That leads me to believe, for the first time on this forum, you have a significantly large BS factor! So why go on? When (IF) you can get back OT and show us how the image I posted as per your request is a perfect example of the problem, and not just this one, I'll attempt to examine if you've provided an ounce of science or a pound of BS. Based on all your postings today, the later would not surprise me a lick.

[digitaldog]

Boiling this quarrel down to real problems, the only one I can see is that ACR engine determines which profile to use from the color temperature, specified at development stage.

ACR/LR doesn't determine this, it's a setting that is sticky OR someone sets it (upon import with a preset etc). And of course anyone can change this profile at any time. But yes, someone can set a profile that's inappropriate for the capture as I did using a daylight profile shot under Fluorescent light (and there was no negative issue I could see after white balancing and applying tint/temp season to taste). So this so called ACR problem is a huge mystery to me and apparently Iliah who can't explain it other to say it's there. A problem that is valid that you point out are users who do things they shouldn't or as Vladimirovich correctly states, produce shitty images due to a lack of understanding how to use the product.