Xrite Color Checker Passport issue?

MarkR

So I got me one of them newfangled Xrite Color Checker Passport gizmos and dutifully created a dual illuminant dng profile for use with Lightroom 3.

Initial tests on some older images were favorable compared to the built in Camera Profile and Adobe Standard. However I have run into some images in which applying the profile creates very blotchy color noise and becomes nearly impossible to white balance.

Here's some examples:

With Adobe standard Profile, even at High ISO (no flash allowed at the museum) I can mostly neutralize the whites and keep the colors relatively cleans:



Moving to the xrite profile, I get colored mess. (Note, noise reduction settings are the same as above, and pretty hefty.) This image I have yet to successfully neutralize the whites as I did above:



Click on the pictures to view them at full size.

WB seems to be ~2500/+16 which is a little beyond the tungsten that I used for the dual illuminant. Is that the issue? Do I need to go back to the museum to create a custom profile? :scratch

MarkR

Here's a better example-- ISO 1600, direct flash on a red shirt. Here's how Adobe Standard renders it:



Here's the Dual Illuminant Color Checker profile:



See the weird yellow-orange flecks that got added in to the shirt?

arodney

That certainly doesn’t look anything like the results I have from the product. Looks like a major white balance issue with the target?

Note too, that even after you apply the custom DNG profile, you still may need to white balance the shot itself. But if you treated both images the same with the exception of the application of the DNG profile, it seems like you may need to build a new one. The results I get are much better than the canned profiles.

MarkR

I would never discount "user error" from the possibilities. :lol

I did create a new dual-illuminant profile at ISO 3200. Applying it gave better results than the original target, not as good as the canned profiles.

MarkR

And, as a follow-up, I just used the Adobe DNG editor dual-illuminant tool, and things are much better. Still some color noise getting exaggerated in high ISO shots, but not to the level that the xrite software was providing. Why? I don't know!

insanefred

I still never can understand why you need to use something to calibrate a raw converter to get reasonably accurate results.
Shouldn't the raw converter to this already.

arodney

insanefred wrote: »

I still never can understand why you need to use something to calibrate a raw converter to get reasonably accurate results.
Shouldn't the raw converter to this already.

Its not really about accurate but pleasing color or visually expected color.

The colors on the Macbeth should render a certain color appearance. The profiles (or previously, tweaking the Calibration tab) produced a closer visual match. There are also differences between various cameras of the same model so a profile that I make for my 5DMII may not produce the same color appearance as a raw from your 5DMII.

insanefred

arodney wrote: »

Its not really about accurate but pleasing color or visually expected color.

The colors on the Macbeth should render a certain color appearance. The profiles (or previously, tweaking the Calibration tab) produced a closer visual match. There are also differences between various cameras of the same model so a profile that I make for my 5DMII may not produce the same color appearance as a raw from your 5DMII.

But can't you accomplish the same thing by using the color palette and create your own color and tone curve to start out with or quickly load somehow?

MarkR

insanefred wrote: »

But can't you accomplish the same thing by using the color palette and create your own color and tone curve to start out with or quickly load somehow?

Well, yes. That's the whole idea. Run the ColorChecker profile once, automagically create a profile, set as default, and "fuggeddaboutit."

It's only when you run into problems, like I did, that it stops being a time saver and starts being a hassle.

I think I did figure out that the ColorChecker created profiles were, for whatever reason, way too saturated, and that was probably causing some or all of my problems. Until I talk to xrite I won't know whether its a problem with me, my system, or their software. However, in the meantime I was able to create a very nice DNG profile using the Adobe tool and the Color Passport. My foliage is now slightly greener, my reds are slightly punchier. It's not a huge difference, but it's enough.

arodney

insanefred wrote: »

But can't you accomplish the same thing by using the color palette and create your own color and tone curve to start out with or quickly load somehow?

You could do the same thing more or less using the Calibrate tab which was the process before DNG profiles were introduced. Or you could try using HLS controls or other rendering controls to match the appearance expect they happen in a different path in the image processing. DNG profiles and calibrate take place before the rendering controls above them take effect. Its like starting with a better baseline prior to setting the individual controls.

malch

MarkR wrote: »

It's only when you run into problems, like I did, that it stops being a time saver and starts being a hassle.

The manner in which you shoot the calibration target is moderately critical. Not sure if that's related to your issue buy maybe it will help others.

1. Set up the target such that it is evenly illuminated. I placed mine against a larger grey background. Then I was able to move the eyedropper around inspecting the RGB values to confirm that the lighting was in fact perfectly even all around the target.

2. You also need to avoid reflected glare and the like.

3. You also want a really good exposure. I shot a bracketed sequence and selected the one where the white patch on the target was as close as possible to 250:250:250.

If you start out with a nice clean capture, you're going to get a better profile out at the end. But, like everything else, if you stuff garbage in... you know what comes out ;-)

In any event, a profile created with a well controlled capture performed visibly better than my earlier (shamefully rather sloppy) attempts. So I recommend taking your time and a good deal of care over that critical exposure.

MarkR

malch wrote: »

The manner in which you shoot the calibration target is moderately critical. Not sure if that's related to your issue buy maybe it will help others.

1. Set up the target such that it is evenly illuminated. I placed mine against a larger grey background. Then I was able to move the eyedropper around inspecting the RGB values to confirm that the lighting was in fact perfectly even all around the target.

2. You also need to avoid reflected glare and the like.

3. You also want a really good exposure. I shot a bracketed sequence and selected the one where the white patch on the target was as close as possible to 250:250:250.

If you start out with a nice clean capture, you're going to get a better profile out at the end. But, like everything else, if you stuff garbage in... you know what comes out ;-)

In any event, a profile created with a well controlled capture performed visibly better than my earlier (shamefully rather sloppy) attempts. So I recommend taking your time and a good deal of care over that critical exposure.

Yes, it looks like the yellow square is somehow getting clipped, leading to oversaturation and general nastiness. Having said that, I'm not sure what to do to prevent that clipping. I'll try again and see what happens. I also have a trouble ticket open with Xrite.

EDIT: I take it back about the yellow channel. I don't know what the heck is going on. Profile looks fine at very low ISO. Anything higher and everything goes to hell. .

malch

MarkR wrote: »

Yes, it looks like the yellow square is somehow getting clipped, leading to oversaturation and general nastiness. Having said that, I'm not sure what to do to prevent that clipping.

Check out these terrific detailed instructions from gollywop at DPReview:

http://forums.dpreview.com/forums/read.asp?forum=1039&message=35818268

MarkR

malch wrote: »

Check out these terrific detailed instructions from gollywop at DPReview:

http://forums.dpreview.com/forums/read.asp?forum=1039&message=35818268

Thanks! . Will read through tomorrow. Having said that, somehow I managed to create a profile with the Adobe dng maker that's looking pretty sweet right now. But I'll see if I can do better with the xrite tools later. (Tempting fate, that's my motto! :lol)

insanefred

arodney wrote: »

You could do the same thing more or less using the Calibrate tab which was the process before DNG profiles were introduced. Or you could try using HLS controls or other rendering controls to match the appearance expect they happen in a different path in the image processing. DNG profiles and calibrate take place before the rendering controls above them take effect. Its like starting with a better baseline prior to setting the individual controls.

Hmmm... I have read somewhere that a DNG profile is actually a second from the based color.
And you are working with the raw data, so the LCH should be on the same level, right? It's not like your doing layers.

Or maybe I just don't "get it"

arodney

insanefred wrote: »

Hmmm... I have read somewhere that a DNG profile is actually a second from the based color.

I don’t understand what you are saying.

DNG profiles (and the Calibrate sliders) are applied in a different part of the raw processing chain by design.

You can alter the controls in any order you wish, but the recommended order is top down. LR and ACR will always apply the processing in an optimal order. Altering sliders before altering Calibration or even loading a DNG profile is allowed but the processing order is fixed. You will generally find that setting the DNG profile/Calibration sliders first, then the other sliders will result in getting to your rendering goals faster, its not a two steps forward, one step back, then one step forward approach.