Color checker passport dual illumination on Canon 350D / Rebel XT
KMBphotographic
Registered Users Posts: 17 Big grins
I have recently started using an X-rite Color Checker Passport with my Canon 550D (Rebel T2i) to make a dual illuminant profile and it works remarkably well. However, I have not had the same success with making a dual illuminant profile with my Canon 350D (Rebel XT, but has Kiss Digital N branding on the outside)...
...the error message I get when I try to follow the same method on my Canon 350D as I (successfully) followed for my 550D reads: "Could not generate profile because one or more images have unsupported illuminants". I have tried using different pairs of illuminants (all of which worked for my 550D) such as flash + tungsten, flash + fluorescent, cloudy + tungsten etc, and setting the white balance to the type of light (flash, or tungsten etc so that the EXIF keyword corresponds to the correct type of illumination).
Does anyone know if the header/EXIF information is written differently in the earlier Canons compared with the later ones? I am wondering if the Color Checker Passport software isn't correctly understanding the header/EXIF information from the older camera, but what may argue against this is that the software does a very good job with making a single illumination profile for my 350D (astoundingly good actually, and this 350D has been infra-red adapted so there is no longer an infra-red blocking filter on this camera - everything looks jolly red unless corrective measures are taken).
I should be very grateful if anyone can shed any insights into how I might be able to successfully generate a dual illuminant profile for my 350D from the X-rite Passport software.
Many thanks, Katherine
...the error message I get when I try to follow the same method on my Canon 350D as I (successfully) followed for my 550D reads: "Could not generate profile because one or more images have unsupported illuminants". I have tried using different pairs of illuminants (all of which worked for my 550D) such as flash + tungsten, flash + fluorescent, cloudy + tungsten etc, and setting the white balance to the type of light (flash, or tungsten etc so that the EXIF keyword corresponds to the correct type of illumination).
Does anyone know if the header/EXIF information is written differently in the earlier Canons compared with the later ones? I am wondering if the Color Checker Passport software isn't correctly understanding the header/EXIF information from the older camera, but what may argue against this is that the software does a very good job with making a single illumination profile for my 350D (astoundingly good actually, and this 350D has been infra-red adapted so there is no longer an infra-red blocking filter on this camera - everything looks jolly red unless corrective measures are taken).
I should be very grateful if anyone can shed any insights into how I might be able to successfully generate a dual illuminant profile for my 350D from the X-rite Passport software.
Many thanks, Katherine
0
Comments
Author "Color Management for Photographers"
http://www.digitaldog.net/
No - I hadn't tried this (or indeed come across Adobe DNG Profile Editor previously...). Many thanks for this suggestion - I'll investigate and report back... Actually, just a question - their documentation says to "Photograph [their 24 patch] ColorChecker Chart". Obviously for this exercise I can't photograph my monitor, but presumably I can simply use the X-rite ColorChecker 24-patch Passport (the arrangement of colours seems to correspond between both the Adobe and the X-rite charts)?
Many thanks again,
Katherine
Author "Color Management for Photographers"
http://www.digitaldog.net/
DNG files from my 550D work perfectly when fed into Adobe DNG PE.
I'd be grateful for any suggestions of what I could try next.
Many thanks, Katherine
Author "Color Management for Photographers"
http://www.digitaldog.net/
That would be very kind.
Um... I can't attach a DNG file to this dgrin forum (it's not gif, jpeg, jpg, mml, kHz, pdf, png or txt) and when I try to upload to Smugmug it says "wrong file type" (for the 550D DNGs as well).
Should I be able to upload a DNG file to dgrin?
Lightroom-4.2 (for both my 350D and 550D exports from RAW to DNG).
Many thanks,
Katherine
What settings used for conversion?
Author "Color Management for Photographers"
http://www.digitaldog.net/
I've just PM'd you with a link.
In File Settings, I used the default parameter values, namely:
Image Format: DNG
Compatibility: Camera Raw 5.4 and Later
Embed Fast Load Data selected
Embed Original Raw File unselected
Many thanks,
Katherine
Author "Color Management for Photographers"
http://www.digitaldog.net/
The gross off-whiteness is because it's an infra-red adapted camera (doing an initial white balance in Lightroom didn't alter any of the behaviour I saw when attempting to build profiles).
Did you succeed in building a dual illuminant profile? I could get the single illumination to work, but it's the dual I had all the problems with and was wanting to overcome.
Many thanks,
Katherine
That alone could be the issue with the errors you got about balance.
I only ended up with one DNG so I didn't make a dual illuminant profile. You can't do that in Lightroom anyway, you have to use the Passport software.
Author "Color Management for Photographers"
http://www.digitaldog.net/
This often happens in the Adobe DNG tool because either a) you forgot to drag the little color dots to the corners of the Colorchecker before trying to build a profile, (I know this because a ... friend ... had this problem once) or b) there is some other thing going on that is preventing the Adobe tool from identifying the color checker patches correctly.
My own limited experience is that the Adobe profile creator will give a much more neutral look, one that should theoretically be similar to the Adobe Standard profile. X-rite profiles seem much contrastier and saturated to my eye.
Jeff Schewe had a wonderful post on some message board (that I can't find) about how Adobe creates it's profiles. The take-away for me was that the Adobe profile is usually preferable, unless the sample camera Adobe used to create it is significantly different in color rendering for your own camera. I actually had this happen to me when I owned a Pentax K20d. The canned standard profile was so far off that every image needed some sort of color correction, and skin tones were a nightmare-- until I created a custom profile. Now I've moved to Nikon and I find that the Adobe Standard to be better than anything I've been able to create in terms of being able to be applied across a range of images.
Katherine, almost certainly, and as Andrew says, the IR converted camera is the issue. While you don't mention who did the conversion, the type of conversion ("IR plus visible" vs "IR only", for instance) or the level of conversion (if "IR only" it may be either near IR or deep IR), it's likely that you will get almost no blue-channel output if it's an "IR only" conversion. Even an "IR plus visible" will have a distorted sensitivity compared to a standard visible light camera.
I'm not sure that creating a dual-illuminant profile for that body has any practical merit for IR digital photography.
I have a Canon XT/350D, with a 720nm Life Pixel, "IR only" conversion. (Their "Standard" IR conversion.)
If you could detail what conversion you have, and who did it, and if you can further explain your intended use for the camera, we might be able to describe a methodology and technique to give you desirable results.
Moderator of the Cameras and Accessories forums
I bought the camera from a friend in its adapted state and I'm not sure which company did the conversion, but I believe the conversion comprised just the removal of the IR blocking filter inside the camera so that it is now IR+visible. When I do IR photography I put an optical blocking filter on the front of the lens (e.g. 720nm or 860nm).
My hope was that if I could colour calibrate the camera in its IR+visible adapted state that it would then become a useable visible camera as well: the colour profiling wasn't intended for IR photography per se. The single-illuminant profile strategy certainly works rather well (I'll upload four images presently that show this) demonstrating that the images from this adapted 350D are colour-calibratable, so my question was why should there be problems with the dual-illumiant strategy. I was hoping to generate a dual-illuminant profile for this camera rather than having to generate a fresh single-illuminant profile for every different lighting situation I would be shooting in.
A camera profile only really makes sense when the camera used is working to the range of sensitivity and limits of the profiling software. An IR+visible camera will produce copious additional red channel in most lighting situations, skewing the results of profiling software designed for visible-only sensors. Profiling for the visible-only component of the sensor (with the IR cutout filter on the lens) should allow the software to do its job within its design parameters.
Using the visible-only profile on a visible+IR capture should result in a fairly normal blue channel, with slightly to somewhat saturated green channel, but highly saturated red channel (because there is no IR channel, so the IR information is translated as red channel.)
An additional custom curve for both green channel and (mostly) red channel should give you some control over the IR component of the image. The custom curve set would probably vary by the IR to visible ratio, which should vary by the lighting situation. You would further adjust the red curve by the intent for the image.
One problem I should mention is that most IR+visible bodies are not AF compensated for the IR component. Using the body for IR only or IR plus visible will probably not properly AF the IR subjects. This may add to the IR "glow" of IR+visible captures, but IR only captures will likely need manual focus.
Moderator of the Cameras and Accessories forums
Doubtful. I think the rabbit hole you've been lead through with DNG Profiles is based on this assumption.
Author "Color Management for Photographers"
http://www.digitaldog.net/
The fourth image I uploaded (marked "CCP then WB") shows that the profile works well when generated from a single illuminant - and hence becomes a useable visible camera. What puzzles me is why a single illuminant gives a good calibration for this camera but that dual illuminants should fail.
Happy New Year,
Katherine
Interesting thought... but wouldn't this calibration then only be appropriate to use for images shot with an IR-blocking filter in place?
I agree that the red channel would be higher than the green and blue ones, but would it inevitably be saturated? I checked the histogram that the red channel wasn't clipping.
Yes indeed - I have a lower fraction of IR keepers than that from my visible-only camera!
Happy New Year,
Katherine
"Hi [Tech Support person's name],
My question is whether a visible-light-plus-infrared dSLR is compatible with the X-Rite Color Checker system? In other words, can you do dual-illuminant profiles which give meaningful results regarding a camera with a sensor capable of both visible and infrared sensitivity?
Unfortunately, I don't see any mention of infrared in any of your Color Checker documentation, including the PDF you included in your response. As such, it's probably not fair to draw conclusions.
Please consider forwarding this request for information to your research and development and product engineering departments, since this is a rather unusual area of expertise.
Thanks for your kind attention,
[ziggy]"
... to which they responded:
"Case Subject: Dual illuminant profiles
Product:
Dear [ziggy],
In an effort to resolve your support Case, the following comment has been added. Please review this comment, and respond as indicated.
Comment: We have not done any testing of the dual-illuminant mode with infrared light conditions. This would not be a common situation for the target audience for ColorChecker Passport users.
The ColorChecker Passport is designed for people who shoot RAW in either Lightroom or Photoshop, and will create a DNG profile that these Adobe programs can use.
You can certainly try to build such a profile, but we have no "yardstick" for you to use in terms of determining "how accurate" such a profile would be.
Thank you,
[Representative's name]
Technical Support Rep - Level 2
X-Rite Incorporated
4300 44th St SE
Kentwood MI 49512 US"
I interpret this as meaning that the Color Checker system is not designed for, and does not take into account, a camera with InfraRed sensitivity, including a camera with both Visible and IR sensitivity.
My own thought is that, while you may encounter "some" situations where the Color Checker system seems to work fairly well (and I qualify those situations as lighting with little to no IR component and subjects with little to no IR reflectance), you will most certainly encounter situations of lighting with high levels of IR "and" subjects with high IR reflectance, and your Color Checker system profiles will provide unusual to even unpleasant results in some of the second scenario (IR rich lighting with IR reflective subjects).
However, some of the second situation may be pleasant and even beautiful, since you may experience a non-normal tonality. For instance, human skin may experience a very nice "glow" in a high-IR-plus-visible lighting situation.
I recommend experimentation, but if you truly want to use the Color Checker system to color balance visible light images with that body, then you must use an IR-cutout/cut-off filter. I suggest using an X-Nite "CC1" color correcting filter, although I have no direct experience with it. (Reviews are generally good.)
Moderator of the Cameras and Accessories forums
It's very kind of you to have written to X-rite about this - thanks so much!
I certainly intend to do some experimentation (when I have got hold of an IR-blocking filter for one of my lenses...) and try and figure out how to decouple the illumination-independent calibration of the relative responses of the blue, green and (super-)red pixels in my detector from the different illumination characteristics.
I'll report back in due course... many thanks again.