Ad
Ad
Ad
Pages: [1] 2 3 »   Bottom of Page
Print
Author Topic: Camera Calibration for Copy Work  (Read 6145 times)
RogerEle
Newbie
*
Offline Offline

Posts: 31


« on: July 21, 2013, 09:28:56 PM »
ReplyReply

Hello everyone, I am hoping for some advice as to what steps I might take to improve my current camera profile, or any other workflow suggestions ... I just don't want to waste money or time on anything that won't make much difference - and looking at all of the different products our there for camera color management that all promise improvement and so on ... Wink ... just  hoping for some guidance to help sort this out.

I have a portrait studio, we do lots of copy work of old photos.  Most of the photos we get are faded / old / dirty, so we are fooling with the colors and tones no matter what.  20% of what we get we are hoping to match the original.  We can currently do that pretty close with a few tweaks - current profile skin tone is a little more red and a little less yellow and the blues / greens are not as strong as they should be, I can live with this and adjust the images but since these are important colors in a lot of images, I am hoping for improvement.  

We copy everything using a Nikon D3 with a nikon 60mm macro, 80A filter, polarizing filter on camera and polarizing sheets in front of tungsten lights (halogen i think). We have the X-Rite ColorChecker and the Datacolor Studio Kit that includes the SpyderCheckr and the print calibration spectro meter.  A number of years age I created profiles using the adobe dng profile editor with the color checker target and also profiles with SpyderCheckr target and software.  I ended up liking the Spyder better for color, just had to dial the saturation way down after I opened it up in Photoshop.  A week ago I started testing again to see if I could improve things and made some interesting discoveries ... I read in another thread here that the Datacolor system is 'bogus' - I am not trying to defend Datacolor here, just giving a history of what I have and what I have done so far Smiley

For the camera white balance I set the kelvin in the camera, fine tuned and re-shot until I got it as close as possible to neutral.  These are the steps I have done (omitting random testing to figure out what I am doing) Smiley ...

1) Made a profile with the ColorChecker and the X-rite passport software per instructions.  
2) Made 3 profiles with the SpyderCheckr per instructions using 3 different camera profiles, first using Adobe Standard, then neutral and finally mode 1 (since we use that in our studio for our portraits).  Datacolor documentation says it doesn't matter which camera profile you use, just use the same one you created the profile with - not what I found, there is some variation between each profile created.
3) I then read the ColorChecker patches with my spectro and recorded each of the lab values (they were pretty close to the published values on the X-rite site).  Used Photoshop in Adobe (1998) space to determine RGB values for each of those readings.
4) Opened ColorChecker image in each profile in ACR in the Adobe (1998) space (which adjustes the HSL slides in ACR, set the camera profile to the same one used to create the 'profile', used Tone Curve / Point Curve to create a curve to match the actual readings from the spectra readings for the grey values across the bottom of the chart (#19 to #24) - saved each as a preset.  The greater the contrast of the camera profile the bigger the curve ended up being to reverse it.

All 4 presets are slightly different visually when clicking back and forth in a few of the color patches - no difference perceived in the grey patches.  When physically looking at the ColorChecker none of the profiles match #2 and the blue green in the upper right corner - the #2 skin tone patch looks more red and the blue green patch looks different (writing this from home - don't remember what direction it is off).  All of the new profiles do give me more separation in the shadows so that is a good thing! Smiley

My budget is up to around $400 for improvements if appropriate, I am looking for suggestions on whether I should purchase a different target and/or on using/purchasing different profiling software.

Thanks in advance!
~Roger
Logged
Vladimirovich
Sr. Member
****
Offline Offline

Posts: 1320


« Reply #1 on: July 22, 2013, 10:12:17 AM »
ReplyReply

topic popcorned...

rawdigger (-> CGATS), (CGATS -> argyll), rpp
« Last Edit: July 22, 2013, 10:13:55 AM by Vladimirovich » Logged
Tim Lookingbill
Sr. Member
****
Offline Offline

Posts: 1158



WWW
« Reply #2 on: July 22, 2013, 10:51:36 AM »
ReplyReply

Quote
All 4 presets are slightly different visually when clicking back and forth in a few of the color patches - no difference perceived in the grey patches.  When physically looking at the ColorChecker none of the profiles match #2 and the blue green in the upper right corner - the #2 skin tone patch looks more red and the blue green patch looks different (writing this from home - don't remember what direction it is off).  All of the new profiles do give me more separation in the shadows so that is a good thing!

My budget is up to around $400 for improvements if appropriate, I am looking for suggestions on whether I should purchase a different target and/or on using/purchasing different profiling software.

You mention two color patches that are off and you now want to solve this by throwing more money at it because you think correcting these two patches is going to reduce the added time of color correcting? Save your money.

Outside of what the CCchart looks like how off are your actual Raw captures using the profiles as is from what you see under those halogens which have a dominant reddish spectra compared to daylight flash. Because of this it is to be expected you would see red in #2 patch and blue green patch off due to the halogens diminished blue spectra.

How about switching to flash or use a Solux 4700K setup? Did you set your illuminant for 2800K in the PE CCchart Wizard? Or did you create a dual illuminant camera profile?

I can tell you this from experience shooting a lot of stuff in my studio in and around next to my display with my 6 year old Pentax K100D DSLR with available light from window light to GE Soft White 2800K and daylight balanced flash and flotubes, I rarely have a problem getting an exact match to what I see to what's on my display. It does take some post processing, but then I'm not setting up my camera as a copy stand. If I did, then I'ld have FAR LESS post processing to do between shots. Fix one frame, save settings and apply to the rest.

The most trouble I have is shooting materials whose color gamut is beyond my camera's AND display's ability to reproduce namely deep, rich and vibrant cyans (turn to a dark baby blue). Skin tone can come out slightly green or red but can easily be tweaked with HSL. Sometimes I just use ACR's default "ACR 4.4" profile and get more accurate results.

You're shooting photos that may have chemical residue that can reflect back spectra you're eyes don't see but the camera records. A profile is not going to fix this. Profiles expect uniform spectra from both subject and light. The CCchart's patches are created with uniform spectra.

A DSLR is not a precision scientific instrument so you can't expect perfection and consistency shooting all types of reflected materials.

« Last Edit: July 22, 2013, 10:53:13 AM by Tim Lookingbill » Logged
Vladimirovich
Sr. Member
****
Offline Offline

Posts: 1320


« Reply #3 on: July 22, 2013, 10:56:49 AM »
ReplyReply


How about switching to flash or use a Solux 4700K setup?


he can just gel halogen lights
Logged
Vladimirovich
Sr. Member
****
Offline Offline

Posts: 1320


« Reply #4 on: July 22, 2013, 11:14:50 AM »
ReplyReply

We copy everything using a Nikon D3 with a nikon 60mm macro, 80A filter, polarizing filter on camera and polarizing sheets in front of tungsten lights (halogen i think).

about the filter 80A on lens to correct halogen - here is your brother in arms = www.luminous-landscape.com/forum/index.php?topic=80130

PS: just skip the feud between Iliah and his best friends  Grin
Logged
opgr
Sr. Member
****
Offline Offline

Posts: 1125


WWW
« Reply #5 on: July 22, 2013, 11:15:40 AM »
ReplyReply

The contrast curve in LR seems to produce oversaturation of reds, yellow, and cyan for D800 files. Could this be a factor in your observations for the D3 as well? Does reducing the contrast in the RAW converter change the colors significantly (while not necessarily producing a pleasing result)?
Logged

Regards,
Oscar Rysdyk
theimagingfactory
RogerEle
Newbie
*
Offline Offline

Posts: 31


« Reply #6 on: July 22, 2013, 12:03:34 PM »
ReplyReply

topic popcorned...

rawdigger (-> CGATS), (CGATS -> argyll), rpp

Thanks Vladimirovich,
I see on the rawdigger web site a mention of CGATS, if rpp is output parameters for Argyll, it seems I will be able to work it Smiley  ... if not rpp?
~Roger
Logged
RogerEle
Newbie
*
Offline Offline

Posts: 31


« Reply #7 on: July 22, 2013, 12:21:16 PM »
ReplyReply

How about switching to flash or use a Solux 4700K setup? Did you set your illuminant for 2800K in the PE CCchart Wizard? Or did you create a dual illuminant camera profile?

Haven't researched Solux, but am open to it ... I am not copying paintings or watercolors, but if it is a better light source for photos I am all ears Smiley
Didn't use the Adobe Profile Editor on this round of testing, used the Xrite Passport profile builder, no illuminant setting that I could see, and no with the one color of light source I wouldn't know how to make a dual illuminate camera profile (not trying to solve calibrating to a range of color temp light sources)

Quote
You're shooting photos that may have chemical residue that can reflect back spectra you're eyes don't see but the camera records. A profile is not going to fix this. Profiles expect uniform spectra from both subject and light. The CCchart's patches are created with uniform spectra.

Don't expect that to be fixed, but when I see it on both the CC Chart and on original it becomes a light source / camera sensor / profile issue though, not one of chemical residue.

Thanks Tim,
~Roger
Logged
Tim Lookingbill
Sr. Member
****
Offline Offline

Posts: 1158



WWW
« Reply #8 on: July 22, 2013, 12:22:55 PM »
ReplyReply

You want a visual match using lights (halogens) that don't deliver a match to D50 and that your eyes see totally different to D50 reference numbers? And you want all that to be precise? Precise to what? D50 numbers or what you see with your eyes? So how do you know you have a match?

I just rattled off a shot of one of my self promo 4x6 one hour minilab color print on glossy Kodak Royal Digital silver halide paper. It's a perfect match shot under daylight flotubes which have a spikey green spectra which WB correction indicates with a huge shift to ACR magenta tint slider).

I used incamera AWB and adjusted in ACR 4.6 using default ACR4.4 profile. I get pinky skin also clicking for R=G=B but the differences between the warmer eyeball WB adjust and the precision version only affects the amount yellow in my own skin but not much in the skin tone restoration shot.

And your expecting precision color using halogen lights?
Logged
RogerEle
Newbie
*
Offline Offline

Posts: 31


« Reply #9 on: July 22, 2013, 12:28:15 PM »
ReplyReply

about the filter 80A on lens to correct halogen - here is your brother in arms = www.luminous-landscape.com/forum/index.php?topic=80130

PS: just skip the feud between Iliah and his best friends  Grin

Thanks Vladimirovich,

Took a quick look, with my untrained eyes it was hard to tell where said feud starts and stops, but I will give it a read.  My in camera white balance is 4260K with the 80A filter ... I read through the first page of 9 pages, will get to the rest tonight ...
Logged
Tim Lookingbill
Sr. Member
****
Offline Offline

Posts: 1158



WWW
« Reply #10 on: July 22, 2013, 12:39:40 PM »
ReplyReply

You're making this harder than it has to be.
Logged
Vladimirovich
Sr. Member
****
Offline Offline

Posts: 1320


« Reply #11 on: July 22, 2013, 12:43:53 PM »
ReplyReply

Thanks Vladimirovich,
I see on the rawdigger web site a mention of CGATS, if rpp is output parameters for Argyll, it seems I will be able to work it Smiley  ... if not rpp?
~Roger

rawdigger output CGATS /read instructions how to/, then either argyll (command line) or Iliah made a GUI frontend for that purpose = http://sail2ithaki.livejournal.com/188062.html, then feed custom icc profile to RPP... you can always ask technical questions to authors (if that is a good /= technically challenging/ question then you will get a good answer)

« Last Edit: July 22, 2013, 12:46:35 PM by Vladimirovich » Logged
RogerEle
Newbie
*
Offline Offline

Posts: 31


« Reply #12 on: July 22, 2013, 12:47:30 PM »
ReplyReply

The contrast curve in LR seems to produce oversaturation of reds, yellow, and cyan for D800 files. Could this be a factor in your observations for the D3 as well? Does reducing the contrast in the RAW converter change the colors significantly (while not necessarily producing a pleasing result)?

I don't have the spectro readings with me, I will look tonight so that I can compare the effect on saturation of reducing the contrast using the tone curve ... my memory is that it reduces the saturation to close to matching the color checker spectro numbers.  There will be a slight varience because the tone curve is built by hand from only the 6 grey patches allowing for a little human error.

Not sure which way to take 'pleasing results', if I was photographing real world scenes the profile/settings for copy work would look flat and very unpleasing - shooting copy though of a perfect original I will be more 'pleased' the closer they match the original.  Smiley

Note that the polarizing filters used increase the contrast, so some of the reason for the lowering of the contrast is the polarizing.  (from memory) the darkest grey square number 24 comes up from around 30 to 54 rgb, patch 23 is close to the same, a dip in the lighter tone with little change in the lightest tone at patch number 19

Thanks!
« Last Edit: July 22, 2013, 12:59:18 PM by RogerEle » Logged
RogerEle
Newbie
*
Offline Offline

Posts: 31


« Reply #13 on: July 22, 2013, 01:01:48 PM »
ReplyReply

rawdigger output CGATS /read instructions how to/, then either argyll (command line) or Iliah made a GUI frontend for that purpose = http://sail2ithaki.livejournal.com/188062.html, then feed custom icc profile to RPP... you can always ask technical questions to authors (if that is a good /= technically challenging/ question then you will get a good answer)



Thaks, it looks like rpp = RAW PHOTO PROCESSOR and is mac only ... I am on pc, is there a second choice? Smiley
Logged
Vladimirovich
Sr. Member
****
Offline Offline

Posts: 1320


« Reply #14 on: July 22, 2013, 01:06:37 PM »
ReplyReply

Thaks, it looks like rpp = RAW PHOTO PROCESSOR and is mac only ... I am on pc, is there a second choice? Smiley
well...  I run RPP on PC/Win using VmWare... in your case then try to find the right method for repro work with ACR/LR (people do that - so there is a way) - we have a number of experienced people here (see nicknames in that feud between Iliah and others in the thread above) who master that art or rerpoduction work using just colorchecker + Adobe PE and proper target illumination.
« Last Edit: July 22, 2013, 01:08:40 PM by Vladimirovich » Logged
RogerEle
Newbie
*
Offline Offline

Posts: 31


« Reply #15 on: July 22, 2013, 01:34:02 PM »
ReplyReply

You want a visual match using lights (halogens) that don't deliver a match to D50 and that your eyes see totally different to D50 reference numbers? And you want all that to be precise? Precise to what? D50 numbers or what you see with your eyes? So how do you know you have a match?

You must be implying that halogens with and 80A filter doesn't get me there? ... Wink
I first look with my eyes and then check the numbers to make sure it is not a viewing issue ... my poor brain needs to translate the rgb color to hsl to really make sense of it so that can be a slower process Wink ...

It may be that the polarizing filter adds some of the contrast that you are not seeing in your test.  So many of the originals I see are improved with polarizing that I sure do not want to lose the polarizing from my normal workflow ...

I appreciate your test but am confused ... are you saying in both methods you used you get a match to the original that is better that using tungsten with an 80a filter?  And they are better because your test image matches close except for your skin tone?  It is not conclusive without comparing the same test with tungsten lighting and an 80a filter ... I am not suggesting you need to do this testing for me, just that it doesn't really prove a point.

It does suggest that I would be smart to create a profile using flash with the polarizing filters (I have flash, would need to purchase the solux), if there is a difference in the profiles created then that will prove your proof of concept if I am understanding you correctly.

Thanks!
Roger
Logged
opgr
Sr. Member
****
Offline Offline

Posts: 1125


WWW
« Reply #16 on: July 22, 2013, 01:40:50 PM »
ReplyReply

Do you happen to have a test-shot available for us to try? Something including the CC from which you made the profiles?
Logged

Regards,
Oscar Rysdyk
theimagingfactory
RogerEle
Newbie
*
Offline Offline

Posts: 31


« Reply #17 on: July 22, 2013, 01:55:37 PM »
ReplyReply

Do you happen to have a test-shot available for us to try? Something including the CC from which you made the profiles?

Sure this is the .nef, I use Google drive personally so I made the file public with this link ...

https://docs.google.com/file/d/0B-YoFPgh003pUzZwV1RIRWlPNVk/edit?usp=sharing

If there is any consensus that I should first shoot the CC with flash with polarizing rather than the tungsten I am happy to before anyone goes to any work on my behalf.  I do really appreciate the effort!

~Roger
Logged
opgr
Sr. Member
****
Offline Offline

Posts: 1125


WWW
« Reply #18 on: July 22, 2013, 03:15:51 PM »
ReplyReply

Okay, this is what I get from a calibrator using default plain-vanilla ICC conversion in linear space.

The Gray1 image shows the result of the gray patches. It does show a contrast increase. Normally I would associate this with an unevenly lit chart. If you know this is a result of the polarizer, then that must be accepted as is, otherwise I would double check whether the chart is evenly lit.

I wonder though what is the advantage of using the polarizer? I get a distinct impression that the benefits may not outweigh the lower consistency of calibration.

Logged

Regards,
Oscar Rysdyk
theimagingfactory
RogerEle
Newbie
*
Offline Offline

Posts: 31


« Reply #19 on: July 22, 2013, 03:49:33 PM »
ReplyReply

Okay, this is what I get from a calibrator using default plain-vanilla ICC conversion in linear space.

The Gray1 image shows the result of the gray patches. It does show a contrast increase. Normally I would associate this with an unevenly lit chart. If you know this is a result of the polarizer, then that must be accepted as is, otherwise I would double check whether the chart is evenly lit.

I wonder though what is the advantage of using the polarizer? I get a distinct impression that the benefits may not outweigh the lower consistency of calibration.



Very interesting, I have never used calibrator software so this is all new to me ... Smiley  Thank you!

Regarding the polarized lights and filter - I started running a b&w custom lab in the early 70's, 14 years of that and then copy and restoration work from when we opened our studio in '86 ... so many tests and so many times polarizing has saved my butt!  Polarizing fixes photos with a texture to the surface of the paper (the top of each bump in the texture reflects the light source and that becomes part of the image without polarizing) or silvering (dark areas of the print take on a metallic silver glow).  Even photos with wrinkles and tears handle better with polarization. Sometimes this is subtle, sometimes a huge difference but it is always appreciated!  So no, just got to figure the best with polarizing.

I did photograph the target with the polarizing filter on the camera turned to let in all light (effectively off).  I have the .nef file if it is some use, I tested it so I could see if there was a difference ... my memory is that the colors were very similar once the point curve reversed the increase in contrast.  As I always use polarization I thought it would make more sense to concentrate on getting that right Smiley

Thanks again,
Roger 
Logged
Pages: [1] 2 3 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad