Generating X-Rite DNG camera profiles for LR. Did I miss something?
Hi, Some of you had warned me that the DNG profiles were not near as good as the proper ICC camera Profiles. Since there is no other way to calibrate the camera sensor for LightRoom, I got the ColorChecker Passport to give it a go. I hastily made some DNG profiles via the X-Rite 1.0.2 export module (out of unprocessed zeroed RAW). Here are my first impressions, tell me please if you think that I must have missed some important step on the way, or if that's simply the way it is. What strikes me, is that contrarily to the ICC profiles I have previously built for Capture One for instance, the DNG camera profiles seem to only affect the chroma values and they leave the camera to RAW gamma curve mostly unaffected. Since the hues aren't automatically pinned at their proper place on a corrected luminance curve, they seem to be floating and the result can be caricatural sometimes. Attempts to reinforce the gamma curve on images linked to those profiles bring up inevitably an over-saturation, since this affects the RGB channels and not a Lab L curve. The profiles leave also the white balance shift uncorrected. Does that first impression merge with what others have seen, or did I miss perhaps some step on the way to producing those test profiles? Paul Schilliger
On May 8, 2013, at 10:59 AM, Paul Schilliger <pschilliger@smile.ch> wrote:
I hastily made some DNG profiles via the X-Rite 1.0.2 export module (out of unprocessed zeroed RAW).
The profiles leave also the white balance shift uncorrected.
Are you using the same "unprocessed zero'd" raw settings? Is there a reason why you're using zeroed raw settings (I suspect you mean all sliders are set to zero? What PV?). Andrew Rodney http://www.digitaldog.net/
Is there someone on the list that knows the goal of the DNG profile? Responding to: Re: Generating X-Rite DNG camera profiles for LR. Did I miss something? I agree with what has been said about the results of using the DNG profile by others. Remember that the DNG profile was made with 'free software' after the purchase of a $100 target system. Yet another CC-24, added to a collection of CC-24s. The problem described with color saturation seems to be a deal breaker for me. The first of two. I had not noticed that the gamma was 'intentionally uncorrected,' only that it was NOT corrected. Coming from someone who always self-neutralizes the grays on the CC-24... THE GRAYS WERE NOT NEUTRALIZED! The second deal breaker. Doing the grays myself takes time. I was hoping for a little new technology to help me out, but no such luck. Is there some use of the DNG profile that I'm unaware? It was said before: "did I miss something." Tim Vitale Conservator Oakland, CA 510-594-8277
Maybe this might be related? (" Hue Twists in DNG Camera Profiles") http://dcptool.sourceforge.net/Hue%20Twists.html http://blog.thomaslesterphotography.com/photography/untwisted-adobe-camera-p... Grtx, René Damkot www.damkot.com www.getcolormanaged.com On 08-05-13 (w 19) 19:57, Tim Vitale wrote:
Is there someone on the list that knows the goal of the DNG profile?
Responding to: Re: Generating X-Rite DNG camera profiles for LR. Did I miss something?
I agree with what has been said about the results of using the DNG profile by others.
Remember that the DNG profile was made with 'free software' after the purchase of a $100 target system. Yet another CC-24, added to a collection of CC-24s.
The problem described with color saturation seems to be a deal breaker for me. The first of two.
I had not noticed that the gamma was 'intentionally uncorrected,' only that it was NOT corrected.
Coming from someone who always self-neutralizes the grays on the CC-24...
THE GRAYS WERE NOT NEUTRALIZED! The second deal breaker.
Doing the grays myself takes time. I was hoping for a little new technology to help me out, but no such luck.
Is there some use of the DNG profile that I'm unaware? It was said before: "did I miss something."
Tim Vitale Conservator Oakland, CA 510-594-8277
_______________________________________________ Do not post admin requests to the list. They will be ignored. Colorsync-users mailing list (Colorsync-users@lists.apple.com) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/colorsync-users/rene%40damkot.com
This email sent to rene@damkot.com
Jim: For my workflow, the X-Rite software just does not work! ! ! I must use the Adobe DNG profile editor. The sales pitch by the woman at x-rite (she is no longer at x-rite) was/is the software was the one to use. My question to her, never really answered, why is there a difference in results? Will, I think the reply was something like Adobe DNG profile editor is just different. Me. I was expecting the same results. Nope! So, Adobe DNG profile editor is what a must use. David B Miller, Pharm. D. member Millers' Photography L.L.C. dba Spinnaker Photo Imaging Center Bellingham, WA www.spinnakerphotoimagingcenter.com 360 714 1345 On May 8, 2013, at 10:57 AM, Tim Vitale <tjvitale@ix.netcom.com> wrote:
Is there someone on the list that knows the goal of the DNG profile?
Responding to: Re: Generating X-Rite DNG camera profiles for LR. Did I miss something?
I agree with what has been said about the results of using the DNG profile by others.
Remember that the DNG profile was made with 'free software' after the purchase of a $100 target system. Yet another CC-24, added to a collection of CC-24s.
The problem described with color saturation seems to be a deal breaker for me. The first of two.
I had not noticed that the gamma was 'intentionally uncorrected,' only that it was NOT corrected.
Coming from someone who always self-neutralizes the grays on the CC-24...
THE GRAYS WERE NOT NEUTRALIZED! The second deal breaker.
Doing the grays myself takes time. I was hoping for a little new technology to help me out, but no such luck.
Is there some use of the DNG profile that I'm unaware? It was said before: "did I miss something."
Tim Vitale Conservator Oakland, CA 510-594-8277
_______________________________________________ Do not post admin requests to the list. They will be ignored. Colorsync-users mailing list (Colorsync-users@lists.apple.com) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/colorsync-users/spinnakerphotoimagin...
This email sent to spinnakerphotoimagingcenter@dnmillerphoto.com
On May 8, 2013, at 11:57 AM, Tim Vitale <tjvitale@ix.netcom.com> wrote:
Is there someone on the list that knows the goal of the DNG profile?
Adobe raw processors use a DNG profile as part of the image processing chain. You can use the one's supplied for your camera type by Adobe, but your camera may not behave just like the sample used by Adobe to build the DNG profile. So create your own. I make one for Daylight + Tungsten then just for really odd illuminates (CFL, metal halide). Plus you may wish to edit the profile using the free DNG profile editor.
Remember that the DNG profile was made with 'free software' after the purchase of a $100 target system. Yet another CC-24, added to a collection of CC-24s.
DNG profiles are created with a Macbeth 24 patch target. Anyone with such a target can build a profile for their cameras using two free software options (Adobe or X-rite). The 24 patch Macbeth use in Adobe raw processors dates back many years, prior to DNG profiles and the work of the late great Bruce Fraser. He figured out a way to 'tweak' the calibration settings in early versions of ACR so that the ProPhoto processing space used in ACR would match what ProPhoto RGB of each 24 patch value should be out the back end. Then we got all kinds of scripts and finally Adobe enhanced the Calibration area with the ability to use custom DNG profiles.
The problem described with color saturation seems to be a deal breaker for me.
DNG profiles shouldn't affect saturation this way per se. Images still need a decent ACR or LR default setting. Zero settings probably isn't it, certainly not with PV2010 or PV2003. If one see's an image with too much saturation, there are no less than three controls in the ACR engine to affect this. IF you see a saturation issue on many images, you simply alter, then update the ACR defaults.
I had not noticed that the gamma was 'intentionally uncorrected,' only that it was NOT corrected.
Coming from someone who always self-neutralizes the grays on the CC-24...
THE GRAYS WERE NOT NEUTRALIZED! The second deal breaker.
Again, the idea of not White Balancing images by application of a DNG profile is faulty, the Adobe raw processing engine doesn't work that way. Many images will need a custom WB no matter the DNG profile, the two are separate. IOW, it isn't the role of a DNG profile to 'correct' WB. That's part of rendering the image. In Adobe's way of processing raw, the idea is we are working with a negative, much as some of us did in the old analog dark room with color negs. The filter pack played a huge role in the color appearance. Nearly every neg needed a different filter pack recipe. The correct color is the color the printer desired, not necessarily the right color based on some measurement. I've built a number of DNG Profiles for my cameras, I've yet to see a case where such a custom profile didn't produce a better appearing, more acceptable starting point for rendering then the supplied DNG profiles. Andrew Rodney http://www.digitaldog.net/
On May 8, 2013, at 11:43 AM, Andrew Rodney <andrew@digitaldog.net> wrote:
In Adobe's way of processing raw, the idea is we are working with a negative, much as some of us did in the old analog dark room with color negs. The filter pack played a huge role in the color appearance. Nearly every neg needed a different filter pack recipe. The correct color is the color the printer desired, not necessarily the right color based on some measurement.
I've built a number of DNG Profiles for my cameras, I've yet to see a case where such a custom profile didn't produce a better appearing, more acceptable starting point for rendering then the supplied DNG profiles.
This is an excellent summary of my experience with both the intent and practice of Adobe's raw processing engine. I would only add to it that modern DSLRs are capable of more more precise, much more linear response than any type of film ever. As a result...well, if you're looking to do the same sorts of things with modern cameras as has always been done with film (and there are a great many reasons why that's what most people are most interested in doing), then Adobe's approach is probably best. But if you're looking to do that which modern cameras can that film couldn't, then Adobe's approach ties you down with all the hoary old colorimetric problems of film and prevents you from taking advantage of the new capabilities of modern cameras. Cheers, b&
On May 8, 2013, at 12:53 PM, Ben Goren <ben@trumpetpower.com> wrote:
I would only add to it that modern DSLRs are capable of more more precise, much more linear response than any type of film ever. As a result...well, if you're looking to do the same sorts of things with modern cameras as has always been done with film (and there are a great many reasons why that's what most people are most interested in doing), then Adobe's approach is probably best.
There's film and there's film. A color neg and a color transparency are vastly different in terms of the photographers control over the rendering (an integral part of photography). They are not necessarily exposed the same way, processed the same way nor printed the same way. A JPEG from a camera is a better cousin to a transparency while a raw file is a closer cousin to the color neg. Adobe raw processors and a product appropriately called Lightroom treat raws as a photorapher trained to handle color negs would treat that kind of film. Anyone on this list ever tried building an ICC profile for a scanner that has to scan a color neg? Not the same as profiling a scanner using a good transparency. Andrew Rodney http://www.digitaldog.net/
I have no idea, I only have on Canon 5D and one Canon 5DMII and both greatly benefit from a custom DNG profile. YMMV. Since I've owned Macbeth's for a good 20+ years, nothing to lose, everything to gain. Andrew Rodney http://www.digitaldog.net/ On May 8, 2013, at 1:00 PM, Iliah Borg <iliah.i.borg@gmail.com> wrote:
On May 8, 2013, at 2:43 PM, Andrew Rodney wrote:
your camera may not behave just like the sample used by Adobe to build the DNG profile
How much the sensor response should differ to make sample variation visible?
-- Best regards, Iliah Borg
On May 8, 2013, at 12:00 PM, Iliah Borg <iliah.i.borg@gmail.com> wrote:
On May 8, 2013, at 2:43 PM, Andrew Rodney wrote:
your camera may not behave just like the sample used by Adobe to build the DNG profile
How much the sensor response should differ to make sample variation visible?
An interesting project to gauge that would be for somebody to collect raw files of images of ColorCheckers shot outdoors at noon on a sunny day. Or, for one of the rental companies to make a series of such images with their inventory. I'd be happy to assist with the analysis of such a project, should anybody else be interested in soliciting / collecting / hosting the files. Personally, I suspect negligible variation with bodies and much more variation (but still relatively small) with lenses. But there's no way to confirm or refute those suspicions without an empirical analysis.... Cheers, b&
On May 8, 2013, at 3:38 PM, Ben Goren wrote:
On May 8, 2013, at 12:00 PM, Iliah Borg <iliah.i.borg@gmail.com> wrote:
On May 8, 2013, at 2:43 PM, Andrew Rodney wrote:
your camera may not behave just like the sample used by Adobe to build the DNG profile
How much the sensor response should differ to make sample variation visible?
An interesting project to gauge that would be for somebody to collect raw files of images of ColorCheckers shot outdoors at noon on a sunny day.
Lenses and light can make a great difference. Stray light in the camera, too. I usually profile from several cameras of the same make and model, using a monochromator and no lens. For older CCD cameras sample variation is indeed a factor. For Canon, Nikon, Olympus and Sony cameras made during last 4 years the sample variation is small and hardly can make a huge difference in profiles. I would be glad to provide hosting and systemize ColorChecker / ColorChecker SG / other targets raw shots. -- Best regards, Iliah Borg
On May 8, 2013, at 9:59 AM, Paul Schilliger <pschilliger@smile.ch> wrote:
Here are my first impressions, tell me please if you think that I must have missed some important step on the way, or if that's simply the way it is.
You nailed it. You can, if you like, use Adobe's DNG Profile Editor to create another DNG Profile that uses the color tables of your new profile but with a hand-drawn tone curve that re-linearizes the curve. Your best bet is to use Apple's Digital Color Meter to check known samples, and use the fewest control points you can...if I remember right, three is about the right number for best results. Start with midtones and then refine highlights and shadows. Too many points and you'll introduce nonlinear discontinuities; much better to live with a less-than-perfect fit. You've also got to set a linear ``curve'' in ACR, and there might be some other things you need to do (apply ACR adjustments in DNG Profile Editor?). As for white balance...crop the image so you just see the entire ColorChecker, including the ``creative'' half. Crank the saturation as high as it'll go and fiddle with the white balance knobs until it looks as close to correct as you can get it, and then return saturation to the neutral position. The click method will get you in the ballpark, but no amount of clicking will get it perfect. This won't, either, but it's as good as you're going to get with Adobe tools. Then adjust exposure until the N5 patch has R/G/B values as close to 118/118/188 as you can get...and then re-adjust white balance and keep looping back and forth between the two until you reach the point of diminishing returns. Obviously, this all needs to be done before you build your DNG profile.... Cheers, b& P.S. Still haven't had a chance to work on my workflow document...but it's looking good for me to get to it either tomorrow or Friday, and hopefully finish up this weekend. Maybe.... b& P.P.S. A similar workflow as what I described above is also your best bet (that I've found) for other non-ICC-aware raw development engines, such as Canon's Digital Photo Processor / Picture Style Editor. b&
I consider DNG a file format. Is this correct? Not an .icc profile. Please advise. Next. As many of you know, I had corrupted files. I won't go into that anymore. However, which is more stable, PSD, or TIFF? Thanks to all David B Miller, Pharm. D. member Millers' Photography L.L.C. dba Spinnaker Photo Imaging Center Bellingham, WA www.spinnakerphotoimagingcenter.com 360 714 1345 On May 8, 2013, at 10:13 AM, Ben Goren <ben@trumpetpower.com> wrote:
On May 8, 2013, at 9:59 AM, Paul Schilliger <pschilliger@smile.ch> wrote:
Here are my first impressions, tell me please if you think that I must have missed some important step on the way, or if that's simply the way it is.
You nailed it.
You can, if you like, use Adobe's DNG Profile Editor to create another DNG Profile that uses the color tables of your new profile but with a hand-drawn tone curve that re-linearizes the curve. Your best bet is to use Apple's Digital Color Meter to check known samples, and use the fewest control points you can...if I remember right, three is about the right number for best results. Start with midtones and then refine highlights and shadows. Too many points and you'll introduce nonlinear discontinuities; much better to live with a less-than-perfect fit.
You've also got to set a linear ``curve'' in ACR, and there might be some other things you need to do (apply ACR adjustments in DNG Profile Editor?).
As for white balance...crop the image so you just see the entire ColorChecker, including the ``creative'' half. Crank the saturation as high as it'll go and fiddle with the white balance knobs until it looks as close to correct as you can get it, and then return saturation to the neutral position. The click method will get you in the ballpark, but no amount of clicking will get it perfect. This won't, either, but it's as good as you're going to get with Adobe tools.
Then adjust exposure until the N5 patch has R/G/B values as close to 118/118/188 as you can get...and then re-adjust white balance and keep looping back and forth between the two until you reach the point of diminishing returns.
Obviously, this all needs to be done before you build your DNG profile....
Cheers,
b&
P.S. Still haven't had a chance to work on my workflow document...but it's looking good for me to get to it either tomorrow or Friday, and hopefully finish up this weekend. Maybe.... b&
P.P.S. A similar workflow as what I described above is also your best bet (that I've found) for other non-ICC-aware raw development engines, such as Canon's Digital Photo Processor / Picture Style Editor. b& _______________________________________________ Do not post admin requests to the list. They will be ignored. Colorsync-users mailing list (Colorsync-users@lists.apple.com) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/colorsync-users/spinnakerphotoimagin...
This email sent to spinnakerphotoimagingcenter@dnmillerphoto.com
On May 8, 2013, at 10:19 AM, Spinnaker Photo Imaging Center <spinnakerphotoimagingcenter@dnmillerphoto.com> wrote:
I consider DNG a file format. Is this correct? Not an .icc profile. Please advise.
DNG is a file format, and part of the format specification is for a completely-unrelated-to-ICC color profile called ``DNG color profile'' (or something like that). DNG profiles and ICC profiles vaguely do the same sort of thing, but DNG profiles are intentionally designed to be loose-fitting general-purpose tools for creating ``pleasing'' color in photographs, and ICC profiles are designed for precision. Cheers, b&
However, which is more stable, PSD, or TIFF?
To me, TIFF ... because: * it's more likely to be supported for the duration of your grand-children's children's lifetime. * if you ever get into the position of changing pp tools away from Adobe, TIFF's are much more likely to be supported. * If Adobe ceases to exist, or decides to abandon support of PSD's, a TIFF is still likely to be supported. * My understanding is that PSD's are more complicated than TIFF's, in that a PSD is something of a "wrapper" of a tiff (as are many if not most RAW files). The PSD may be smaller than a TIFF, but my speculation is that would be because it uses a different compression algorithm. Typically, complexity == less stable. Regarding vulnerability to file corruption ... that's typically proportionally related to file size. So I suppose a smaller PSD would have a smaller risk of file corruption. Or not? Also, "consider the source" ... others on this list may inform both of us about "the error of my ways".
participants (8)
-
Andrew Rodney
-
Ben Goren
-
Iliah Borg
-
LdaSignup
-
Paul Schilliger
-
René Damkot
-
Spinnaker Photo Imaging Center
-
Tim Vitale