Colorchecker calibration, Gamma and Other correction techniques

Sdiver2489 said:

The only thing that bothers me is I don't quite understand how the scaling of the image works in Lumariver. If I am interpreting the manual correctly, it scaled the image so the whitest page on the target doesn't clip. This is definitely a nice feature but I would rather have it scale based on the 18% grey chip so the overall exposure was standardized rather than the absolute highlight. From what I see, after applying the Lumariver profile, 18% grey is usually sitting around 48% Luminance...close but just a little dark. Again, I can apply the tone curve to fix this.

Click to expand...

If you are going for a Reproduction workflow don't apply tone curves to correct exposure, just expose properly. If Exposure is too low, increase it, using LPD Headroom just for fine tuning if you wish. The Baseline Exposure tag in dcp profiles ('Headroom in LPD) is used to compensate for differences in exposure treatment by different camera manufacturers. It works linearly on raw data so it does not change chromaticities. Not so curves. Also not so LR.

If you expect to shoot a color target and all images that will use the resulting profile in the exact same conditions forever, get the lighting and exposure where you want it and make your profile. Then open RT, load your raw file, in Processing Profiles top right select the Neutral entry, head over to the Color tab, pick the White Balance eyedropper and click on CC24 patch D4, then go down to the Color Management section, select the Custom profile you just made and make sure all checkboxes that are tickable below it are ticked. Looks flat, right?

Now head back to the Exposure tab and click on the Auto Matched Tone Curve button, et voilà, an approximation of the Standard Adobe Profile tone curve gleaned from the jpeg embedded in the Raw file. If all your images are always going to be seen on a calibrated monitor and your monitor is calibrated, this last curve I would want to incorporate in the LPD profile, because curves introduce chromaticity shifts while LPD is great at minimizing those by using a TMO in its profiles, albeit only in General Purpose 2.5D mode. So the next time that you go through this RT workflow the image will no longer look flat at the end of the paragraph above, nor will you need the Auto Matched tone curve - and your colors should be a little better. Now save a TIFF or print it to compare. How are the oranges?

Sdiver2489 said:

I can absolutely provide you a RAW file I am working with if you like.

Click to expand...

Great. Ideally you would put the two raw files shot in the same conditions that you consider best, one with the CC24 and one with the subject, plus your favorite dcp profiles in a drive for download.

Jack

Jack Hogan said:

Sdiver2489 said:

The only thing that bothers me is I don't quite understand how the scaling of the image works in Lumariver. If I am interpreting the manual correctly, it scaled the image so the whitest page on the target doesn't clip. This is definitely a nice feature but I would rather have it scale based on the 18% grey chip so the overall exposure was standardized rather than the absolute highlight. From what I see, after applying the Lumariver profile, 18% grey is usually sitting around 48% Luminance...close but just a little dark. Again, I can apply the tone curve to fix this.

Click to expand...

If you are going for a Reproduction workflow don't apply tone curves to correct exposure, just expose properly. If Exposure is too low, increase it, using LPD Headroom just for fine tuning if you wish. The Baseline Exposure tag in dcp profiles ('Headroom in LPD) is used to compensate for differences in exposure treatment by different camera manufacturers. It works linearly on raw data so it does not change chromaticities. Not so curves. Also not so LR.

Click to expand...

This is an interesting point. So my goal has always been to get exposure absolutely correct. The question has always been how to judge that. The A7riv has a notoriously unreliable LCD for judging exposure on the camera itself. So I, of course, use the histogram. I'll post the histogram for one shot below here but it looked fairly accurate. Fairly close to the right without clipping any whites. If I want to get it PERFECT however, the only way I can think to do that with flash is to adjust the power of the flash taking a picture of solely a 18% gray card and make sure its perfectly centered in the histogram. Even then, it'll be my best judgement of what is exactly the center. If you have any other ideas of how to make sure each exposure is absolutely perfect out of camera I'm open to all suggestions as this is a topic I've struggled with finding a technical solution to.

Jack Hogan said:

If you expect to shoot a color target and all images that will use the resulting profile in the exact same conditions forever, get the lighting and exposure where you want it and make your profile. Then open RT, load your raw file, in Processing Profiles top right select the Neutral entry, head over to the Color tab, pick the White Balance eyedropper and click on CC24 patch D4, then go down to the Color Management section, select the Custom profile you just made and make sure all checkboxes that are tickable below it are ticked. Looks flat, right?

Click to expand...

Yup this is exactly what I did in RAWTherapee. Glad I got it more or less right...little bit of a learning curve to that program when you pick it up.

Jack Hogan said:

Now head back to the Exposure tab and click on the Auto Matched Tone Curve button, et voilà, an approximation of the Standard Adobe Profile tone curve gleaned from the jpeg embedded in the Raw file. If all your images are always going to be seen on a calibrated monitor and your monitor is calibrated, this last curve I would want to incorporate in the LPD profile, because curves introduce chromaticity shifts while LPD is great at minimizing those by using a TMO in its profiles, albeit only in General Purpose 2.5D mode. So the next time that you go through this RT workflow the image will no longer look flat at the end of the paragraph above, nor will you need the Auto Matched tone curve - and your colors should be a little better. Now save a TIFF or print it to compare. How are the oranges?

Click to expand...

Generally speaking I'm not concerned with the image looking flat. The purpose of these files is pretty much solely to print. There is a second purpose to act as a digital backup but they would again be printed in the case of damage to the original.

Jack Hogan said:

Jack Hogan said:

I can absolutely provide you a RAW file I am working with if you like.

Click to expand...

Great. Ideally you would put the two raw files shot in the same conditions that you consider best, one with the CC24 and one with the subject, plus your favorite dcp profiles in a drive for download.

Jack

Click to expand...

Sent via PM Thanks for your willingness to help and your patience with me as I do my best to improve my technique.

Sdiver2489 said:

Jack Hogan said:

Sdiver2489 said:

The only thing that bothers me is I don't quite understand how the scaling of the image works in Lumariver. If I am interpreting the manual correctly, it scaled the image so the whitest page on the target doesn't clip. This is definitely a nice feature but I would rather have it scale based on the 18% grey chip so the overall exposure was standardized rather than the absolute highlight. From what I see, after applying the Lumariver profile, 18% grey is usually sitting around 48% Luminance...close but just a little dark. Again, I can apply the tone curve to fix this.

Click to expand...

If you are going for a Reproduction workflow don't apply tone curves to correct exposure, just expose properly. If Exposure is too low, increase it, using LPD Headroom just for fine tuning if you wish. The Baseline Exposure tag in dcp profiles ('Headroom in LPD) is used to compensate for differences in exposure treatment by different camera manufacturers. It works linearly on raw data so it does not change chromaticities. Not so curves. Also not so LR.

Click to expand...

This is an interesting point. So my goal has always been to get exposure absolutely correct. The question has always been how to judge that. The A7riv has a notoriously unreliable LCD for judging exposure on the camera itself. So I, of course, use the histogram. I'll post the histogram for one shot below here but it looked fairly accurate. Fairly close to the right without clipping any whites. If I want to get it PERFECT however, the only way I can think to do that with flash is to adjust the power of the flash taking a picture of solely a 18% gray card and make sure its perfectly centered in the histogram. Even then, it'll be my best judgement of what is exactly the center. If you have any other ideas of how to make sure each exposure is absolutely perfect out of camera I'm open to all suggestions as this is a topic I've struggled with finding a technical solution to.

Click to expand...

With raw files, there is no correct exposure.

It could be argued, that correct exposure is the one that one reproduces the grey patches optimally without exposure correction.

I would do this:

Take a test image.

• Open in LR, Process Version v, with no curve.
• Check Lab on fourth neutral patch
• Adjust exposure until L in Lab value is 55.
• Apply that exposure correction to the exposure used for the test image

Best regards

Erik

--
Erik Kaffehr
Website: http://echophoto.dnsalias.net
Magic uses to disappear in controlled experiments…
Gallery: http://echophoto.smugmug.com
Articles: http://echophoto.dnsalias.net/ekr/index.php/photoarticles

Sdiver2489 said:

Sdiver2489 said:

Jack Hogan wrote: [edited]

If you are going for a Reproduction workflow don't apply tone curves to correct exposure, just expose properly. If Exposure is too low, increase it, using LPD Headroom just for fine tuning if you [wish must]. The Baseline Exposure tag in dcp profiles ('Headroom in LPD) is used [in the DNG spec] to compensate for differences in exposure treatment by different camera manufacturers. It works linearly on raw data so it does not change chromaticities. Not so curves. Also not so LR.

Click to expand...

This is an interesting point. So my goal has always been to get exposure absolutely correct. The question has always been how to judge that. The A7riv has a notoriously unreliable LCD for judging exposure on the camera itself. So I, of course, use the histogram. I'll post the histogram for one shot below here but it looked fairly accurate. Fairly close to the right without clipping any whites.

Click to expand...

Yes, in this very controlled studio context with no mixed or specular highlights, Exposing To The Right is a good strategy. For best IQ the raw histogram should be used as a guide, RT's raw histogram is not bad to work off of. Looking at the jpeg histogram should work as long as a bit of headroom is left to the right and the camera is set up to use a color space larger than the intended output device's. Most cameras only allow sRGB and Adobe RGB, I would go for the latter if you are printing.

Sdiver2489 said:

If I want to get it PERFECT however, the only way I can think to do that with flash is to adjust the power of the flash taking a picture of solely a 18% gray card and make sure its perfectly centered in the histogram. Even then, it'll be my best judgement of what is exactly the center. If you have any other ideas of how to make sure each exposure is absolutely perfect out of camera I'm open to all suggestions as this is a topic I've struggled with finding a technical solution to.

Click to expand...

As Erik says, it's all relative, meaning that it often is not possible to match the absolute Exposure at the scene: think for instance of a mountain landscape on a sunny day: a monitor would never be able to reproduce the Luminance at the scene. Fortunately the Human Visual System adapts approximately linearly. So typically we ETTR, which produces the best possible SNR without clipping tones reflected diffusely. Then adjust brightness linearly during raw conversion. If everything is really fixed, you can use the Exposure Compensation slider in RT (not necessarily in LR) to match your reference*.

Save the EC as a preset in RT with your profile and white balance settings*, therefore reducing the workflow below to a single click. Or you can use the Headroom feature of LPD. Careful with EC if using a 3D profile though.

Sdiver2489 said:

Sdiver2489 said:

If you expect to shoot a color target and all images that will use the resulting profile in the exact same conditions forever, get the lighting and exposure where you want it and make your profile. Then open RT, load your raw file, in Processing Profiles top right select the Neutral entry, head over to the Color tab, pick the White Balance eyedropper and click on CC24 patch D4, then go down to the Color Management section, select the Custom profile you just made and make sure all checkboxes that are tickable below it are ticked. Looks flat, right?

Click to expand...

Yup this is exactly what I did in RAWTherapee. Glad I got it more or less right...little bit of a learning curve to that program when you pick it up.

Sdiver2489 said:

Now head back to the Exposure tab and click on the Auto Matched Tone Curve button, et voilà, an approximation of the [Standard Adobe Profile] tone curve [used in camera] gleaned from the jpeg embedded in the Raw file. If all your images are always going to be seen on a calibrated monitor and your monitor is calibrated, this last curve I would want to incorporate in the LPD profile, because curves introduce chromaticity shifts while LPD is great at minimizing those by using a TMO in its profiles, albeit only in General Purpose 2.5D mode. So the next time that you go through this RT workflow the image will no longer look flat at the end of the paragraph above, nor will you need the Auto Matched tone curve - and your colors should be a little better. Now save a TIFF or print it to compare. How are the oranges?

Click to expand...

Generally speaking I'm not concerned with the image looking flat. The purpose of these files is pretty much solely to print.

Click to expand...

It seems that you are using your own printer. What is it? Is it properly profiled? If so you may consider the suggestion above, building a TMO that works well for the monitor and one that works well for the printer into two additional GP profiles: one for the printer and one for the monitor. The two should match well visually side by side. Now you would have three presets in RT: 1 for viewing, 1 for printing and the original flat one as a base for archiving.

Sdiver2489 said:

There is a second purpose to act as a digital backup but they would again be printed in the case of damage to the original.

Sdiver2489 said:

Sdiver2489 said:

I can absolutely provide you a RAW file I am working with if you like.

Click to expand...

Click to expand...

Sent via PM Thanks for your willingness to help and your patience with me as I do my best to improve my technique.

Click to expand...

My pleasure, I don't have time right now but I will take a look as soon as I can.

Jack

* About white balance: in RT, it would be best to have a truly neutral target to white balance off of, because none of the CC24 patches are really 100% neutral. The dedicated page in the Passport is pretty good. I also like the WhiBal card, which can sit right next to the CC24 when taking the profiling shot. You can use that target to figure out EC/Headroom: Whibal is around 50% reflectance (L*75.8), I can't remember the Passport's.

I tried my suggested workflow and I get bad results, whatever I do in Lightroom.

Sorry for that.

Best regards

Erik

Erik Kaffehr said:

Sdiver2489 said:

Jack Hogan said:

Sdiver2489 said:

The only thing that bothers me is I don't quite understand how the scaling of the image works in Lumariver. If I am interpreting the manual correctly, it scaled the image so the whitest page on the target doesn't clip. This is definitely a nice feature but I would rather have it scale based on the 18% grey chip so the overall exposure was standardized rather than the absolute highlight. From what I see, after applying the Lumariver profile, 18% grey is usually sitting around 48% Luminance...close but just a little dark. Again, I can apply the tone curve to fix this.

Click to expand...

If you are going for a Reproduction workflow don't apply tone curves to correct exposure, just expose properly. If Exposure is too low, increase it, using LPD Headroom just for fine tuning if you wish. The Baseline Exposure tag in dcp profiles ('Headroom in LPD) is used to compensate for differences in exposure treatment by different camera manufacturers. It works linearly on raw data so it does not change chromaticities. Not so curves. Also not so LR.

Click to expand...

This is an interesting point. So my goal has always been to get exposure absolutely correct. The question has always been how to judge that. The A7riv has a notoriously unreliable LCD for judging exposure on the camera itself. So I, of course, use the histogram. I'll post the histogram for one shot below here but it looked fairly accurate. Fairly close to the right without clipping any whites. If I want to get it PERFECT however, the only way I can think to do that with flash is to adjust the power of the flash taking a picture of solely a 18% gray card and make sure its perfectly centered in the histogram. Even then, it'll be my best judgement of what is exactly the center. If you have any other ideas of how to make sure each exposure is absolutely perfect out of camera I'm open to all suggestions as this is a topic I've struggled with finding a technical solution to.

Click to expand...

With raw files, there is no correct exposure.

It could be argued, that correct exposure is the one that one reproduces the grey patches optimally without exposure correction.

I would do this:

Take a test image.

• Open in LR, Process Version v, with no curve.
• Check Lab on fourth neutral patch
• Adjust exposure until L in Lab value is 55.
• Apply that exposure correction to the exposure used for the test image

Best regards

Erik

Click to expand...

This is interesting. This really comes back to my previous methodology I used pre-lumariver where I took my shots, made a profile via X-rite. Applied the profile and white balanced. Then I scaled the image to match the correct Lab value.



I just tried it with profile 5 and the results look fairly promising. A bit brighter than RT doing the same thing. RT I cannot get the swatch up past 50% without the highlights clipping. See below.















So at least Lightroom works. It just bothers me from a "science" point of view of why, if RT is truly the most true to its purpose, that it clips with a 2.5D profile when the captured image is low in dynamic range and should be linear as we have acknowledged. I'm happy LR doesn't do this (on version 5...on version 2 it behaves more like RT) but I'm also wanting to figure out what's going on with all the differences in processing.

Jack Hogan said:

Sdiver2489 said:

Jack Hogan said:

Jack Hogan wrote: [edited]

If you are going for a Reproduction workflow don't apply tone curves to correct exposure, just expose properly. If Exposure is too low, increase it, using LPD Headroom just for fine tuning if you [wish must]. The Baseline Exposure tag in dcp profiles ('Headroom in LPD) is used [in the DNG spec] to compensate for differences in exposure treatment by different camera manufacturers. It works linearly on raw data so it does not change chromaticities. Not so curves. Also not so LR.

Click to expand...

This is an interesting point. So my goal has always been to get exposure absolutely correct. The question has always been how to judge that. The A7riv has a notoriously unreliable LCD for judging exposure on the camera itself. So I, of course, use the histogram. I'll post the histogram for one shot below here but it looked fairly accurate. Fairly close to the right without clipping any whites.

Click to expand...

Yes, in this very controlled studio context with no mixed or specular highlights, Exposing To The Right is a good strategy. For best IQ the raw histogram should be used as a guide, RT's raw histogram is not bad to work off of. Looking at the jpeg histogram should work as long as a bit of headroom is left to the right and the camera is set up to use a color space larger than the intended output device's. Most cameras only allow sRGB and Adobe RGB, I would go for the latter if you are printing.

Click to expand...

Yup this is exactly what I do (and did)

Jack Hogan said:

Jack Hogan said:

If I want to get it PERFECT however, the only way I can think to do that with flash is to adjust the power of the flash taking a picture of solely a 18% gray card and make sure its perfectly centered in the histogram. Even then, it'll be my best judgement of what is exactly the center. If you have any other ideas of how to make sure each exposure is absolutely perfect out of camera I'm open to all suggestions as this is a topic I've struggled with finding a technical solution to.

Click to expand...

As Erik says, it's all relative, meaning that it often is not possible to match the absolute Exposure at the scene: think for instance of a mountain landscape on a sunny day: a monitor would never be able to reproduce the Luminance at the scene. Fortunately the Human Visual System adapts approximately linearly. So typically we ETTR, which produces the best possible SNR without clipping tones reflected diffusely. Then adjust brightness linearly during raw conversion. If everything is really fixed, you can use the Exposure Compensation slider in RT (not necessarily in LR) to match your reference*.

Save the EC as a preset in RT with your profile and white balance settings*, therefore reducing the workflow below to a single click. Or you can use the Headroom feature of LPD. Careful with EC if using a 3D profile though.

Click to expand...

You can see my comments to Erik. My success with 2.5D profiles seems to so far limited to LR version 5. Haven't tried to print it yet but it looks like it should print well comparatively to other edits I have done. The problem I've been having time and time again with certain combinations of profiles and processes/software is clipping highlights before I'm able to average the exposure to 18% grey.

That's why I liked the headroom feature of Lumariver. It seemed like it would standardize my exposures (I assume it uses the brightest white patch at 96% Lab to pick the gain needed for the image. However, as you have pointed out this doesn't appear to be entirely reliable from image to image.

Jack Hogan said:

Jack Hogan said:

Jack Hogan said:

If you expect to shoot a color target and all images that will use the resulting profile in the exact same conditions forever, get the lighting and exposure where you want it and make your profile. Then open RT, load your raw file, in Processing Profiles top right select the Neutral entry, head over to the Color tab, pick the White Balance eyedropper and click on CC24 patch D4, then go down to the Color Management section, select the Custom profile you just made and make sure all checkboxes that are tickable below it are ticked. Looks flat, right?

Click to expand...

Yup this is exactly what I did in RAWTherapee. Glad I got it more or less right...little bit of a learning curve to that program when you pick it up.

Jack Hogan said:

Now head back to the Exposure tab and click on the Auto Matched Tone Curve button, et voilà, an approximation of the [Standard Adobe Profile] tone curve [used in camera] gleaned from the jpeg embedded in the Raw file. If all your images are always going to be seen on a calibrated monitor and your monitor is calibrated, this last curve I would want to incorporate in the LPD profile, because curves introduce chromaticity shifts while LPD is great at minimizing those by using a TMO in its profiles, albeit only in General Purpose 2.5D mode. So the next time that you go through this RT workflow the image will no longer look flat at the end of the paragraph above, nor will you need the Auto Matched tone curve - and your colors should be a little better. Now save a TIFF or print it to compare. How are the oranges?

Click to expand...

Generally speaking I'm not concerned with the image looking flat. The purpose of these files is pretty much solely to print.

Click to expand...

It seems that you are using your own printer. What is it? Is it properly profiled? If so you may consider the suggestion above, building a TMO that works well for the monitor and one that works well for the printer into two additional GP profiles: one for the printer and one for the monitor. The two should match well visually side by side. Now you would have three presets in RT: 1 for viewing, 1 for printing and the original flat one as a base for archiving.

Click to expand...

Using my own printer is a sort of "Draft". I use it as a check to make sure my colors/brightness/contrast are "in the ballpark" before shipping it off to a professional print house. The printer I'm using is a Canon Pro-100 with their ink and paper using their profile. I fully realize that profiling the printer would be the next best step if my final output was my printer but it's just a draft copy. Generally speaking I'm able to get very close. But, as you pointed out, I'm looking for a standardized workflow which the science follows and also results in accurate color and brightness reprints of painted artwork. My technique has improved with time and you all have taught me a few things that I'm going to take into account in the next images I shoot.

Jack Hogan said:

Jack Hogan said:

There is a second purpose to act as a digital backup but they would again be printed in the case of damage to the original.

Jack Hogan said:

Jack Hogan said:

I can absolutely provide you a RAW file I am working with if you like.

Click to expand...

Click to expand...

Sent via PM Thanks for your willingness to help and your patience with me as I do my best to improve my technique.

Click to expand...

My pleasure, I don't have time right now but I will take a look as soon as I can.

Jack

* About white balance: in RT, it would be best to have a truly neutral target to white balance off of, because none of the CC24 patches are really 100% neutral. The dedicated page in the Passport is pretty good. I also like the WhiBal card, which can sit right next to the CC24 when taking the profiling shot. You can use that target to figure out EC/Headroom: Whibal is around 50% reflectance (L*75.8), I can't remember the Passport's.

Click to expand...

Not a problem. Take your time or not at all if you don't have the time. I would never ask anything of you all. You have already "donated" a ton of time to my efforts which I appreciate. There isn't much material out there on this stuff. Most of it is far too "amateur" for me where there is no effort to calibrate, standardize or use science to help out. I've done that method too...it works but it's a pain.

I've heard that about the white balance targets. I honestly haven't tried the white balance card because it seemed unnecessary unless I was white balancing in camera...but I might give it a go and see if it results in any substantial differences. Now that I have Lumariver, I can also play around with different color targets to see which works best.

The bottom image is brighter overall and looks clipped, which means that after Headroom adjustments there are values above L100, which should not happen with a wholly diffuse image, especially one like yours. What did the histogram look like? I suggest ETTR wrt the raw data, and apply one of your 'flat' profiles without Headroom. Then adjust Exposure Compensation in RT to fine tune brightness, making sure rendered highlights are not clipped. Reshoot with such EC and apply the GP profile.

You may need to do this because even though the ETTR raw data may not be clipped - usually meaning that the green color raw channel is just about to be clipped while Red and Blue are lower - after the white balance multipliers are applied it may end up that G is still around 100% but R and B are higher than that (see figure 8 here for instance). Maximum diffuse white cannot be higher than L100 by definition, so if you want to stay true to the input you need to lower exposure to bring R and B back under 100.

LR may look ok because it introduces highlight recovery and a curve with non linear responses (and related chromaticity shifts). It is a useful fallback - but why fallback if you don't need to? Get compensated Exposure right and you have the best of both worlds.

Jack

Sdiver2489 said:

Sdiver2489 wrote: I just tried it with profile 5 and the results look fairly promising. A bit brighter than RT doing the same thing. RT I cannot get the swatch up past 50% without the highlights clipping. See below.

Click to expand...

Do you mean patch D4 in the Passport? Depending on your unit, that's around 18-19% reflectance, which is about L*49.5-50.7. If you push it higher than that maximum diffuse white will clip by definition. So you shouldn't. To be safe, keep it below L*50, or around 18% reflectance***.

Jack

*** 116 * [0.18^(1/3)] - 16 = 49.5, so L*49.5. If you don't know the value for your copy of the Passport stay below that. % diffuse reflectance is linear and Y/Yn in the link. It is different than L*, which is non-linear. Though it is true that 100% diffuse reflectance is L*100.

Sdiver2489 said:

I've heard that about the white balance targets. I honestly haven't tried the white balance card because it seemed unnecessary unless I was white balancing in camera...but I might give it a go and see if it results in any substantial differences. Now that I have Lumariver, I can also play around with different color targets to see which works best.

Click to expand...

It's always good to have at least one image containing the white balance target for the lighting of the scene. Now, if you're using your own lighting all the time, you should only need to do that once, but even then if there's the chance of stray 'other' lighting it's good thing to have, OOC JPEG or raw...

Once you've left the scene behind, white balance compensation is all just guessing...

ggbutcher said:

Sdiver2489 said:

I've heard that about the white balance targets. I honestly haven't tried the white balance card because it seemed unnecessary unless I was white balancing in camera...but I might give it a go and see if it results in any substantial differences. Now that I have Lumariver, I can also play around with different color targets to see which works best.

Click to expand...

It's always good to have at least one image containing the white balance target for the lighting of the scene. Now, if you're using your own lighting all the time, you should only need to do that once, but even then if there's the chance of stray 'other' lighting it's good thing to have, OOC JPEG or raw...

Once you've left the scene behind, white balance compensation is all just guessing...

Click to expand...

Perhaps my terminology wasn't clear. I have a white balance target on the CC24 Passport. I just have not used the larger target which is on a different "page" of the passport device. I always have a white balance target for each "scene" I shoot.

Jack Hogan said:

Jack Hogan said:

Sdiver2489 wrote: I just tried it with profile 5 and the results look fairly promising. A bit brighter than RT doing the same thing. RT I cannot get the swatch up past 50% without the highlights clipping. See below.

Click to expand...

Do you mean patch D4 in the Passport? Depending on your unit, that's around 18-19% reflectance, which is about L*49.5-50.7. If you push it higher than that maximum diffuse white will clip by definition. So you shouldn't. To be safe, keep it below L*50, or around 18% reflectance***.

Jack

*** 116 * [0.18^(1/3)] - 16 = 49.5, so L*49.5. If you don't know the value for your copy of the Passport stay below that. % diffuse reflectance is linear and Y/Yn in the link. It is different than L*, which is non-linear. Though it is true that 100% diffuse reflectance is L*100.

Click to expand...

Yes, and I'm just referencing the data I was sent from X-rite's certificates on the device. I don't have a device to measure swatches to verify the accuracy of the data they provided. I am taking them at their word that this is relatively correct:

Jack Hogan said:

The bottom image is brighter overall and looks clipped, which means that after Headroom adjustments there are values above L100, which should not happen with a wholly diffuse image, especially one like yours. What did the histogram look like? I suggest ETTR wrt the raw data, and apply one of your 'flat' profiles without Headroom. Then adjust Exposure Compensation in RT to fine tune brightness, making sure rendered highlights are not clipped. Reshoot with such EC and apply the GP profile.

You may need to do this because even though the ETTR raw data may not be clipped - usually meaning that the green color raw channel is just about to be clipped while Red and Blue are lower - after the white balance multipliers are applied it may end up that G is still around 100% but R and B are higher than that (see figure 8 here for instance). Maximum diffuse white cannot be higher than L100 by definition, so if you want to stay true to the input you need to lower exposure to bring R and B back under 100.

LR may look ok because it introduces highlight recovery and a curve with non linear responses (and related chromaticity shifts). It is a useful fallback - but why fallback if you don't need to? Get compensated Exposure right and you have the best of both worlds.

Jack

Click to expand...

Depends as always where I am viewing it and with what profile. If we are just talking what the histogram looks like in RT after applying the neutral profile and nothing from Lumariver, I get this:







If I load the same image into LR with its defaults I get this which is what the histogram looks like on the back of the camera:





The odd thing is these trials were done with the 2.5D profile from Lumariver which, by definition, has no headroom setting. However, to get the D4 swatch to 51% Lab (which is what X-rite quoted me in their certification data) I had to boost the exposure by 1.5 EV or so in RT.

Hi,

I made a small experiment, using Lumariver and using a repro profile on a quick and dirty setup.

I created a TIFF in RawTherapy, trying to do a linear conversion and printed that TIFF.

The print looks flat compared to the original.

I also measured a color checker passport included in the image. Here are the Lab values:




So, it seems that we have a good match between the TIFF and the reference values regarding lightness. But highlights in the print are less than optimal. Paper white on the print is 94.5.

So, the issue may mainly be the mapping of grey scale to print and screen.

I can share the DNG and DCP files, would any one be interested.

Best regards

Erik

--
Erik Kaffehr
Website: http://echophoto.dnsalias.net
Magic uses to disappear in controlled experiments…
Gallery: http://echophoto.smugmug.com
Articles: http://echophoto.dnsalias.net/ekr/index.php/photoarticles

Sdiver2489 said:

Jack Hogan said:

The bottom image is brighter overall and looks clipped, which means that after Headroom adjustments there are values above L100, which should not happen with a wholly diffuse image, especially one like yours. What did the histogram look like? I suggest ETTR wrt the raw data, and apply one of your 'flat' profiles without Headroom. Then adjust Exposure Compensation in RT to fine tune brightness, making sure rendered highlights are not clipped. Reshoot with such EC and apply the GP profile.

You may need to do this because even though the ETTR raw data may not be clipped - usually meaning that the green color raw channel is just about to be clipped while Red and Blue are lower - after the white balance multipliers are applied it may end up that G is still around 100% but R and B are higher than that (see figure 8 here for instance). Maximum diffuse white cannot be higher than L100 by definition, so if you want to stay true to the input you need to lower exposure to bring R and B back under 100.

LR may look ok because it introduces highlight recovery and a curve with non linear responses (and related chromaticity shifts). It is a useful fallback - but why fallback if you don't need to? Get compensated Exposure right and you have the best of both worlds.

Jack

Click to expand...

Depends as always where I am viewing it and with what profile. If we are just talking what the histogram looks like in RT after applying the neutral profile and nothing from Lumariver, I get this:

If I load the same image into LR with its defaults I get this which is what the histogram looks like on the back of the camera:

The odd thing is these trials were done with the 2.5D profile from Lumariver which, by definition, has no headroom setting. However, to get the D4 swatch to 51% Lab (which is what X-rite quoted me in their certification data) I had to boost the exposure by 1.5 EV or so in RT.

Click to expand...

I've only been skimming the progress of discourse, so I may have missed it, but is there any accounting for the display profiles used by RT and LR?

Both programs were loaded with my display profile for color management.

Erik Kaffehr said:

Hi,

I made a small experiment, using Lumariver and using a repro profile on a quick and dirty setup.

I created a TIFF in RawTherapy, trying to do a linear conversion and printed that TIFF.

The print looks flat compared to the original.

I also measured a color checker passport included in the image. Here are the Lab values:

So, it seems that we have a good match between the TIFF and the reference values regarding lightness. But highlights in the print are less than optimal. Paper white on the print is 94.5.

So, the issue may mainly be the mapping of grey scale to print and screen.

I can share the DNG and DCP files, would any one be interested.

Best regards

Erik

Click to expand...

Sounds similar to the issues I have in RT, print is a bit dark compared to the original. So far LR still seems to give the best results. Both Process 2 with the headroom from reference and Process 5 2.5D profile manually scaled to match 18% gray. I still need to print this one to see how it compares but both are at least in the ballpark with no clipping.

Sdiver2489 said:

If we are just talking what the histogram looks like in RT after applying the neutral profile and nothing from Lumariver, I get this:

Click to expand...

If that's the raw histogram this image is underexposed by about a stop. This weekend I should have some time to comment on the shared files.

Sdiver2489 said:

Depends as always where I am viewing it and with what profile. If we are just talking what the histogram looks like in RT after applying the neutral profile and nothing from Lumariver, I get this:

Click to expand...

This is dark because the capture is about one stop underexposed, maximum raw value around 7000 DN (it's a weird format I don't know btw). Next your 'Cinderella Front' dcp profile is added, without Baseline Exposure ticked, the image remains one stop underexposed:


Your 3D dcp profile is applied but without ticking Baseline Exposure

So now we tick the Baseline Exposure/Headroom box and the exposure is corrected +1 EV per the instructions in your profile (see below). With lots of provisos, this is equivalent to moving the EC slider up 1. Were you to apply this profile to a properly exposed similar image, it would blow the highlights.

Now relative brightness is correct but the image above looks flattish because your profile's curve is near linear and we are viewing it on a monitor.


Your 3D dcp profile is fully applied

Below instead is the General Purpose profile that I generated via LPD with a TMO based on the classic Adobe curve. After selecting the Neutral Processing Profile in RT, my dcp profile was loaded and the white balance settings copied from the previous one, nothing else.


Loaded image in RT, Neutral Processing Profile, copied WB from above, loaded my GP dcp profile as-is

Now the image finally looks good on my monitor, though I have no idea what it looks like in reality. Some of the orange above and below the yellow reflection of the lantern is out of gamut.

Keep in mind that the profile above is for Adobe RGB but it was then saved as sRGB. The TMO is generic and it and some of the other parameters I chose may not be appropriate for your monitor or printer, so some iteration in LPD is necessary to get it right.

This is less than ideal, it would have been much better to have properly exposed images to work with, but I hope it gives you an idea of how to use Lumariver Profile Designer with RT.

Jack

PS Your Cinderella Front dcp on the left, my GP dcp on the right

I noticed that above I left the 'Auto Black Subtraction' tickbox accidentally ticked in LPD's Tone Curve menu. This is what it looks like with it unticked:


GP dcp profile Same as above but with Auto Black Subtraction unticked

The earlier one is obviously more contrasty, but likely less accurate as you can tell by the increased saturation.