Ad
Ad
Ad
Pages: [1] 2 »   Bottom of Page
Print
Author Topic: Comparing custom DNG camera profiles for ACR/LR.  (Read 9874 times)
Rhossydd
Sr. Member
****
Offline Offline

Posts: 1965


WWW
« on: January 09, 2009, 07:50:06 AM »
ReplyReply

Having been diligently shooting Macbeth color checker charts on many shoots for years now. Iíve built several calibration settings from different shoots with various ACR calibration scripts for Photoshop (Fors etc).
Iíve now moved on to using the DNG profile editor for ACR, which delivers results so much faster.

The visual differences between each of the resultant profiles, or calibration settings, for my camera are generally very subtle, although they are all superior to the default ACR profiles.

One of the things Iím curious about is how different these profiles are from one another. Previously when building calibration sets for ACR via the various Photoshop scripts it was easily possible to see the different numbers from each and see the general areas where changes were needed, but the DNG editor makes seeing these commonalities of settings much less easy to see.

Can anyone suggest a method of comparing the differences between ACR profiles to help understand what corrections are being applied in each profile ?
My aim is to be able to settle on a single 'best average' for my particular camera body, to avoid ending up with very many profiles built for each shoot cluttering up the menu.

Any suggestions or views ?
Logged
madmanchan
Sr. Member
****
Offline Offline

Posts: 2110


« Reply #1 on: January 09, 2009, 08:25:12 AM »
ReplyReply

Certainly. All you need to do is open a single image which has interesting colors (at least, interesting to you -- e.g., things that are representative of the subject matter you often shoot) in Camera Raw or Lightroom and then go to the Camera Calibration tab and flip through the various profiles that you've built. You can then compare how each of the colors in the image change with the profiles.

You can also do the same with an image of the ColorChecker, but of course the ColorChecker is not very representative of your normal subject matter (I hope ...)
Logged

Rhossydd
Sr. Member
****
Offline Offline

Posts: 1965


WWW
« Reply #2 on: January 09, 2009, 09:26:41 AM »
ReplyReply

Thanks Eric.

I've used the ability to see the different effects of the various profiles already.
I'm just particularly interested to see what's going on underneath and numbers can sometimes be easier to understand than the subtleties of small visual changes on the limited range of colours in one image that are also constrained by the limitations of the display system used.
I know the recipes are in XML format (I think) so it ought to be possible to write a utility to help understand what's going on and how multiple recipes compare.
A bit like the compare function in Profile Maker Pro.
Logged
madmanchan
Sr. Member
****
Offline Offline

Posts: 2110


« Reply #3 on: January 09, 2009, 06:31:15 PM »
ReplyReply

That's right; the DNG Profile Editor was only designed with visual-based tweaking/editing in mind, but doesn't contain visualizations / analysis / statistics tools that more advanced editors would have.
Logged

Rhossydd
Sr. Member
****
Offline Offline

Posts: 1965


WWW
« Reply #4 on: January 11, 2009, 07:50:43 AM »
ReplyReply

Thanks Eric, I do appreciate that the DNG editor, especially when itís still only in beta) isnít going to arrive with every possible facility anyone could ever want. Itís pretty damn good already.
I had just hoped that someone else might have had similar curiosities about it and might already have made a utility to make more sense of the differences between them, I guess that may come in time.

Iíve ended up creating a simple spreadsheet based on the data present when viewing the recipes in the editor.
Itís shown me pretty much what I was expecting; There is a commonality in values between all the various profiles created for a specific camera body, but one or two show significant deviation from the others for some reason.
As a result Iíve been able to create a profile based on the average and median values for the adjustments to create a generic custom profile (if that isnít an oxymoron).

Iíve had a closer look at the data in the recipe files and wonder if you could confirm a couple of observations.
The automatic profile creation only creates a profile based on the 18 point colour edits based on the coloured patches. Those edits will only be for hue and saturation, never lightness.
Thereís no corrections made in tone curve response or the colour matrices by the automatic chart function.
Logged
madmanchan
Sr. Member
****
Offline Offline

Posts: 2110


« Reply #5 on: January 11, 2009, 10:10:12 AM »
ReplyReply

Quote from: Rhossydd
I had just hoped that someone else might have had similar curiosities about it and might already have made a utility to make more sense of the differences between them, I guess that may come in time.

In time, you just may get your wish. (Hint: Adobe is not the only company building DNG profile tools.)


Quote
Iíve had a closer look at the data in the recipe files and wonder if you could confirm a couple of observations.
The automatic profile creation only creates a profile based on the 18 point colour edits based on the coloured patches. Those edits will only be for hue and saturation, never lightness.
Thereís no corrections made in tone curve response or the colour matrices by the automatic chart function.

All correct.
Logged

Rhossydd
Sr. Member
****
Offline Offline

Posts: 1965


WWW
« Reply #6 on: January 11, 2009, 10:31:37 AM »
ReplyReply

Thanks Eric :-)
Logged
Nat Coalson
Full Member
***
Offline Offline

Posts: 195


WWW
« Reply #7 on: January 16, 2009, 08:32:54 AM »
ReplyReply

You can use virtual copies to see the different profiles side by side; I find this makes it much easier to compare results.
Logged

digitaldog
Sr. Member
****
Offline Offline

Posts: 9191



WWW
« Reply #8 on: January 16, 2009, 09:03:09 AM »
ReplyReply

I too have been playing with the Macbeth profile generation in the new editor since I just got a new 5DMII. The custom profile I built is IMHO superior to the supplied one but I'm still on the fence about a few areas until I shoot more images and apply the new profile. But certainly a very easy, quick process and vastly preferable to the old scripts. I built a single and dual illuminant profile(s), still seeing what that brings to the party. Really need to shoot a lot more images and compare, but as a default, I'm pretty happy with the custom profile.
Logged

Andrew Rodney
Author ďColor Management for PhotographersĒ
http://digitaldog.net/
mcmorrison
Full Member
***
Offline Offline

Posts: 152


« Reply #9 on: June 03, 2009, 06:24:47 PM »
ReplyReply

Hello,

I have finally gotten to trying out the DNG Profile Editor, and am very happy with the results. I have wrestled with the color in a number of images, feeling that they somehow did not come through correctly. Now, using a custom profile, they seem to be much truer and better behaved. Thanks for this!

I have made a profile at noon in direct sunlight using a Mini Color Checker, and have been applying this to a range of photos. I wonder what the intended strategy is: should I also be making profiles for overcast? For tungsten lighting? For high altitude lighting? What is the "ideal" lighting to make a single-table profile?

Also, I use mostly Solux lighting, and so don't have a 6500 source to make the two-table profile. How are the two tables created from one image when using the one-table approach, and what is given up by doing so? Would it be valuable to make a two table profile using a 5K Solux and a regular tungsten image, or would this confuse the 6500 assumptions in the software?

Thanks in advance,

Michael Morrison
Logged
madmanchan
Sr. Member
****
Offline Offline

Posts: 2110


« Reply #10 on: June 04, 2009, 10:19:05 AM »
ReplyReply

I would not bother to create separate profiles for different flavors of daylight, unless you are extraordinarily picky. The differences will generally be subtle.

I would, however, suggest having a separate profile for tungsten, or as a rule of thumb any time you do a click-WB on the gray patch of your mini-CC and the temperature value is below 3500 K.

If you make a 2-table profile, the two tables will be interpolated automatically based on the image-specific white balance. In practice this works well for the ~ illuminant A / D65 case (substituting D55 or D75 for D65 also works fine), as long as you are photographing under some flavor of natural daylight or a tungsten bulb. If you switch to something wacky like an office fluorescent tube or a CFL bulb from Home Depot, then all bets are off.
Logged

mcmorrison
Full Member
***
Offline Offline

Posts: 152


« Reply #11 on: June 04, 2009, 10:45:06 AM »
ReplyReply

Hello Eric,

Thanks, this is very helpful.

What would you think of using a Solux 5K for the 6500 table? Or would I be better off using a cloudy day shot?

Tungsten I can arrange!

Thanks again,

Michael
Logged
madmanchan
Sr. Member
****
Offline Offline

Posts: 2110


« Reply #12 on: June 04, 2009, 11:46:00 AM »
ReplyReply

The Solux bulb should work fine. Note that I've found in practice the Solux bulbs tends to have lower actual CCT compared to what is stated on the box. For example, I have two 3500 K Solux bulbs that measure at about 3300 K (pretty close) and two 4700 K bulbs that measure at 4100 K (not that close). I don't know how the 5000 K bulbs do.
Logged

mcmorrison
Full Member
***
Offline Offline

Posts: 152


« Reply #13 on: June 04, 2009, 11:57:01 AM »
ReplyReply

Hello Eric,

Thanks.

Yes, I too find the Soluxes have a slightly lower temperature than rated. As I understand it, increasing the voltage will increase the temperature at the expense of bulb life. I haven't got a transformer handy to push a bulb higher for the color table, but I'll probably keep my eye out for one now.

Another option to up the voltage as I understand it is to put 9 bulbs on a 120 volt circuit (no transformer) so that each bulb gets 13.3 volts instead of its rated 12. I haven't got the gear for this at the moment either. . .

Thanks,

Michael
Logged
mcmorrison
Full Member
***
Offline Offline

Posts: 152


« Reply #14 on: June 12, 2009, 11:56:25 AM »
ReplyReply

Hello Eric,

Do you have any thoughts on using sunrise or sunset lighting for the 2800 K profile table? Most of my low temperature shots are in just this type of light, and it occurs to me that it might be preferable to use that light directly for the DNG profile creation rather than tungsten. Do the routines depend on the tungsten spectrum somehow, or would they adjust well to sunrise/sunset light?

Also, assuming effective control of glare, would there be any meaningful improvement in the DNG profiles created if a Color Checker SG option was added?

Thanks again for this great tool. It is proving valuable in many ways!

Michael Morrison
Logged
madmanchan
Sr. Member
****
Offline Offline

Posts: 2110


« Reply #15 on: June 12, 2009, 03:21:07 PM »
ReplyReply

Michael, I would recommend building a single-illuminant profile for the common low temperature conditions you shoot in. I would not expect much improvement, if any, by using the SG chart.
Logged

mcmorrison
Full Member
***
Offline Offline

Posts: 152


« Reply #16 on: June 12, 2009, 04:02:35 PM »
ReplyReply

Hello Eric,

Thanks for the insights!

I understand that you recommend a one-table profile for sunrise/sunset lighting.  Do I understand that if I want to create a two-table profile, I should use tungsten for the low-temperature table, not sunrise/sunset lighting?

Thanks again,

Michael
Logged
madmanchan
Sr. Member
****
Offline Offline

Posts: 2110


« Reply #17 on: June 17, 2009, 09:07:20 PM »
ReplyReply

That's right, Michael.
Logged

mcmorrison
Full Member
***
Offline Offline

Posts: 152


« Reply #18 on: September 02, 2009, 01:23:21 PM »
ReplyReply

Hello Eric,

I photographed a Color Checker this morning with skylight, and the CCT came to 11,000 įK (I'm at 7,000 feet altitude). I made a two chart profile using tungsten as the other end point. I've looked at a few files with the resulting profile and they look reasonable so far.

What are your thoughts on using 11,000 as the high-temperature end point? Would it usefully extend the profile, or would it distort the common temperature ranges around 5,500? Or?

Thanks,

Michael
Logged
madmanchan
Sr. Member
****
Offline Offline

Posts: 2110


« Reply #19 on: September 02, 2009, 02:52:54 PM »
ReplyReply

Hi Michael, it is probably going to affect color saturation to some degree for daylight images in the 4000 K to 7000 K range, but only subtly so. I do not think it will extend the profile by too much.
Logged

Pages: [1] 2 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad