Ad
Ad
Ad
Pages: « 1 [2]   Bottom of Page
Print
Author Topic: Adobe DNG Profile Editor vs X-Rite Passport Editor  (Read 15086 times)
neil snape
Sr. Member
****
Offline Offline

Posts: 1432


WWW
« Reply #20 on: November 14, 2009, 11:44:23 AM »
ReplyReply

The Passport profiles are too contrasty compared to the base calibrations I agree.

But For the above images and many of my own the Passport profiles in some regions are more visually correct than the Adobe ones and vice versa as noted above .


Yes also to the use in studio.

I never thought that I would want to use a color correction for input, other than creating a custom calibration.

As it turns out I wasn't expecting a chart to build DNG profiles so easily which is exactly what is needed for correcting the camera input.

Any testing I had done years ago with camera profiles, showed that the only really solid use was studio copy lighting or very standardised lighting.

I am not yet convinced that these profiles work in so many different lights with variable ISOs, but do know the Passport is easy enough to include in any series and it builds a profile that approximates what is needed to align color input. I have already seen here that the Passport profile shot at ISO 800 is not working at other speeds so it tells me that trying a one size fits all is not the method I am going to use.

The Passport is too new to me to know how effective it will be in extreme conditions, time will tell.
Since I use a lot of gels on some pictures, I'll have to rely on a profile at that ISO then let it go. In that case whether or not it is an edited  DNG profile or PassPort would work out well.
Logged
madmanchan
Sr. Member
****
Offline Offline

Posts: 2100


« Reply #21 on: November 14, 2009, 04:49:58 PM »
ReplyReply

In a sense, it is not too different from the printer profile scenario, where one can take a set of (possibly averaged) measurements from a target and feed it to different profile-building software. They will all result in different "perceptual intents" in the resulting printer profile, optimized for different metrics, with different subjective results. See Bill Atkinson's site for more info.
Logged

bjanes
Sr. Member
****
Offline Offline

Posts: 2714



« Reply #22 on: November 19, 2009, 04:51:44 PM »
ReplyReply

Quote from: neil snape
Both are going about correcting colors by modifying the color matrix. The Passport is fully automated, easily transported, easily learned without any understanding of manually editing color patch values.

Quote from: sandymc
Not the case for the DNG profile editor. The matrixes are unchanged; what is written is a new HueSatMap table.
Sandy
I looked at the sizes of a Passport generated .dcp profile and one generated by the DNG profile editor.

The Passport profile was 1.47 KB whereas the DNG profile editor generated was 27.6 KB. 1.47 KB is not a lot of room for much of a lookup table, and is not that much larger than a matrix profile. I do not currently have any profile inspector software installed, but can you tell us exactly what is in these profiles?

Bill
Logged
TheSuede
Newbie
*
Offline Offline

Posts: 16


« Reply #23 on: November 19, 2009, 06:37:41 PM »
ReplyReply

The passport profiles are quite low-resolution, they divide the HSV colour-space into 6 divisions of hue (six huepoints or specific colours), 6 divisions of saturation of each specified hue, and 3 division of value (brightness) - making a very sparsely populated LUT of 6*6*3=108 points of correction.
+one matrix of course, but they're only 3x3 big - not much "space" needed there!

A typical dcp created/modified by the Adobe profile Editor will typically contain at least four matrices, and two LUT's of at least a thousand correction points each, depending on how densely you populate the HSV cylinder.
« Last Edit: November 19, 2009, 06:38:13 PM by TheSuede » Logged
TheSuede
Newbie
*
Offline Offline

Posts: 16


« Reply #24 on: November 19, 2009, 06:46:23 PM »
ReplyReply

Just had to add this quick remark also; Colour accuracy does not mean "colour exactly as reality" to me, as I learned CM in pre-press environments - it means that specific colours (a "colour" being a specified hue, saturation and radiation intensity) retain their approximate distances from each other all the way from reality to print. THIS orange is approximately this much brighter and more yellow than THAT orange. And when these oranges look "correct", intense red has the right hue - not too blue, not too yellow, not too weak/saturated. When this condition of "equal distances" is roughly satisfied for all the reproduceable colours all the way around the colour space that you have available in your presentation format (digital OR print), then you have a "good" profile.
« Last Edit: November 19, 2009, 06:46:43 PM by TheSuede » Logged
digitaldog
Sr. Member
****
Offline Offline

Posts: 8027



WWW
« Reply #25 on: November 19, 2009, 06:54:08 PM »
ReplyReply

Quote from: TheSuede
Just had to add this quick remark also; Colour accuracy does not mean "colour exactly as reality" to me, as I learned CM in pre-press environments - it means that specific colours (a "colour" being a specified hue, saturation and radiation intensity) retain their approximate distances from each other all the way from reality to print. THIS orange is approximately this much brighter and more yellow than THAT orange.


OK fine, but how do you measure or place a matrix on accuracy here? We can all measure things with a rubber ruler but anyone else can call into question the accuracy.

Or you can say “from a scale of one to ten, ten being best...” That’s fine but again, its subjective.

How do you propose one describes and measures “their approximate distances from each other“?
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
TheSuede
Newbie
*
Offline Offline

Posts: 16


« Reply #26 on: November 19, 2009, 08:17:46 PM »
ReplyReply

Well, that's easy from one point of view, and hard from another... :-)

A certain "hue", "saturation" and "brightness" can be measured, but only if you give the measurement some certain starting points. Then you have to decide by what metric the distances are to be measured in, which is not easy as humans are a lot more sensitive to hue-change in some hues, and more sensitive to other hue changes in darker lighting conditions or other lighting temperatures. same goes for brightness changes in different hues. That's also why it CAN'T be objectively measured, the metrics are (and cannot be!) the same as the ones used for measuring reality. You have to stay within the target output referred space, with the set whitepoint of that media. Those will never be the same as "reality". Well, they CAN be, but that's only the case in very specific situations, not likely to correspond to any photograph that you can take in an actual, physical situation...

"Approximate difference" means that when something that looks a certain green in a certain light and another thing looks a certain red in the same light, they will appear to have the same inter-relation to each other when reproduced in/on another media. That if they seemed to have the same saturation and brightness in reality, that that impression is preserved in the output referred media. How that is done, or "measured" is up to you. Others have spent years of their lives doing this, and there are still several thousand man-hours per day spent around the world on this in research centers, software development companies, print-shops, pre-press development departments and so on doing exactly this.... When you have a definite answer, give me a call. I'll be wanting to get in on that.
« Last Edit: November 19, 2009, 08:20:30 PM by TheSuede » Logged
sandymc
Full Member
***
Offline Offline

Posts: 235


« Reply #27 on: November 20, 2009, 03:49:40 AM »
ReplyReply

Quote from: bjanes
I looked at the sizes of a Passport generated .dcp profile and one generated by the DNG profile editor.

The Passport profile was 1.47 KB whereas the DNG profile editor generated was 27.6 KB. 1.47 KB is not a lot of room for much of a lookup table, and is not that much larger than a matrix profile. I do not currently have any profile inspector software installed, but can you tell us exactly what is in these profiles?

Bill

A typical DNG profile editor based profile has:

a. Two color matrixes
b. Two forward matrixes
c. A HueSatDelta table of 90x25x1
d. A tone curve with 96 entries

Plus the other bits and pieces you'd expect to create a valid profile.

As regards the Passport version, post one somewhere and I'll be happy to deconstruct it, but I would think that what TheSuide says a post above would be right.

Sandy
« Last Edit: November 20, 2009, 03:50:23 AM by sandymc » Logged
neil snape
Sr. Member
****
Offline Offline

Posts: 1432


WWW
« Reply #28 on: November 20, 2009, 04:04:16 AM »
ReplyReply

Quote from: sandymc
A typical DNG profile editor based profile has:

a. Two color matrixes
b. Two forward matrixes
c. A HueSatDelta table of 90x25x1
d. A tone curve with 96 entries

Plus the other bits and pieces you'd expect to create a valid profile.

As regards the Passport version, post one somewhere and I'll be happy to deconstruct it, but I would think that what TheSuide says a post above would be right.

Sandy
Good to know.

So in the end, what exactly is changed when you use the DNG editor. I had thought it would have been some type of curve, but can't be if there are only matrix grids. Are the forward matrices and HueSatDelta changed according to your user input?

you can download a dcp here: PassPort profile

I am very curious to find out what is going on in the profiles.
Logged
madmanchan
Sr. Member
****
Offline Offline

Posts: 2100


« Reply #29 on: November 20, 2009, 08:55:36 AM »
ReplyReply

Neil, if you select a Passport profile as your "Base Profile" in DNG Profile Editor, it will respect all the contents of that profile as the starting point for your edits. This includes the color matrices, tone curve, and lookup table. Any edits you apply get "folded in" on top of the base profile.

However, if you choose to use the Chart Wizard feature of the DNG PE (as opposed to your own edits in the first 3 tabs of the DNG PE), then it will start over from scratch. You'll know this because if you go back to check the first tab of the DNG PE, you'll see that the Base Profile has been switched to "ColorChecker" instead of whatever your Passport profile is named.
« Last Edit: November 20, 2009, 08:56:08 AM by madmanchan » Logged

neil snape
Sr. Member
****
Offline Offline

Posts: 1432


WWW
« Reply #30 on: November 20, 2009, 09:00:30 AM »
ReplyReply

Quote from: madmanchan
Neil, if you select a Passport profile as your "Base Profile" in DNG Profile Editor, it will respect all the contents of that profile as the starting point for your edits. This includes the color matrices, tone curve, and lookup table. Any edits you apply get "folded in" on top of the base profile.

However, if you choose to use the Chart Wizard feature of the DNG PE (as opposed to your own edits in the first 3 tabs of the DNG PE), then it will start over from scratch. You'll know this because if you go back to check the first tab of the DNG PE, you'll see that the Base Profile has been switched to "ColorChecker" instead of whatever your Passport profile is named.


Thanks!

I just noticed that this afternoon. I like the logic behind the editor. It is easy to use, and perhaps combined with the new chart (PassPort) it gives you chance to tweak or personalise a profile.

Early on in this thread someone wrote that the matrices are not changed. I am curious as to what happens when we fold in our colour preferences into the newly created profile?
Logged
sandymc
Full Member
***
Offline Offline

Posts: 235


« Reply #31 on: November 20, 2009, 09:57:58 AM »
ReplyReply

Quote from: neil snape
Good to know.

So in the end, what exactly is changed when you use the DNG editor. I had thought it would have been some type of curve, but can't be if there are only matrix grids. Are the forward matrices and HueSatDelta changed according to your user input?

you can download a dcp here: PassPort profile

I am very curious to find out what is going on in the profiles.

Neil,

Thanks for that file. It has:

1. a single color matrix, with the light source set to D55

2. No Forward matrix

3. A HueSatDelta table of 6x6x3

4. No tone curve.

Pretty interesting, actually - that they are using a three-D table is a bit of a surprise. Although the delta Vs that they are using are very small, and closely clustered to unity. So e.g., the a sample of the three V entries are:

    <Element HueDiv="5" SatDiv="5" ValDiv="0" HueShift="0.000000" SatScale="1.000000" ValScale="1.000000"/>
    <Element HueDiv="5" SatDiv="5" ValDiv="1" HueShift="0.753359" SatScale="0.997588" ValScale="1.029834"/>
    <Element HueDiv="5" SatDiv="5" ValDiv="2" HueShift="0.635227" SatScale="0.997903" ValScale="1.022837"/>

Not sure what they're getting from such small shifts.

Sandy
Logged
TheSuede
Newbie
*
Offline Offline

Posts: 16


« Reply #32 on: November 20, 2009, 06:55:09 PM »
ReplyReply

My understanding by looking at the passport profiles is that they are trying to avoid channel clipping at the black end, as you would/could get by trying to modify a HSV matrix when RGB values are zero or close to zero. Deconvoluting the HSV to RGB would push values below zero if you try to "shift" the hues/saturations.
Logged
mcmorrison
Full Member
***
Offline Offline

Posts: 152


« Reply #33 on: December 19, 2009, 01:56:15 PM »
ReplyReply

Hello,

If, as Sandy indicates, the Passport software makes a much smaller profile with fewer parts, we might expect this to translate into different quality. So far, I can see visual differences between PP profiles and profiles made by DNG PE, but can't tell much about the differences or whether there is a "better". I sure like the PP user interface better, showing the sampling squares that will be used. Does anyone have insight into this?

Also, Sandy indicated that the PP profile she dissected used a light source of D55. Is this always used? Or does the editor choose a light source based on the white balance patch in the ColorChecker? Likewise, what light sources does it use for two-chart profiles?

Thanks,

Michael
Logged
b2martin
Full Member
***
Offline Offline

Posts: 130


« Reply #34 on: December 20, 2009, 11:46:12 AM »
ReplyReply

Eric, I am trying to understand your comment about the "Chart Wizard" tab of the Adobe DNG Profile Editor.  If I use the "Chart Wizard" tab of the Adobe DNG Profile Editor does you comment mean that it makes no difference what I select as a base profile, the resulting profile is independent of the base profile selected?

If this is the case, is there any way to use the Adobe DNG Profile Editor to calibrate the colors of the "Base Profile" to a standard like the 24 patch color checker chart and keep all the other characteristics of the "base Profile"?
Logged
clayh
Newbie
*
Offline Offline

Posts: 18


« Reply #35 on: December 22, 2009, 12:28:18 PM »
ReplyReply

FWIW, last weekend I used the Passport Editor to create some dual-illuminant profiles for my D3x, M9 and GF-1. The results were sort of mixed. The D3X profile it created only varied slightly from the canned Adobe profile. The M9 profile was better than the Adobe standard, but not as good as the 'embedded' profile. And the GF-1 seemed to cause more highlight clipping than the Adobe standard. I'm beginning to think that the main use of this little plastic folder will be in creating custom white balances when shooting. As a previous poster pointed out, the creation of a standard color profile is not always a strictly objective numerical task. Ultimately, it has to look right, no matter what your numbers may be telling you.
Logged
TheSuede
Newbie
*
Offline Offline

Posts: 16


« Reply #36 on: December 22, 2009, 08:02:26 PM »
ReplyReply

Quote from: b2martin
Eric, I am trying to understand your comment about the "Chart Wizard" tab of the Adobe DNG Profile Editor.  If I use the "Chart Wizard" tab of the Adobe DNG Profile Editor does you comment mean that it makes no difference what I select as a base profile, the resulting profile is independent of the base profile selected?

If this is the case, is there any way to use the Adobe DNG Profile Editor to calibrate the colors of the "Base Profile" to a standard like the 24 patch color checker chart and keep all the other characteristics of the "base Profile"?

You can - but only by applying some "manual" work. You need the reference values for the target you're shooting, and then you just add control-points in the HSL-adjustment pane and start tweaking manually.
Logged
Pages: « 1 [2]   Top of Page
Print
Jump to:  

Ad
Ad
Ad