Ad
Ad
Ad
Pages: [1] 2 »   Bottom of Page
Print
Author Topic: Prophoto 1.8 gamma with 2.2 gamma displays  (Read 5967 times)
texshooter
Full Member
***
Offline Offline

Posts: 200


« on: August 05, 2013, 05:44:05 PM »
ReplyReply

I understand it is conventional wisdom to use the Prophoto 1.8 gamma color workspace (Even Lightroom requires it). But all displays use 2.2 gamma (Even Apple  switched to 2.2). Yet I hear nobody complain about the mismatch of using the 1.8 colorspace with 2.2 displays. Why is that? Wouldn't it be better if they were the same? Either both 1.8 or both 2.2?
Logged
digitaldog
Sr. Member
****
Online Online

Posts: 8107



WWW
« Reply #1 on: August 05, 2013, 05:45:32 PM »
ReplyReply

The TRC (gamma) of a working space and that of a display are totally separate, not related and don't have to match.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
Mac Mahon
Jr. Member
**
Offline Offline

Posts: 90


« Reply #2 on: August 05, 2013, 09:02:48 PM »
ReplyReply

I understand it is conventional wisdom to use the Prophoto 1.8 gamma color workspace (Even Lightroom requires it). But all displays use 2.2 gamma (Even Apple  switched to 2.2). Yet I hear nobody complain about the mismatch of using the 1.8 colorspace with 2.2 displays. Why is that? Wouldn't it be better if they were the same? Either both 1.8 or both 2.2?
As I understand it, Lightroom uses a linear TRC version of ProPhoto (not even 1.8 ) as its working space.  That's to ensure optimum adjustability of tonal data during processing.  The reason monitors are generally profiled with a non-linear TRC (gamma =2.2 typically) is to accommodate the fact that human visual response is non-linear.  The reason the two gammas don't have to match is that the colour management system, in applying a monitor profile, converts the image data, colour and TRC, from the working space to the monitor space. There's no significant loss or disadvantage in that conversion AFAIK.

Tim
Logged
Simon Garrett
Sr. Member
****
Offline Offline

Posts: 290


« Reply #3 on: August 06, 2013, 02:36:59 AM »
ReplyReply

The reason monitors are generally profiled with a non-linear TRC (gamma =2.2 typically) is to accommodate the fact that human visual response is non-linear.  
I'm not sure that's quite the reason.  The end-to-end TRC of the complete system - camera to raw processor to editor to monitor - is linear.  It must be, or the image on the monitor wouldn't look like the original scene.  

You're right that the response of our eyes is non-linear, but the same non-linear eyes view the original scene, so why would the monitor need non-linear response when the original isn't non-linear?  

In a colour-managed system, when an image is converted from one profile to another, the image data is converted from the TRC of the source data to the TRC specified in the target.  So as I understand it, if a monitor is calibrated to either a gamma of 1.8 or 2.2 then, provided the profile says that, the conversion process will convert the data to the correct TRC and the image will look right.  
« Last Edit: August 06, 2013, 08:31:34 AM by Simon Garrett » Logged
bjanes
Sr. Member
****
Offline Offline

Posts: 2718



« Reply #4 on: August 07, 2013, 08:02:38 AM »
ReplyReply

I'm not sure that's quite the reason.  The end-to-end TRC of the complete system - camera to raw processor to editor to monitor - is linear.  It must be, or the image on the monitor wouldn't look like the original scene.  

You're right that the response of our eyes is non-linear, but the same non-linear eyes view the original scene, so why would the monitor need non-linear response when the original isn't non-linear?  

In a colour-managed system, when an image is converted from one profile to another, the image data is converted from the TRC of the source data to the TRC specified in the target.  So as I understand it, if a monitor is calibrated to either a gamma of 1.8 or 2.2 then, provided the profile says that, the conversion process will convert the data to the correct TRC and the image will look right.  


Yes, the end to end TRC of the complete system must be linear for reasons you state. As explained in Poynton's Gamma FAQ gamma compression is performed when the image is encoded, and the data are re-expanded for display as shown in the figure below.

Where the non-linearity of human perception comes into play is in the matter of perceptual uniformity. With linear encoding, the perceived change in brightness of steps in the lower range is bigger than that of upper steps as shown in Figure 2 of Greg Ward's article on encodings. For editing an image, we want the perceived brightness imposed by a change in steps to be the same at the low end as in the high end. With linear encoding, everything is scrunched up towards the lower end.

Regards,

Bill
Logged
Simon Garrett
Sr. Member
****
Offline Offline

Posts: 290


« Reply #5 on: August 07, 2013, 10:39:01 AM »
ReplyReply

Thanks for the reference links.
For editing an image, we want the perceived brightness imposed by a change in steps to be the same at the low end as in the high end. With linear encoding, everything is scrunched up towards the lower end.
For storage it makes sense to provide perceptually uniform encoding - especially in 8-bit encoding - but I wonder if that's so for editing?  Lightroom, for example, uses a linear-encoded working space (for editing), I believe.  
Logged
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3021


« Reply #6 on: August 07, 2013, 11:03:54 AM »
ReplyReply

Thanks for the reference links.For storage it makes sense to provide perceptually uniform encoding - especially in 8-bit encoding - but I wonder if that's so for editing?  Lightroom, for example, uses a linear-encoded working space (for editing), I believe.  

Hi Simon,

Lots of image manipulations are mathematically much easier to perform in Linear gamma space, so I have to assume that good software uses that when needed/beneficial. Besides profile conversions, interpolation/resampling is another example, if one wants to avoid color artifacts.

Cheers,
Bart
Logged
bjanes
Sr. Member
****
Offline Offline

Posts: 2718



« Reply #7 on: August 07, 2013, 01:25:41 PM »
ReplyReply

Thanks for the reference links.For storage it makes sense to provide perceptually uniform encoding - especially in 8-bit encoding - but I wonder if that's so for editing?  Lightroom, for example, uses a linear-encoded working space (for editing), I believe.  

Lightroom does employ a linear editing space, but the readouts and histograms are in terms of an sRGB tone curve so that it appears that one is editing in a gamma encoded space. Very clever programming, indeed.

Bill
Logged
Mac Mahon
Jr. Member
**
Offline Offline

Posts: 90


« Reply #8 on: August 07, 2013, 04:12:26 PM »
ReplyReply

Where the non-linearity of human perception comes into play is in the matter of perceptual uniformity. With linear encoding, the perceived change in brightness of steps in the lower range is bigger than that of upper steps as shown in Figure 2 of Greg Ward's article on encodings. For editing an image, we want the perceived brightness imposed by a change in steps to be the same at the low end as in the high end. With linear encoding, everything is scrunched up towards the lower end.
Bill

Thanks for the useful references.

To be clear for the OP:  we edit in a linear (or more nearly linear) space because image data manipulation easier to control (mathematically), as Bart says;  but a linear space would not work for output because human perception (did  I say vision?) would see everything as scrunched up at the lower end, as you say.

The answer to the OP's question, then, is we have to have something like gamma 2.2 on the monitor to account for the (non-linear) feature of human perception, and we get better editing control in (near) linear space. So it is sensible that the gammas of the editing and monitor spaces are not the same.

No?

Tim
Logged
texshooter
Full Member
***
Offline Offline

Posts: 200


« Reply #9 on: August 07, 2013, 04:25:45 PM »
ReplyReply

Bill

Thanks for the useful references.

To be clear for the OP:  we edit in a linear (or more nearly linear) space because image data manipulation easier to control (mathematically), as Bart says;
Tim

So why is the Prophoto space not 1.0 gamma instead of 1.8, since linear is better for editing? Would that be a case where more is not better? Why not use a 1.0 gamma space in Photoshop similar to Lightroom's linear space?
« Last Edit: August 07, 2013, 04:33:30 PM by texshooter » Logged
xpatUSA
Full Member
***
Offline Offline

Posts: 244



WWW
« Reply #10 on: August 07, 2013, 06:33:16 PM »
ReplyReply

So why is the Prophoto space not 1.0 gamma instead of 1.8 . . .

The ProPhoto space, formerly known as "Reference Output Medium Metric RGB (ROMM RGB)", was published by Kodak. They say:

Quote
This color space is intended to be used for performing manipulations on images that exist in a rendered image state.

. . and a rendered image needs a non-linear gamma and 1.8 was chosen so as to match the independent profile connection spaces XYZ or Lab. It follows therefore that there can not be such an animal such "Linear ProPhoto". That is to say, another space might have the same white reference and the same primaries as ProPhoto but, if it's gamma is not 1.8, it ain't ProPhoto.

You can read the whole paper here: kronometric.org/phot/xfer/4Chris/Kodak%20ROMM%20white%20paper.pdf

Kodak also published a matching un-rendered space - "RIMM RGB". They say:

Quote
A new color encoding known as Reference Input Medium Metric RGB (RIMM RGB)
is defined. This color encoding is intended to be used for manipulating images that exist
in a device-independent un-rendered image state. This color encoding was chosen to
provide a large enough color gamut to encompass most common input devices, and is
defined to be suitable as an input space to the ICC profile connection space (PCS).
Examples of manipulations that might be applied in this color encoding include scene
balance algorithms, manual color/density/contrast/ tone scale adjustments, red-eye
correction, and dust/scratch removal. The color encoding is also appropriate for archiving
and/or interchanging unrendered images. 8-bit, 12-bit and 16-bit versions of this color
encoding are defined.

Curiously, RIMM has a stated encoding gamma of 1/0.45 = approx 2.2 with a small linear portion.

So it may be that "image data" is always non-linear but, when editing, is converted to linear
and back (I think I read that somewhere).
« Last Edit: August 07, 2013, 09:12:49 PM by xpatUSA » Logged

best regards,

Ted
bjanes
Sr. Member
****
Offline Offline

Posts: 2718



« Reply #11 on: August 08, 2013, 09:52:49 AM »
ReplyReply

So it may be that "image data" is always non-linear but, when editing, is converted to linear
and back (I think I read that somewhere).

In Lightroom and ACR the editing space is linear, but Photoshop can edit in either linear or gamma encoded spaces. The preferred editing space in PS is ProPhotoRGB, but if one needs to use a linear (scene referred) space, one can convert to linear_RIMM_v4 as outlined in this post on the ICC web site. It is interesting to look at the histograms of a full range image in the gamma encoded space and that of the linear space. The ProPhotoRGB histogram is on top and the linear histogram is on the bottom. Note that the linear image, contrary to what is often stated, does  not appear dark if the image is tagged with the linear profile and color management is used. However, the linear histogram is scrunched up to the left, and it is difficult to determine if clipping is present as judged by the histogram. The gamma encoded space is preferred for editing since it is nearly perceptually uniform. However, as Bart has pointed out, some image manipulations are best done in a linear space.

Regards,

Bill
Logged
xpatUSA
Full Member
***
Offline Offline

Posts: 244



WWW
« Reply #12 on: August 08, 2013, 10:23:44 AM »
ReplyReply

Thanks for the clarification, Bill.

Logged

best regards,

Ted
tho_mas
Sr. Member
****
Offline Offline

Posts: 1692


« Reply #13 on: August 09, 2013, 04:53:16 PM »
ReplyReply

Lightroom does employ a linear editing space, but the readouts and histograms are in terms of an sRGB tone curve so that it appears that one is editing in a gamma encoded space. Very clever programming, indeed.
you call it "clever" when the histogram of a RAW software does not display the actual image data of the output??? Now, this is certainly a sophisticated way to blandish a serious shortcoming of a software.
Logged
Guillermo Luijk
Sr. Member
****
Offline Offline

Posts: 1273



WWW
« Reply #14 on: August 09, 2013, 05:25:29 PM »
ReplyReply

you call it "clever" when the histogram of a RAW software does not display the actual image data of the output??? Now, this is certainly a sophisticated way to blandish a serious shortcoming of a software.

While linear edition may have advantages, a linear histogram representation hasn't. LR takes the best of both worlds.
Logged

elied
Full Member
***
Offline Offline

Posts: 236


« Reply #15 on: August 09, 2013, 06:47:45 PM »
ReplyReply

you call it "clever" when the histogram of a RAW software does not display the actual image data of the output??? Now, this is certainly a sophisticated way to blandish a serious shortcoming of a software.
But it does display it. It is called soft proofing. Or are you still on LR3?
« Last Edit: August 09, 2013, 06:49:42 PM by elied » Logged

Roll over Ed Weston,
Tell Ansel Adams the news
WombatHorror
Sr. Member
****
Offline Offline

Posts: 297


« Reply #16 on: September 05, 2013, 09:14:25 PM »
ReplyReply

I understand it is conventional wisdom to use the Prophoto 1.8 gamma color workspace (Even Lightroom requires it). But all displays use 2.2 gamma (Even Apple  switched to 2.2). Yet I hear nobody complain about the mismatch of using the 1.8 colorspace with 2.2 displays. Why is that? Wouldn't it be better if they were the same? Either both 1.8 or both 2.2?

Because you have to use fully-color managed software to view them to begin with and fully color managed software manages everything including translating from file tone response curve to monitor tone response curve. And most people store them as 16bits so you don't really get any banding during such translations.

Also all displays don't use gamma 2.2. Out of the box they can be all over the place. Most people do calibrate them to gamma 2.2 although people sometimes use sRGB TRC calibration at times instead too (and some use gamma 2.3 or 2.4 for movie viewing in cave-dark rooms, etc.)
Logged
madmanchan
Sr. Member
****
Offline Offline

Posts: 2100


« Reply #17 on: September 06, 2013, 09:19:19 AM »
ReplyReply

The TRC (gamma) of a working space and that of a display are totally separate, not related and don't have to match.

This is the crucial point.  Lightroom and other image editing apps may use various response curves for the purposes of image editing.  This is completely separate from the response curve of the display.
Logged

Tim Lookingbill
Sr. Member
****
Offline Offline

Posts: 1055



WWW
« Reply #18 on: September 06, 2013, 08:40:09 PM »
ReplyReply

This is the crucial point.  Lightroom and other image editing apps may use various response curves for the purposes of image editing.  This is completely separate from the response curve of the display.

I'ld take a guess this can primarily affect the appearance of color managed previews editing tonal roll off into black for shadow detail and highlight detail into 255 white with tweaks to the curve point tool, right?
Logged
Schewe
Sr. Member
****
Online Online

Posts: 5258


WWW
« Reply #19 on: September 06, 2013, 08:44:09 PM »
ReplyReply

I'ld take a guess this can primarily affect the appearance of color managed previews editing tonal roll off into black for shadow detail and highlight detail into 255 white with tweaks to the curve point tool, right?

If you start off with an 8-bit display pipeline and apply a white balance and gamma correction on top in the video LUTs, yes...if you have a display that does it's internal white balance and gamma adjustment in 10-bit precision, no, it won't really be a problem. That's the difference in the displays from EIZO & NEC that do their internal adjustments in 10-bit.
Logged
Pages: [1] 2 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad