Different results of color space conversion

GS
Posted By
G Sch
Mar 12, 2009
Views
416
Replies
10
Status
Closed
I am converting a raw image.

1. First in ProPhoto, passing it to PS CS3, accepting ProPhoto (against the working color space), and then I convert it in sRGB in Edit.

2. Next, converting it in ProPhoto, but when CS3 receives it, I ask for immediate conversion in sRGB, the working space.

3. Third, I change the color sapace in ACR to sRGB and pass the image to CS3.

Of course, the ACR adjustment parameters are identical in the three processes.

1 and 3 are almost identical (a difference layer does show differences, but I don’t see them on the results without huge boosting, and that shows quite random, noise-like difference).

However, 1 and 2 are *vastly* different. The difference, boosted by 2 EV clearly shows the original texture, which is determined by a pecularity in the blue channel.

What is the explanation for the difference between the two conversion from ProPhoto to sRGB?

The conversion engine is Adobe (the conversion immediately at receiving the image does not ask me for the engine).

< http://www.panopeeper.com/Download/ProPhoto_to_sRGB_Discrepa ncy.tif> contains three layers with the three versions.

< http://www.panopeeper.com/Download/ProPhoto_to_sRGB_inProPho to.tif> is the unconverted, i.e. ProPhoto version.
Related Tags
color conversion difference result space

MacBook Pro 16” Mockups 🔥

– in 4 materials (clay versions included)

– 12 scenes

– 48 MacBook Pro 16″ mockups

– 6000 x 4500 px

DS
Dennis_S
Mar 12, 2009
The conversion engine is Adobe (the conversion immediately at receiving the image does not ask me for the engine).

Hi Gabor,

If I am not mistaken, the "immediate conversion" uses the conversion options you have set in the Edit>Color Settings. Are you sure those matched the options you chose when you did the Edit>Convert to Profile?

I played around a bit with your samples and I could get close to your "Converted when receiving" version by using the Microsoft ICM engine (other options like Dither and Black point comp didn’t produce big differences that I could see). Is it possible that is what you have as the engine in Edit>Color Settings?
F
Freeagent
Mar 12, 2009
Gabor,

I reproduced your exact steps (but in CS4), and there was no difference whatsoever between the three. Pitch black in difference blend mode.

What puzzles me with #2 is that the green channel shows no change, but the blue channel shifts downward, and the red channel shifts dramatically and actually goes to solid clipping.

This is weird.
GS
G Sch
Mar 12, 2009
I played around a bit with your samples and I could get close to your "Converted when receiving" version by using the Microsoft ICM engine (other options like Dither and Black point comp didn’t produce big differences that I could see). Is it possible that is what you have as the engine in Edit>Color Settings?

As I posted, I am using the Adobe engine.

I reproduced your exact steps (but in CS4), and there was no difference whatsoever between the three. Pitch black in difference blend mode.

I don’t understand how you reproduced these steps. The file I uploaded is already in sRGB.

Anyway, I repeated the entire procedude carefully, the result is the same.

The raw file can be downloaded from <http://www.panopeeper.com/Download/CCC_ISO0100_01208.ARW>, the adjustment parameters are in <http://www.panopeeper.com/Download/CCC_ISO0100_01208.xmp>

With these files it is possible to repeate the entire process.

Pls note, that the conversion from raw to TIFF occured in 16bit mode, I converted the demo file to 8bit in order to reduce the size.
GS
G Sch
Mar 12, 2009
Note: the ARW file is 36 MB large; although it is a very inspiring shot (a color checker sheet), please don’t download it just for fun, for my site will block further access for the day if the download activity is too high.
DS
Dennis_S
Mar 13, 2009
Gabor,

I ran your raw file using method 1 and 2 and they came out identical.

I understood that you used the Adobe engine for the Edit>Convert to Profile but your comment about the immediate conversion not asking for the engine led me to believe that perhaps you had forgotten that the immediate conversion uses the parameters as established in Edit>Color Settings.

So just to be sure, are the options you are using when you do the Edit>Convert to Profile the same as the options you have in Edit>Color Settings?
GS
G Sch
Mar 13, 2009
Dennis,

I misunderstood your justified question; specifically, I did not think, that Edit Color settings stores the settings independently of the Edit Convert option.

Big thanks, that was it! In the color settings the Microsoft engine was selected. I don’t even remember when I have played with this last time.

Now, this gives me a headache: where did I use the conversion immediately when opening the file, and what are the consequences?

Anyway, thanks again; case closed
PZ
Ping Zheng
Mar 13, 2009
The conversion engine is Adobe (the conversion immediately at receiving the image does not ask me for the engine).

I played around a bit with your samples and I could get close to your "Converted when receiving" version by using the Microsoft ICM engine (other options like Dither and Black point comp didn’t produce big differences that I could see). Is it possible that is what you have as the engine in Edit>Color Settings?

I got the same result as Gabor(The two images are different) when sue the Microsoft ICM engine Probably your default engine in Edit > Color Settings is Microsoft ICM
F
Freeagent
Mar 13, 2009
I don’t understand how you reproduced these steps

Well, I have ACR too and some raw files lying around. But nevermind, it’s cleared up now.
GS
G Sch
Mar 13, 2009
I have ACR too and some raw files lying around

I don’t think a difference will be apparent with *any* raw file. I have been doing this nonsense for a while and this is the first case that a discrepancy was obvious. If you loaded this Sony raw file in ACR, ProPhoto, you have seen the huge difference between that and sRGB. In my praxis the effect of the sRGB limits appears usually in the very bright range, for example with red.

Anyway, thanks. I have learned now not only to avoid the MS engine, but to watch for out-of gamut indication even in images, which do not appear suspicious to me.
F
Freeagent
Mar 13, 2009
In my praxis the effect of the sRGB limits appears usually in the very bright range, for example with red

Not necessarily. A dark saturated color can easily clip in the low end in sRGB. This one, for instance. The ProPhoto histogram on top, same converted to sRGB below.

Another real-life problem is yellow-green foliage, which will often show low-end blue clipping in sRGB. I do a lot of nature photography, forest interiors and so on, and this low-end blue clipping is a constant problem. But that’s another discussion.

Must-have mockup pack for every graphic designer 🔥🔥🔥

Easy-to-use drag-n-drop Photoshop scene creator with more than 2800 items.

Related Discussion Topics

Nice and short text about related topics in discussion sections