[Note: this post has been rewritten to correct an error. The cause of the error was that Bruce Lindbloom’s desk image on his web site is in the sRGB, gamma 2.20 color space, while I thought it was in the sRGB IEC61966-2.1 color space. That caused some rather large errors, which are much smaller once the image is converted to sRGB IEC61966-2.1.]
In yesterday’s post, we saw that model-based color space conversion accuracy as performed in Matlab using 64-bit floating point intermediate values, are dominated by the quantizing error of 16-bit per color plane images.
But what about the accuracy such conversions in Photoshop? I took a look.
I loaded the same test image that I used for yesterday’s experiments into Ps (thanks to Bruce Lindbloom for the image):
I took the image from its native sRGB to Adobe (1998) RGB and back, using the Adobe (ACE) color engine, then I loaded the original image and the round-trip image into Matlab, converted both to Lab, and computed the deltaE for each pixel. Then I did that a few more times, starting with the last conversion and always comparing the latest image to the original one. I computed some stats on the deltaE image, and here’s what I got:.
Then I took the original image to Lab in Photoshop, then back to sRGB. I did that iteration several more times, comparing with the result for the first round trip, and got this:
With this test image, Photoshop’s profile conversion accuracy seems to be pretty darned good.