Lab to RGB conversion (16 bit) is different in CS and CS2

MB
Posted By
Mauro_Boscarol
Oct 3, 2006
Views
402
Replies
11
Status
Closed
I start with a Lab 16 bit color: 79 4 39.

If I convert this in Adobe RGB (16 bit) with ACE, rel col, bpc, I get two different results:

with Photoshop CS
R=27245 G=24457 B=16073

with Photoshop CS2
R=27232 G=24448 B=16057

This is only an example. Generally, Photoshop CS and CS2 converts from Lab to RGB differently.

Anyone can help me?

Thank you.

Mauro Boscarol

Master Retouching Hair

Learn how to rescue details, remove flyaways, add volume, and enhance the definition of hair in any photo. We break down every tool and technique in Photoshop to get picture-perfect hair, every time.

CC
Chris_Cox
Oct 3, 2006
CS2 will preserve more precision in the transforms than CS did. Also, bugs are fixed in the color engine with every release, and some of those fixes lead to improved precision or accuracy.
AR
alan_ruta
Oct 4, 2006
I’m not sayting quality doesn’t matter, but these are minute differences that I can’t believe would be noticeable on press.

alan
MO
Mike_Ornellas
Oct 4, 2006
Maybe not on press, but during the proofing process it may show a difference and that difference may be the make or break point in the editing and approval process of really hogh end work. That and having mutiple builds of the application in the real world creates small inconsistencies that may or may not create problems.
AR
alan_ruta
Oct 4, 2006
It curious. I’m going to run some tests and when I get a chance I slip them onto a proof here and there.

alan
MB
Mauro_Boscarol
Oct 4, 2006
I’m not interested in press or proofing or whatever, but I’m concerned with the Lab to RGB conversion only.

My point is: with 16 bit, the Lab to RGB conversion make with Photoshop CS is different from same conversion make with Photoshop CS2, and I am curious why the two are different.

Perhaps a new version of ACE?

In any case the differences appears evident in the histogram. With most images, the histograms *are* differerent: CS is more smooth, CS2 is not so smooth. See

<http://www.boscarol.com/test.html>

My question is: why is it so?

Thank you.

Mauro Boscarol
R
Ram
Oct 4, 2006
Chris Cox gave you the correct answer in post #1.
MB
Mauro_Boscarol
Oct 4, 2006
Sure, thank you Chris, the answer about Lab to RGB conversion is convincing.

If I can, I would append another question:

is it likely that a so small number difference (about 0.05 of a RGB unit) gives histograms apparently so different? (see <http://www.boscarol.com/test.html> ).

Thank you.

Mauro Boscarol
CC
Chris_Cox
Oct 4, 2006
Quite likely – that looks like the result of improved precision (leading to fewer quantization problems).
MB
Mauro_Boscarol
Oct 5, 2006
Frankly, quantization problems seems increased from CS to CS2. If a histogram in CS is smooth, in CS2 is more spiky.

Perhaps I don’t know the meaning of "quantization problem".

Thank you.

Mauro Boscarol
CC
Chris_Cox
Oct 5, 2006
The spikeyness of the histogram can result from many things. More accurate and precise results could actually lead to more spikes, it all depends on the data and the profiles involved.

Unless you’re measuring the exact values resulting from the conversions, and know the values inside the source and destination profiles to evaluate the "correct" results – you’re not going to be able to figure out where quantization issues might affect the result.
MB
Mauro_Boscarol
Oct 5, 2006
Thank you Chris for help me understand this subject.

Regards.

Mauro Boscarol

MacBook Pro 16” Mockups 🔥

– in 4 materials (clay versions included)

– 12 scenes

– 48 MacBook Pro 16″ mockups

– 6000 x 4500 px

Related Discussion Topics

Nice and short text about related topics in discussion sections