|
Mark,
You could be right but I (we) don't know if PS
works with real number until the data needs to be save or whatever. If it's not
the case and this would be surprising, then each editing step would effect the
data and each additional editing steps would have a cumulative effect. I
don't think what you propose below is correct even if PS works with real number
because the starting point would be different.
Let's say I choose arbitraly a
normalised [0..1] pixel level of 0.5, lets see what our
starting value could be in 8 bit resolution. This would mean we started
with a value of 0.5*255=127.5 => 127 or 128, now in 16 bit this would
be 0.5*65535=32767.5 => 32767 or 32768. Lets say our original pixel
value is 32767, PS would transform this in 32767/65535 = 0.49999237 and we could
proceed with editing from this value but if we convert this value to 8 bit and
save it for later comparison, PS would need to convert this to a value to an
integer value between [0..255], so lets do this by 0.49999237 * 255 =
127.4980545 and to 127 by rounding. If we need to apply the same editing to
the 8 bit data file, PS will convert 127 to 127/255 = 0.498039215, a difference
of 0.001953155, though very small this difference could be increased by
editing the image further. Just out of curiosity say we apply a small gamma
correction of say 1/1.45 and see what effect it as.
0.49999237^(1/1.45) = 0.6199955
0.4980392^(1/1.45) = 0.6183242
for a difference of 0.001671312
The difference is now lower, this could be
surprising and it's why a full analysis including both color errors and
quantization errors migth produce interesting results. I made a search on Google
to see if someone did this kind of analysis but as of now (4) I saw only
perceptual comparisons and no number crunching but interestingly the
few test I've seen seem to indicate the difference are imperceptible from 8
or 16 bit editing on printed image. Food for thought don't you
think?
Regards
Yves
PS I don't recall any discussion of this topic in
particular but I have a bad memory so don't take my word for it.
----- Original Message -----
Sent: Sunday, December 16, 2007 3:37
PM
Subject: Re: OT: 16 bit editing myth or
reality?
Yves,
It would be easy to do such a
comparison....in the end, convert the 16 bit version to 8 bit, then count how
many levels of tone are contained in each file.
Mark
In a message
dated 12/16/07 2:19:30 PM, gauvreau-yves@cgocable.ca
writes:
Mark,
Happy holidays to you and all others
on the list.
In my last message, I hope I was
able to tell every one the context in which this (my) experiment was
done. Though I think most of the list member had the intuitive or
learned knowledge that working in higher bit resolution would minimise
quantization errors or simply result in higher quality image, it wasn't
obvious to me and probably to most of us here how this improved quality
translated into actual quantitative values.
As for doing what you
suggest below, it should be obvious that such an exegarated transform would
result in poor quality image for both resulution, though the 8 bit would
probably suffer the most. (I like very much porcupines...)
If I where to do an in depth
analysis of the "benefit" of using 16 bit resolution on B&W
image, I would try to recreate a typical set of editing transformation
as would be done on an actual image. Any out of the ordinary editing would
be counter productive in my opinion and the result wouldn't be that useful
since they wouldn't have any connection to real world image and real world
editing. In this type of analysis I would also use PS for the simple reason
that recreating the math behind some of the editing transform would be way
to much work and it wouldn't be relevant to the initial goal. When all
four image would be ready to print that is image #1: original
unedited 16 bit image, image #2: image #1 converted to 8 bit, image #3:
edited from image #1 and image #4: edited using the same editing step as for
image #3 but from image #2 instead, to be print ready image #3 would have to
be converted to 8 bit. Then I would load image #2 to #4 (image #1 is
redoundant) in my program or whatever to count the looses if any.
I could also do a little bit better by recording each individual editing
step by itself, this would allow one to evaluate the contribution of each
type of editing transform and the more global result of applying all of
these transform steps together. To be even more complete, this analysis
could be done on a color image and the stats should also
include CIE delta errors (1976) this would give a more complete picture
then just measuring quantization errors as I did.
I'll first seach to see
it as been done before and if not, I just might give it a try out of
curiosity and provide you with the result.
Regards
Yves
Best Wishes,
Mark
Nelson
Precision Digital Negatives -
The System
PDNPrint
Forum at Yahoo Groups
www.MarkINelsonPhoto.com
**************************************
See
AOL's top rated recipes
(http://food.aol.com/top-rated-recipes?NCID=aoltop00030000000004)
|