Re: Photographer needs monitor recommendations(Eye-One/iCColor)
Re: Photographer needs monitor recommendations(Eye-One/iCColor)
- Subject: Re: Photographer needs monitor recommendations(Eye-One/iCColor)
- From: Armand Rosenberg <email@hidden>
- Date: Fri, 21 Nov 2003 19:45:46 -0500
At 1:30 PM -0800 11/21/03, email@hidden wrote:
> What point are you making though ? Number of samples doesn't equate to
anything in particular. 20 measurements of 50 msec duration are
going to capture the same number of photons as a single 1 second
measurement.
So, YOU'RE saying more samples per patch is no better than a single
measurement?
I guess my point was that when measuring inkjet targets that may have slight
density differences across the patches OR when measuring a press/proof
target that has halftone dots, more samples per patch can mean a better
average of that patch. I don't have a scientific study to back that up, but
intuition tells that SHOULD result in better data. It's just a hunch though.
Well, I do have a scientific background. Aside from boundaries of
patches (position of detector), there are 2 general (major) types of
error in any measurement: random and systematic ("drift"). The first
comes from the statistical nature of photons and electrons, the
second from equipment drift usually. In open-loop measurements like
those done here, neither error can be completely eliminated but
intelligent measurement methods can minimize both. Random errors are
minimized by gathering as many photons as possible, which simply
means a long measurement time. This can be accomplished by many short
measurements or fewer longer ones -- it's only the TOTAL measurement
time that matters for the random errors. This is what Graeme Gill
correctly implies in his message. It does not matter what you choose,
as long as the detector can handle the measurements. (For example,
you would use a CCD detector differently than a CMOS detector or a
photodiode array, etc.) Drift errors can be minimized by repeating
measurements as many times as possible, although you would have to
repeat an entire set of measurements (on all patches), not do a whole
series on a single patch then move on to the next one, in order to
have an effect on drift, usually. [More detail: the measurement
repetition time scale would really have to be appropriate to the
drift time scale to have an effect on drift errors. Drift is often
very difficult to eliminate however. Changing the order of measuring
patches each time can sometimes help with drift also.] So in this
particular case, it sounds like it really does not matter from a
measurement standpoint whether you do 1000 measurements of 1 ms
duration each, or 50 measurements of 20 ms duration each, since you
are not going back through ALL the patches 50 or 1000 times anyway
(it's one patch at a time) -- so the drift issue is not really
addressed, only the random errors. In either case you are gathering 1
s worth of photons and that's all that matters. The higher rep rate
will help in distinguishing steps (patch borders), as has been
pointed out, but it's only the total integration time per patch
(total photons captured per patch) that will matter to the
measurement accuracy. So number of samples without knowing the dwell
time per sample is a meaningless number.
There are complications: longer measurements may saturate some types
of detectors more quickly than others, forcing you to take shorter
measurements to avoid saturation (simply put, a condition where a
detector has gathered too many photons and is no longer useful --
please don't think saturation as in the pure color sense). There is
also electrical noise added in some detectors every time you do a
measurement, and this noise is independent of the light intensity,
which would make it preferable to take fewer but longer measurements.
Et al.
Basically, however, quoting a single number like number of samples
per patch is not meaningful by itself. You would have to know more
about the detector used, etc. to draw a useful conclusion. Unless
every brand/model of instrument is built exactly the same (not
likely), the samples per patch number is just one more toy in the
hands of the marketing department.
Armand
_______________________________________________
colorsync-users mailing list | email@hidden
Help/Unsubscribe/Archives:
http://www.lists.apple.com/mailman/listinfo/colorsync-users
Do not post admin requests to the list. They will be ignored.