Take any image, cut the resolution by some factor, resampling with nearest neighbor interpolation (so you retain actual image data), and compare histogram to original, then cut it again... and so on and so on... You will see that the histogram remains unchanged for a good number of progressive subsamples. You could make a whole statistical study of why this works if you chose to:) Of course an arbritary section would not be an actual sample of the entire image and would probably not be representative of the image as a whole... Yes, the histogram is a great visualization tool for understanding your image... no arguement there:) mark --- In DigitalBlackandWhiteThePrint@yahoogroups.com, "J Michael Sullivan" <michael@h...> wrote: > Not sure what you mean by a small subsample. If you mean an arbitrary section of an > image, then I can assure you many of your images are being edited under a false > pretense! > > The premise of a histogram is to demonstrate the prevalance of data distribution in > an image. The begin and end points are what most people focus on, but they are only > one small aspect of the usefulness of a histogram. It is VERY useful to look at an > image and "see" that it is mostly high-key, or low-key, or mid-key, or evenly > distributed. Having such knowledge is important as to which is the best method (and > there are many) to enhance the most critical image detail, as there is no way that > EVERY pixel can be important -- to me this is the key of good scanning. > > MJS > (author of "Make Your Scanner a Great Design and Production Tool", now out of print) > > > --- In DigitalBlackandWhiteThePrint@yahoogroups.com, "Mark Hahn" > <markhahn2000@y...> wrote: > > I'm not sure that Adobe does it, but from my tests only a small sub- > > sample of an image is all that is really needed for a accurate > > Histogram... the big question is how useful is it really? I mean, > > are you really not going to apply the specific curve to an image > > which makes it *right* because you don't like the look of the > > histogram? > > > > mark > > > > --- In DigitalBlackandWhiteThePrint@yahoogroups.com, "J Michael > > Sullivan" <michael@h...> wrote: > > > I'm sure the Adobe dudes figured out a way to optimize it for the > > G5 chip too :( > > > > > > No matter, moving Gigabytes of 16bit layers is gonna require one > > kickass machine! > > > > > > mjs > > > > > > > > > > > > --- In DigitalBlackandWhiteThePrint@yahoogroups.com, "Stephen > > Kobrin" > > > <skobrin@h...> wrote: > > > > This question may be rendered moot in Photoshop CS as according > > to > > > > Evening's introduction you can have a "live historgram" running > > while > > > > you use curves. As I understand it, you will be able to see the > > > > impact of any curves ajustment on the histogram in real time. > > > > > > > > The bad news, at least for some of us, is that it looks like CS > > > > requires a Pentium III processor. I have a II with plenty of > > memory > > > > that works just fine with PS 7. > > > > > > > > Steve
Message
Re: levels and curves
2003-10-02 by Mark Hahn
Attachments
- No local attachments were found for this message.