[quote name='wim' timestamp='1285534192' post='3269']
I was quite dissatisfied in the beginning when resizing files, and after some research and experimenting I actually found a way to get around the problems when viewing files on the web. Rather than reduce in one go, I do this in steps of approximately 2/3 of the previous size, and I use bicubic sharpen, every second reduction step as a general rule. 2/3 the original size or slightly bigger, like 0.7X, makes the software work harder by it having to regenerate and interpolate an entirely new image. The danger with downsizing in larger steps, especially steps of 0.5X and smaller, may make it simpler and faster for the software, because it may only have to throw away pixels, but I find the results not as good. I do get artefacts I find when I use the faster step downs.
[/quote]
I believe that this way produced the better results, when programs did the reduction in pixels by simple "resize" operations. But today most (if not all) programs at least give you the choice how the reduction is done. (IrfanView for instance (and because I mentioned it before) will let you choose between "resize" (quick but leading to somewhat dissatisfying results) and "resample ... along with a selection of resample functions" (MUCH better results, but takes noticably longer). Today I belive the cumbersome process of reducing the size in several small steps including a certain amount of sharpening can savely be skipped, if you select the resample-way (rather than the resize-way) to let your images "lose some weight". At least, in the few tests I did (when I heard of the method mentioned by Wim) I couldn't see a noticable difference to the result of IrfanView (with resample + the Lancos-function) or PS/PSE and its bicubic-functions. But ... to each his own ... you should certanly give it a try and see if it works for you.
Rainer
I was quite dissatisfied in the beginning when resizing files, and after some research and experimenting I actually found a way to get around the problems when viewing files on the web. Rather than reduce in one go, I do this in steps of approximately 2/3 of the previous size, and I use bicubic sharpen, every second reduction step as a general rule. 2/3 the original size or slightly bigger, like 0.7X, makes the software work harder by it having to regenerate and interpolate an entirely new image. The danger with downsizing in larger steps, especially steps of 0.5X and smaller, may make it simpler and faster for the software, because it may only have to throw away pixels, but I find the results not as good. I do get artefacts I find when I use the faster step downs.
[/quote]
I believe that this way produced the better results, when programs did the reduction in pixels by simple "resize" operations. But today most (if not all) programs at least give you the choice how the reduction is done. (IrfanView for instance (and because I mentioned it before) will let you choose between "resize" (quick but leading to somewhat dissatisfying results) and "resample ... along with a selection of resample functions" (MUCH better results, but takes noticably longer). Today I belive the cumbersome process of reducing the size in several small steps including a certain amount of sharpening can savely be skipped, if you select the resample-way (rather than the resize-way) to let your images "lose some weight". At least, in the few tests I did (when I heard of the method mentioned by Wim) I couldn't see a noticable difference to the result of IrfanView (with resample + the Lancos-function) or PS/PSE and its bicubic-functions. But ... to each his own ... you should certanly give it a try and see if it works for you.
Rainer