Very interesting, David. I'd love to see what your are doing with this smart blur. FWIW i still scan my MF negs with my Epson Perfection 3200 Photo (flatbed). It's quite good at this - not so good for 35mm negs. I must admit I haven't used it a lot since I got the *ist D :-)
Jens Bladt mailto:[EMAIL PROTECTED] http://hjem.get2net.dk/bladt -----Oprindelig meddelelse----- Fra: David Mann [mailto:[EMAIL PROTECTED] Sendt: 3. januar 2005 10:40 Til: [email protected] Emne: Re: Scanning Question On Jan 3, 2005, at 12:37 PM, Jens Bladt wrote: > A pixel doesn't really have a physical size* - it's a code in a > computer. Yes. We use the ppi figure to relate the dimensionless virtual pixel to the physical pixel which is really just a sampling point of a defined size. > Ppi only makes sence in connection with a media for viewing. It does also relate to the scanning process as it relates the filesize to the size of the media - in the same way as for printing. There is still a physical media involved, however instead of putting an image onto the media we're capturing the image from the media. > What really matters is the number of pixels in > total!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! The number of pixels relates to the square of the scan resolution, as you pointed out. I do agree with you that the filesize is an important factor, especially when comparing files from scanners with files from digital cameras (in which the resolution figure is unimportant as it only relates to the size and density of the sensor). But we need to be careful even when counting pixels. A 5 million pixel file off a digital camera will contain less detail than a 5 million pixel image from a film scanner, as the digital camera makes use of interpolation due to the Bayer pattern sensor. Looking at filesizes means other questions need to be asked, such as the format and sub-formats (eg TIF files can be saved with non-lossy compression) and the colour depth. Perhaps the file contains 12 bits per channel of actual information, padded and saved as 16 bits per channel. > 4000 ppi means 4 times more pixels in you image, than 2000 ppi (see the > above decribed example). > If there's not enough detail or info in your neg/slide/print, that you > are > scanneing, it won't help you to increase the number of pixels in your > scan. > In my experience (my scanner is not the worlds best) anything above > appr. > 2200 ppi is a waste of time and space. You are right here - but of course a better scanner is likely to extract more detail. I remember when I got my 2400ppi film scanner: I was totally blown away by the detail it could pull off the slide when compared to my 1200ppi model (which I still have as it's capable of scanning medium and large format). But that's not really much of a comparison as there is a law of diminishing returns - going from 2000ppi to 4000ppi will see a smaller improvement in image detail, at the price of a greatly enhanced ability at resolving film grain (and 4x the file size). I can see film grain on 100ASA colour slide films with the 2400ppi scanner (400ASA negs are horribly grainy and I have trouble producing a satisfactory 8x10 from them). The 1200ppi can't resolve the grain at all, so my MF scans turn out very smooth. Unfortunately the medium format scans don't do justice to the film. The 6x7 slide is about 4x the area of a 35mm slide, but the 35mm scanner is generating 4x the data per area. So the file sizes end up about the same - the only difference is the tonality as the MF scanner can't resolve the film grain at all. Not exactly a compelling reason to shoot MF but the major reason I bought that scanner was to be able to scan MF to the web. Now that I want to be able to make decent prints I've ordered a much better MF film scanner (a decision I did not make lightly). > If you loose detail is annother question, since the grainstructure in > the > film will never exactly "match" that of the "grid" in your scanner. I > guess > you'll always "loose" a little somthing, grain structure, colour depth > or > whatever. My first thought was that I'd hate it if grain had a regular pattern, as this would produce horrible interference patterns. But then I remembered that you could just use existing de-screening techniques to deal with it. So maybe it wouldn't be so bad. FWIW I've encountered tonality problems due to film grain and I've not really found an easy way to deal with it. I've experimented a bit with the Smart Blur filter in Photoshop. This can help smooth out the grain without losing detail in the photo but it makes the tonality worse as the colour discontinuities become larger in size (but remain the same in magnitude) and thus they become more obvious when viewing on-screen. It might be easier to explain that in pictures, which I'll try to do if anyone is interested. Cheers, - Dave http://www.digistar.com/~dmann/
