From: David J. Littleboy (davidjl@gol.com) Subject: Re: Epson 3200 Photo View this article only Newsgroups: rec.photo.equipment.medium-format Date: 2003-02-04 16:24:13 PST "Gavin Shaw" wrote: > > Got mine a few weeks ago. Took me a while to find the best way to hold > a 617 transparency close to the glass without touching. The speed is > quite good. The raw 3200dpi images are soft and need a fair amount of > unsharp mask in PS - I use a radius of about 3 pixels. You probably know this, but try light sharpening with wide radius, downsampling by about 70% (sqrt(2)), light sharpening with a smaller radius, and repeat with an even smaller radius. That'll give you a 3500x9900 or so image that is easier to adjust the final sharpening on, and still prints at almost 300 dpi for 12x34" prints. (And the smaller files will be much easier to deal with.) David J. Littleboy Tokyo, Japan From: Gavin Shaw (gavin.shaw@zetnet.co.uk) Subject: Re: Epson 3200 Photo View this article only Newsgroups: rec.photo.equipment.medium-format Date: 2003-02-05 13:24:04 PST The message from "David J. Littleboy" contains these words: > "Gavin Shaw" wrote: > > > > Got mine a few weeks ago. Took me a while to find the best way to hold > > a 617 transparency close to the glass without touching. The speed is > > quite good. The raw 3200dpi images are soft and need a fair amount of > > unsharp mask in PS - I use a radius of about 3 pixels. > You probably know this, but try light sharpening with wide radius, > downsampling by about 70% (sqrt(2)), light sharpening with a smaller radius, > and repeat with an even smaller radius. That'll give you a 3500x9900 or so > image that is easier to adjust the final sharpening on, and still prints at > almost 300 dpi for 12x34" prints. (And the smaller files will be much easier > to deal with.) > David J. Littleboy > Tokyo, Japan I haven't tried this systematic approach, but in this instance I've come close. The web shots are are result of scanning at 3200, sharpening, resampling to a 32.9 x ~100cm 360 dpi 'print' size, sharpening, resampling to 1280 x ~440 pixels, very slight sharpening. The first step is ~66% but the second is 10%. The processing is mighty slow with the full res 48bit first step because my PC is woefully short of RAM (512MB). Rotating the canvas from portrait is the most painful step! -- Gavin From: Gavin Shaw (gavin.shaw@zetnet.co.uk) Subject: Re: Epson 3200 Photo View: Complete Thread (16 articles) Original Format Newsgroups: rec.photo.equipment.medium-format Date: 2003-02-05 13:24:04 PST The message <43_%9.166892$rM2.134274@rwcrnsc53> from Leonard Evens contains these words: > Gavin Shaw wrote: [snip] > > I've just posted a few images at: > > > > http://www.users.zetnet.co.uk/gshaw/panoramics/panoramics_frame.htm > Great pictures, and you did an excellent job of preparing them for web > viewing. Do you have any hints about how to do that? I still haven't > figured out how to do it myself, and when I ask people who seem to have > the knack, they just tell me to do the usual things. Unfortunately, the > usual things don't seem to work as well for me. Hi Leonard, only the last 3 steps in the process could be considered specific to putting the pictures on a web page. For the record, here are the steps I used: Suspend the 617 transparency across the slot I've cut in a sheet of 0.5mm aluminium (all sharp edges removed and surface matt blackened). The film is held with peel-off 'Post-It' bookmark tabs which are great for size, easy to reposition and cause zero damage. Scan at 3200dpi, 48bit with Vuescan. Neutral colour balance, very minimal clipping of the histogram ends. Takes about 15 minutes, Firewire, 512MB DDR. 'Mirror' option checked as the transparency is emulsion downward. Save as 48bit LZW compressed TIFF. (Takes about 10 minutes... I know I need 2GB RAM!) Open in PS7 and rotate to landscape. (Takes.... well I don't know. I'm not in the room for this bit, I go and have a cup of tea or two.) De-spot the image with the clone tool and save as 48 bit internal ZIP-compressed TIFF - so I never have to scan the image again unless I have access to a better scanner. This compression allows the image to fit on a CD. Tweak the levels a little to try to replicate the original transparency as close as possible. Convert to 24-bit Unsharp mask (~70%, ~3 pixels, threshold 1) Resample to a size for printing (360ppi @ 32.9 x ~100cm) and unsharp ~1 pixel IIRC. Save as TIFF for printing. Resample to 'large monitor width' 72ppi @ 1280 x ~423 pixels and sharpen very slightly. I do this at 200% zoom so the effect is clear to see. Probably 60%, 0.7 pixels, threshold 6. 'Save for web', 'optimized' jpeg, 60-65 quality (enough to make the sky/land boundary 'jpegging' reasonably unobtrusive. Stick it on a web page, where the HTML specifies the actual pixel size, so it should appear as one image pixel per 'screen pixel' for optimum clarity. > > Over the next few days I hope to add a few zoomed-in views from the > > full res > > scans to show how far the detail holds up. > If you have some resolution test negatives which will allow us to count > line pairs per ten pixels or something similar, that would allow us to > get a direct comparison with the Epson 2450. I used to have a 35mm B&W test slide which came with a teleconverter, but I misplaced that a long time ago, sorry. > -- > Leonard Evens len@math.northwestern.edu 847-491-5537 > Dept. of Mathematics, Northwestern Univ., Evanston, IL 60208 -- Gavin