Normal file size????????

[rdenney]

I previously used the method described by Frank, with good results. However some of my earlier LF negs I've had scanned by a lab were all done according to desired print size, i.e. separate scans were done according to whether I wanted 8x10 or 16x20 prints. I believe this is standard practice.

You never really know what size you'll want to print something. Scanning at the maximum capability of the scanning device will give you all there is, and if you need more, you have to find a scanner with more resolution.

But there is a deeper point in Frank's work flow, and that's the difference between correcting and targeting. When you first scan a file, you work at the scanner's best resolution, and make the image look the way you want it to look at that resolution. Sharpening here is done for correction--merely to correct for the smear of pixels caused by the scanner or sensor technology. Also, you correct the color for your calibrated and profiled screen, not for your output devices. You also apply whatever creative process you desire to be included in every display of the print, which I call correcting the image to suit your visualization.

The corrected file, once saved, becomes the base file for future application. You'll only replace it when you need something it doesn't offer, which you can obtain by using a better scanning technology.

If you want to make a print from that file, then copy it, and resize it to the optimum output of the printer. As long as the corrected file and the targeted file are at or above the printer's useful maximum resolution, you'll never see any effects of the resizing. If both are at or above your eye's abilities, then you also won't see any effects. Sharpening and color correction for targeting are designed to make it look good on the output device being targeted.

Example: I scan 4x5 in my Epson flatbed at 2400 pixels/inch. That gives me about 8800x11000 pixels in the file. I correct the color to look correct on my calibrated monitor, adjust image settings (curves, etc.), apply any effects such as dodging and burning, spot the file, and apply just enough sharpening to overcome the mild fuzziness of the scanner at that resolution. That sharpening is done at 100% actual pixels on the screen. It would never be visible if the whole image was shrunk sufficiently to fit on the screen. The resulting corrected file represents all I can get out of my scanner.

Then, if I want to make a 16x20 print, I copy the file, view it in Photoshop using the Epson 3800 profile, adjust the color and image settings to look correct with that profile, and then I resize it (once) to that printer's ideal output. My printer looks very good at 240, but it's at its best at 360. My scan supports over 500. I can resize that to 360, and since both resolutions are beyond anything I can see, any inaccuracies in the interpolation will also be too small to see. After resizing, I sharpen it to look good at that resolution. The resized file that is adjusted to look good on the Epson 3800 becomes my targeted file. An HP printer would prefer 300 rather than 360--the targeted resolution is display device-specific.

Rick "whose workflow avoided many do-overs after separating correcting and targeting steps" Denney

[jim kitchen]

Dear Rick,

My scanning process is very similar, where I acquire an RGB file that becomes a master file without sharpening applied during the original scan, simply because my negatives are 8X10. Did I interpret your comment incorrectly, that you sharpen during a scan? Although my Epson 750 scanner produces a softer image even at the optimum focal point, I am not bothered by this today, and I know that I could acquire a sharper drum scanned image, complete with better shadow detail, but that is outside this discussion...

I do not touch this master file, but copy it, rename it, and make the renamed file my working file, going forward. I implement and apply other processes downstream against the copied working file, but our, et al, final file size thought process, happens to be the same, although my files are set to be output at 720 for each finished print.

jim k

[rdenney]

Dear Rick,

My scanning process is very similar, where I acquire an RGB file that becomes a master file without sharpening applied during the original scan, simply because my negatives are 8X10. Did I interpret your comment incorrectly, that you sharpen during a scan?

No. I do it in Photoshop, where I can tailor the sharpening to just correct for the fuzziness of the scanner sensor and no more. I do this also for digital camera images, where I work from unsharpened RAW files and apply just enough sharpening to correct the effects of the anti-aliasing filter. Again, this is a correction step--just to overcome the obvious effects of the digital capture device, not to adjust its appearance at any particular downstream resolution.

Even if I don't resize the image for a given print, I may sharpen again during targeting just because I see the image with a more sharpened appearance at that size. Also, I sharpen images targeted for smaller-resolution output (including web display) than for bigger prints. That seems contradictory because the effects of the AA filter or the scanner fuzziness disappear during downsampling. But those effects were corrected during my correction stage anyway, and sharpening during targeting is what I do to achieve a certain effect. Smaller prints and web displays seem to me to need a bit more sharpening effect to give the same impression as a larger, more detailed print with less sharpening.

That said, given your output versus mine, any disagreement between my workflow and yours should cause me to reconsider my process more than you.

Rick "who also prints at 720 when the resolution is available" Denney

[Lenny Eiger]

I don't believe that printers have an optimum print resolution. I don't think that 360 is better than 370, or vice versa. I just send what I have at the size I need, usually way over 360, maybe 992 or 774 or 1127. There's no need to bring in another variable during a resampling action when there isn't a clear benefit?

I do believe in targeting, my RIP requires a flattened tiff file, so I always make one at the size I need and sharpen it as part of the process. (Since I use a drum scanner, the files start out sharp and do not require other kinds of sharpening that might be pre-targeting.) However, I make a point to never resample... certainly not for targeting.

Lenny

[Frank Petronio]

Truth is that I don't resize to print to my little Epson, I let it do the crunching. And over time, I sharpen less than I used to.

[rdenney]

I do believe in targeting, my RIP requires a flattened tiff file, so I always make one at the size I need and sharpen it as part of the process. (Since I use a drum scanner, the files start out sharp and do not require other kinds of sharpening that might be pre-targeting.) However, I make a point to never resample... certainly not for targeting.

If sharpening is important for a particular targeting process, though, resampling first is important, it seems to me. The sharpening algorithms are pixel-based and not scalable.

Rick "for whom sharpening is usually not that important for large prints from large-format scans" Denney

[Barry Kirsten]

Thanks, Rick, Lenny, Jim and Frank. I'm out of my depth on this topic, but am learning fast. Can someone explain the importance of the antialiasing filter and whether it should be routinely used for all scans?

Many thanks,

Barry.

[Lenny Eiger]

If sharpening is important for a particular targeting process, though, resampling first is important, it seems to me. The sharpening algorithms are pixel-based and not scalable.

Rick "for whom sharpening is usually not that important for large prints from large-format scans" Denney

Rick,

You simply resize without resampling. You can sharpen at the new size...
Lenny

[pdmoylan]

This is really fodder for another thread, but quelle difference in effect of sharpening with the scan vs post if any other than control? Also, since I don't use post sharpening in PS, I wondered about any evidence of artifacts. My experience with an old 35mm Polarioid film scanner if one was not careful at setting threshold and radius, artifacts became a serious problem. I expect Silverfast, Espson Scan and PS all moderate the effect. Any experience with this?

[Ivan J. Eberle]

An anti-aliasing filter is part of the optical path of most small format dSLR sensors but not medium format ones or scanners (and thus not typically part of the medium-format or large-format or scanner image-processing chain).