Tuesday, August 03, 2010

Quick And Dirty Comparison

For the full effect, click on the images and see them larger, but bottom line is the upper image was made with my 5D2, 50 mm. macro at f11, tripod mounted, live view (ie. no mirror bounce (in fact no shutter at the start of the exposure). The second is the result of literally tossing the 5X7 neg. onto my flatbed 4870 scanner, no holder, no glass to flatten it. It was made with my 210 Symmar S, f 32, 1/2 second exposure on TMAX-100.

At a guess, I'd say I'd need to use at least a double row stitch to get enough pixels to equal the resolution, possibly more. Note that one was taken in the morning, the other the night before, but that's a lot more information, I'd hazard a guess at significantly more than 100 megapixels worth.

Whether the ability to make 35X50 prints that can be viewed from 9 inches is worth the trouble of the 5X7 remains to be seen. I'm waiting on a backpack before I start seriously using the camera.

My plan so far is to use the lenses I have, 90 Nikkor, 135 Sinaron SE (which does in fact cover 5X7, barely), 210 Symmar S, 305 G-Claron and am getting a 450 mm. Fujinon-C.  If this actually becomes practical as well as fun, then I'll get a 120 Nikkor SW for architecture (for movements on the 5X7).

I'll continue to report in.


Paul Bradforth said...

George, I've never posted here before, and I haven't followed preceding posts that closely, but I have to ask: what did you do to the 5D upper shot to make it look like that? You don't say what sort of view it is, and it looks like about 400% pixels? I have a 5D, albeit a mk 1, and I can't think how I'd get a result that looked that bad. I don't mean to be rude, I'm just interested to know a bit more about it.

George Barr said...


that's about exactly right, this is some 241% to get the same as 100% on the film scan. I could do a better job upsizing the 5D2 file to show less pixelation, but not give it more detail.


Paul Bradforth said...

Ah, that clears it up! Thanks George; I somehow imagined that your 5D shots looked like that at 100%. What a relief :-)

Tim said...

Hmmm. So you've got three obvious considerations straight off the bat, then:

a) Is f/11 really the sharpest for your dSLR lens? (I would find f/32 for the LF lens less surprising.)

b) if the two shots are not taken at much the same time of day, then the different lighting would explain the drastic different tonality. Otherwise, the b&w conversion is different between the two processes (unsurprising, but something to consider getting to grips with).

c) I like the idea that the print should be the final arbiter. Maybe scale the scan *down* to match its size, then compare both pixel-peeping (at 100%, no more) and a print, to see which throws better detail through the printer driver at a given enlargement. Additionally, compare size of JPEG files from both sources, at the same pixel-count. (IME JPEG file-size is a reasonable ballpoint measure of the amount of image-data before your eyes; it behaves as expected with varying aperture across the same scene, for example, which is more than can be said for TIFF+LZW.) And of course there are issues of localised contrast/response versus viewing-distance.

I fully expect the LF to beat the digital effort for size/resolution/image-data from just one scan; purely for the sheer amusement of it, however, I'm currently playing with multiple scans offsetting the film-holder a few pixels apart and then re-blending them together, just to see what extra detail can be computed.