Advice sought on Epson V 750 vs Capturing image with my 5D and lightbox

[chuck94022]

I've heard a lot of a priori arguments as to why dslr scanning will be worse than other methods. So far, though, the results have been very encouraging. Using something like a three frame stitch of a 4x5 negative should be easy to shoot and combine. As long as that gets you the resolution you need for the size prints that you want to make, it should be a fast and high quality method. When you start to get more frames, resolvable detail does go up, and it's still pretty fast to shoot, but you'll start spending a fair amount of time in stitching software.

I the the fundamental argument against it is that a DSLR does not measure every pixel position with all color channels. It must interpolate. At your chosen output resolution that may be just fine. For web use it will certainly be more than fine. (4x5 is wasted on the web, there, an iPhone is more than enough...)

But if we are talking ultimate resolution, a Bayer interpolated sensor is a compromise.

[rdenney]

I the the fundamental argument against it is that a DSLR does not measure every pixel position with all color channels. It must interpolate. At your chosen output resolution that may be just fine. For web use it will certainly be more than fine. (4x5 is wasted on the web, there, an iPhone is more than enough...)

But if we are talking ultimate resolution, a Bayer interpolated sensor is a compromise.

Well, there really is no such thing as "ultimate" resolution, until the scanner can sharply delineate each grain. That is beyond any current scanner.

It is tempting to focus on one single aspect of any given system, such as the Bayer array, or the anti-aliasing filter, or the dynamic range, or the mechanical imprecision, or the plastic molded lenses in the strip sensor, or the reflections of the platen glass, or the flare caused by the diffusion light source, and on and on.

Really, though, whether something works or not is a system question as much as a technology question. When applying any technology to a particular use case, some aspects of it will be overdesigned for that application, and others will be marginal or inadequate.

Considering the system as a whole, it is possible using a 5D to get better resolution than using an Epson V750. I have proved that. But it is not easy! And it requires working at 1:1 or even 2:1 and then stitching, with excellent glass. The challenge is not to find a camera with sufficient resolution, Bayer array or AA filter notwithstanding. The challenge is an even light source (or even a light source corrected for the lens's falloff), and a stitching capability that is reasonably efficient, particularly for images with large expanses of smooth tones.

And with negative materials, my 5D had more than enough dynamic range to capture the entire range of densities in a black-and-white negative that I processed for optical printing (read: It was on the dense end of what scanners prefer). I'm quite sure it would be more than sufficient for color negatives, too. Color slides should present the greatest challenge there. I have not yet had time to play with my new-to-me Bowens Illumitran, but in the old days I could control contrast reasonably well duping regular color slides onto Ektachrome (using the contrast unit), and even my 5D has several stops more dynamic range than old 35mm Ektachrome.

Making a single photo of a 4x5 film does indeed beg the question of why the digital camera wasn't used in the first place. One reason is that the intended application does not need more resolution, and another is that the films were made before using a digital camera was an option. But I still use 6x7 and 4x5 because it gives me more detail than I can get from my original 5D, and I don't make enough images for film-based costs to exceed what a better digital camera would cost, even if I had the money.

Several years ago, my wife gave me the task of duplicating a photo album that belonged to her father. I started removing the photos one at a time and scanning them in a flatbed. I was spending about a half hour on each picture. The scanner is designed to record all that can be recorded more than it is designed to make a result that looks like a photo. When that little fact occurred to me, I set up my copy stand and made the photos with my Canon 10D and the 50mm macro lens. Since the camera is already configured to make images that look like photographs instead of images that look like scans, I didn't have to make any corrections at all. I finished the album averaging about 60 seconds for each image. The scanner still provided better resolution, but the camera was still good enough to print those dupes out at their original size and be nearly indistinguishable (from an image standpoint) from the originals. In that use case, the scanner was inefficient overkill.

I'm not sure I understood the OP's point clearly enough to know whether he needed something quick and easy where it was not necessary to extract as much detail from the negative as possible, or whether he really believed a 5D dupe of a 4x5 film on a light table was sufficient to get all the detail that is available in a 4x5 film. Generally, a scanner does better at extracting detail because that's what it was designed to do, and a dupe does better at getting the image onto the web (or into a photo album at snapshot size) because that's what it was designed to do. It's possible, as we have proved, to use the digital camera as a scanner, but it is not easy. The purpose of that effort has not been to make it easy, but to have an alternative when scanners are no longer available.

Rick "who really needs a few hours with that Illumitran" Denney

[Peter J. De Smidt]

Here a dslr scan of a 6x7 Acros negative:


Here's a detail from the image:


That looks like pretty good resolution to me.

Here's a detail from a Screen Cezanne scan at 4000 spi:


Here's a detail from a scan using a dslr and a 40 year old 55mm nikkor lens:

[Serge S]

Can someone recommend a good quality lightbox for digital capturing of transparent originals?