A quick and dirty comparison of scanning a roll of Delta 100 film with a DSLR (a Canon 5D Mark III using the Canon 100mm macro lens) and en Epson v850 flatbed scanner.

The Delta 100 was chosen as it is a very fine grained and sharp film. It was processed in Ilfosol 3 1+4 (unfortunately at a very warm 26 degrees, which meant only a 4:30 developing time).

Three scans were made and normalised to 4800 DPI by upscaling in Photoshop using Bicubic Smoother:

• Canon: single image at f14, ISO 100, using an iPad as a backlight for the negative and a Canon 100mm macro lens. 30Mbyte CR2 RAW file (22MP).
• Epson #1: 3200 DPI image from Epson Scan. 32Mbyte TIFF (16MP).
• Epson #2: 4800 DPI image from Epson Scan. 72.5Mbyte TIFF (37MP).

The Canon image was made by suspending the film in an old negative carrier above an iPad showing a white background. The system was collimated using a mirror and the resulting CR2 file processed in Lightroom via a custom “curves” profile to invert the negative.

The images were then edited for dust and the had the standard clarity and sharpening that I apply to all film scans (clarity +10, sharpening 40 with radius 2.0, detail 20 and masking 30).

The first comparison show the upscaled Canon image on the left, and the Epson 4800 DPI image on the right (click to enlarge):

Comparison Canon macro photograph of a negative (left) and an Epson v850 scan 4800 dpi scan (right).

The difference is very small, but the Canon image perhaps has a slightly better sharpness – despite being upscaled from 22 to more than 37 megapixels. The Canon image is also slightly darker, reflecting a small difference in exposure.

The second image shows the difference between the two epson scans. I think that it is pretty much impossible to tell them apart:

Comparison Epson v850 scans at 3200dpi scan (left) and 4800 dpi scan (right).

Although not shown here, I also tried an experiment with +100 shadow lift on the scans. Here the Epson images perform much better than the Canon, probably reflecting the relatively widely mourned dynamic range from Canon sensors. It would not, however, be very difficult to shoot two images and perform an HDR merge if this was an issue.

To summarise:

•  even a sharp, low grain, low-ISO film is comfortably out-resolved by a DSLR sensor
• a 3200 DPI Flatbed scan gets pretty close to a DSLR macro image (and is much more convenient if scanning many negatives as a batch)
• a 4800 DPI Flatbed scan is just a terrible waste of disk space

So nothing particularly surprising. It is also worth adding that the DSLR images had vastly less dust and fluff on them than the Epson scans. This is because of the design of the Epson film strip holders, which use a clear plastic to support the film that is very prone to static electricity and dust collection.

At a rough guess, the Delta 100 is probably yielding about 16MP of actual useful resolution (vs around 8-10MP for HP5+). It might be interesting to run a comparison of Delta 100 against a u4/3 digital image at some point…