I recently acquired a Minolta 5400 DiMage f4 40mm Scanner Lens and wanted to see if it would be better to use than my current scanning lens, the Sigma 105mm f2·8 DG DN Macro Art.
Following my previous comparison above and with the knowledge that focusing could be an issue I decided to experiment with a measured depth of field comparison.
I drew up a test patten of dashed lines the length of a 35mm film frame diagonal with centre, 24mmØ and 36mmØ marks using Adobe Illustrator. The test pattern is mounted onto a stand made of mount board angled 22·5 degrees from the lens axis. Using “Right Triangle Calculator” a ·25mm line spacing at 22·5º needed to be increased 108·293% to represent ·25mm vertical line spacing, the overall pattern being 8mm high vertically. I printed the pattern with my mono laser printer which has a pitch of 1200dpi so there is some moiré effect, nevertheless, there is sufficient edge between black toner imprint and paper for checking focus.
For each lens at both full frame and APS-C I used a 24x36mm rectangle on the platform to set the magnification, then raised the camera/lens to achieve focus on the centre of the centre line on the angled pattern. I used indirect daylight so there is colour temperature change during the day and between cameras: Sony A7C and Sony A6400.
When I bought the Sigma I tested it for slide/neg copying at all apertures and found it best between f5·6 and f8, I use f8. It has a 10mm extension fitted so focusing at 1:1 doesn’t max-out. The Minolta is fixed at f4.
Four lines sharp corner to corner so ·75mm depth of field then gradual loss of focus to 1·5mm depth of field.
Bonus: this also tests field curvature and chromatic aberration.
At least five lines sharp corner to corner so 1mm depth of field then gradual loss of focus to 2mm depth of field.
Three lines sharp corner to corner so ·5mm depth of field, very rapid loss of focus beyond four lines though. I used the Minolta lens reversed for APS-C as the image quality dropped off beyond 24mmØ in normal orientation. This lens is designed for magnification at 1·8:1 and performs best up to 4:1 (according to Daniel Knop). So it makes sense to me reversing it for 1:1·5 which is a reduction of the subject size.
Not quite three lines sharp in centre. Although this lens covers the subject at 1:1, it’s not its comfort zone, sharpness deteriorates beyond 24mmØ. I tried both lens orientations.
Using the same Kodachrome slide as before, this time the 631x631 pixel crops near the frame edge have been resized to 1262x1262 for easier assessment. Each lens was very carefully focused on this target crop with the aid of Sony Imaging Edge Desktop Remote and my Velmex Unislide.
Conclusion
The winner here for digitising 35mm photos is the lesser magnification of 1:1·5 APS-C. Both the lenses are neck-and-neck, the Sigma maybe showing slightly sharper grain, however, the Sigma maintains that sharpness easily across the entire image with its greater depth of field. It encompasses film curvature and non-parallelism of sensor and subject, especially when scanning multiple photos expeditely. Whereas the Minolta requires a very precise and finely adjustable setup with flat film to achieve sharp focus all-over.
I did try the Minolta lens at its intended magnification of 1·8:1 on this photo, it was as good but didn’t resolve any more information/sharpness from this image. Is then 24mp enough for 35mm photos, mostly hand-held? It does provide higher resolution than 4K TVs and projectors.
For medium and large format negs then obviously full frame or larger sensors with their higher available resolutions working at less magnifications than 1:1 will be more suitable.
Finally, do any lenses out there perform as good or better on full frame at 1:1 than APS-C at 1:1·5 on 35mm photos with the same sensor resolution? I do have a Rodenstock Apo Ronar 150mm f9 in my 4x5 kit???
