Scanning 35mm film – is high-res scanning worth its cost?

Most photo labs propose scans in 3 resolutions: 1000×1500, 2000×3000, 4500×6700. The scans  are saved as jPEGs, with some labs also offering to save 4500×6700 scans as TIFF files.

In theory, those resolutions correspond to an image of 1.5 Million points (1.5 MP), 6 MP, and 30 MP respectively. In general,

  • 1000×1500 scans – when available – are virtually free (they’re included in the processing costs by some labs such as thedarkroom.com )
  • 2000×3000 scans cost roughly $5 for a full roll (in addition to the processing costs), or .50 per individual image scanned
  • 4400×6700 scans cost roughly $11 to $12 per full roll (in addition to the processing costs), or 3.00 per individual scan
  • 4400×6700 (TIFF) are the most expensive at $21 per full roll (oldschoolphotolab.com)

Storage constitutes an indirect cost – which doesn’t hurt until you run out of disk space, and have to upgrade your PC, your home NAS  or you online backup plan. But if storing 36 images at 1.5 Mbytes will not break your storage budget, 36 high res TIFF images represent almost 3 Gbytes. The exact size of a JPEG file is difficult to predict (JPEG is a lossy compression format), but in general, the file size of each type of scan falls within those brackets:

  • 1000x 1500 – JPEG -1.5 to 2 Mbytes
  • 2000 x 3000 – JPEG – 3 to 4 Mbytes
  • 4492 x 6776 -JPEG – 12 to 16 Mbytes
  • 4492 x 6776 (TIFF) – 80 MBytes / image
Scan_2000x3000_Piedmont
Atlanta Piedmont Park – Shot with Canon A-1 – Canon FD 35-105 f/3.5 – Fujicolor 400. Scanned at a resolution of  2000 x 3000 – the pictures of this roll are not really better than when scanned at 1000 x 1500 – probably a limitation of the lens (a 35-105 zoom of the seventies)

The tests

I wanted to have a few pictures I had taken a long time ago scanned, and I asked the lab to scan some images in 2000 x 3000, and some in 4400 x 6700. The pictures had been taken with a Minolta 7xi and the famous Angenieux 28-70 f/2.6 zoom, on Fuji Reala film (the 100 ISO “professional” color film Fujifilm were selling at that time). The pictures had originally been enlarged on photographic paper, and I expected the scans to be good.

I also had a series of images taken recently with a zoom from the early seventies, that had been scanned by the lab at 1000×1500, that I asked the lab to rescan at 2000 x 3000.

Once the jPEGs were ready, I downloaded them in iPhone and iPad photo galleries, in Photoshop and Lightroom on a laptop, and on WordPress, in order to compare the perceived quality. A reminder of the resolution of a few devices compared to print.

  • iPhone 5 S Retina photo gallery : 1136x 640 (720,000 points) at 326ppi
  • 9.7 in iPad Retina Photo gallery:  2048 x 1536 (3,000,000 points)  at 266 ppi
  • Print 8 x 10: 2400 x 3000 points or 7.2 million points at 300ppi
  • the pictures of this blog are generally saved for the “Large” format proposed by WordPress, at 1024 x 680, corresponding to 600,000 points.
Paris – The Seine – scanned at 2000 x 3000. Minolta 7xi – Angenieux Zoom 28-70 – Fuji Reala film (1992). No visible difference in quality with the 4492 x 6700 scan (look at the details of the Eiffel tower compared to the glass house of the Grand Palais in the image below)

Conclusion

  • Scan at 1000×1500 or 2000×3000 ?
    • on an iPhone, on a 4×6 print, or in a blog supporting 1024 x 680 images (such as this one), there is no visible difference between 1500 x 1000 and 3000 x 2000 scans.
    • For all larger screen or print formats (9.7′ iPad Retina, laptop, 8×11 print, blogs offering to view images at native resolution)  the difference between a scan at 1.5 Million points and a scan at 7 Million points is very visible, unless the original is very poor (low lens resolution, very grainy film, subject slightly out of focus, operator shake at slow shutter speeds). It’s even more visible if you crop the image, even slightly.
  • Scan at 2000×3000 or 4400×6700 ?
    • on an iPhone, iPad 9.7′ Retina or on a 8×11 print – the difference is not really visible.
    • Above that (13 x 20 prints, for instance), the theoretical difference in resolution does not  necessarily translate into a difference in print quality: a 13 x 20 print  represents 24 million points at 300 ppi and the 6 million of points of a 2000×3000 scan should theoretically be overwhelmed, but practically the resolution of the film and of the lens play their part, as the technical limitations of the photographer (focus, shake) do. Large prints are often framed and hung on a wall, and you don’t look at a picture on a wall the same way you look at a 8 x 10 print you hold in your hand. And all technical considerations taken apart, with some subjects, images scanned at 2000×3000 may look as good as images taken at 4492×6770 – it depends on the contrast and quantity of fine details in the subject.

Scanning at 2000×3000 is a good compromise for 35mm film, and my choice when I have film processed. It works fine with any support I use day to day (iDevice, laptop, 8 x 11 prints), is not too expensive and generally produces a visible difference with the 1000×1500 scans.

If I wanted to print a really great picture, an image compelling from an artistic point of view and almost perfect technically (fine grain film, sharp lens, subject in focus, no shake), I would have it scanned at the 4492 x6776 resolution, and saved as TIFF. It would give me no guarantee that the print would be great (there are so many variables), but it would give me the best chances of success.


Scan_4492x6770_Paris-22
Paris – Scan 4492 x 6770 – Shot from the Pont Neuf -Minolta 7xi – Angenieux zoom 28-70 F/2.6 – Fuji Reala (July 1992)