Scanner Test

Sprintscan vs Coolscan

I wish I could have had access to comparative scans of astronomical negatives at the time that I was making a purchase decision. Commercial scanners are obviously optimised for the mainstream market,and their performance when presented with thin contrasty astronomical negatives has the potential to yield dramatically different results. Recently, through a very co-operative dealer, I was lucky enough to obtain scans of the same negatives on both the Coolscan and Sprintscan. I thought I would share this information for the benefit of others. I have also had 21 of my negatives scanned by PhotoCD, so I have included notes on these results. The comparative images are based only on the Coolscan and the Sprintscan, since I did not scan any of the PhotoCD negatives during the brief time that I had the Coolscan.

• Kodak PhotoCD - the highest quality scans of all, with excellent definition, extremely good tolerance to negative defects (scratches, etc) and no scanner induced artefacts visible at all. However the scans were not optimised for astronomical images and in the particular case of the UK PhotoCD centre (Ealing), they caused significant damage to my negatives. There is evidence that they use cheap labour as scanner operators who are trained only to use the scanners and have no idea how to handle valuable photographic material. I believe that they damage negatives in this way on a regular basis. In a further update to this page I will include an image of the damaged negatives.
• Polaroid Sprintscan 35 Plus - very high quality scans of both regular and astronomical negatives, and a close second to the quality obtainable from PhotoCD. Excellent definition, very good tolerance to negative defects (scratches, etc) and a very low level of scanner induced artefacts. A very comprehensive range of adjustments and controls for setting optimum scanning characteristics. For regular negatives, a little more work is required to obtain an optimum scan than with the Nikon (which is more automated), but for astronomical negatives the comprehensive adjustments enable an optimum scan to be obtained. Scanning artefacts (such as banding) can become visible if the scanner is allowed to use entirely its default settings on a thin astronomical negative and the resulting scan is aggressively enhanced in Photoshop. The key with the Sprintscan is to optimise all settings as far as possible at the pre-scan stage. All images on the image page were scanned with the Sprintscan and in my very limited experience it is the best $2,000 scanner for astronomical imaging.
• Nikon LS-1000 Super Coolscan - very high quality scans of regular negatives. Excellent definition, but very poor tolerance to negative defects (scratches, etc) and a relatively high level of scanner induced artefacts on thin astronomical negatives. The range of adjustments is not so comprehensive as with the Sprintscan - the big omission as far as astronomical negatives are concerned is the lack of exposure control. For regular negatives, the Coolscan produces a wonderfully exposed and colour balanced image entirely automatically, but with thin astronomical negatives the same level of automation results in a disasterous scan. Some scanner induced artefacts (such as banding) were highly visible (even in the pre-scan) on the very thinnest astronomical negatives. Based on my very limited experience I could not recommend the Coolscan for astronomical imaging.
  Update: Since writing the above I have seen feedback from Tom Krajci and Jerry Lodriguss which indicates that it is possible to obtain good scans of astro negatives by using a technique that essentially 'tricks' the LS-1000 into making the correct exposure and contrast adjustments. Jerry Lodrigussreported the following:
  "Just tried a new trick that worked very well for scanning those hard to scan astrophoto negs with a small bright object against a mostly blank sky background. I took a piece from the beginning of the roll, where the leader was fogged totally black, and I mounted it in a slide mount so that the film in the slide mount showed HALF totally black, exposed film, and HALF showed totally clear, orange film base. Then I prescanned this neg with the scanner's normal defaults as a color neg. Then switch to the real astro neg, and DISABLE the prescan before acquiring the final image. I used it for M51 and got a very nice scan without any prescan manipulations."
  The rest of this report is not modified because it was based on using the LS-1000's normal controls and therefore remains valid.

I compared scans of a couple of difficult negatives - this was purposeful, because sometimes in astronomical imaging, the very best we can get is a very thin negative. In all cases I tried as best I could to optimise the scan using the scanner's controls. None of these images was touched in any way by Photoshop, other than to resize the image - they are pure output from the scanners.

The following scans are from a thin test negative of M51 - the exposure was only 50 minutes - cut short by the rising moon. This is not the same negative as shown on my images page. The negative had some light scratches (put there by the French photo lab).
 

Nikon LS-1000 Super Coolscan	Polaroid Sprintscan 35 Plus

It is difficult to believe that these scans are from the same negative. The light scratches are not only made fully visible, but seem to be actually amplified by the Coolscan. Notice the M51 'spokes' - at 6 o'clock and 7 o'clock positions, which are caused by very light scratches.The Coolscan image is already way overcooked and there is nothing that could be done in Photoshop to retrieve it. Believe it or not, I tried very hard (including an hour long consultation with Nikon tech support) to optimise this scan. Although I tried hard with the contrast, curves, and brightness, they made very little difference - the big killer was lack of any control over exposure (apart from kludging it). Notice also the banding visible on the Coolscan image - this appeared on most scans of thin negatives. The Sprintscan image is suitable for further enhancement (and colour balance) in Photoshop and a fairly decent image could be made out of it, although because of the thin negative it will never yield an optimum image.

Nikon LS-1000 Super Coolscan	Polaroid Sprintscan 35 Plus

In a further attempt to get optimised scans I used the scanners controls to select and scan only a portion in the centre of each negative. In each case I used the scanners controls as much as possible to get the very best out of the image, as though I had no recourse to further imaging enhancement software. In the case of the Coolscan image, the scanner automatically set the contrast to a very high level and the exposure to a very low level and there was no means using the scanner's controls to correct these values any more than is shown. In the case of the Sprintscan image, I purposely overdid the contrast to try to yield an optimum image, with the result that the core of the galaxy is blocked up. With the contrast at a more reasonable level, the image shows excellent gradation right into the core.

The following scans were from another very thin negative of the Antares - Rho Ophiuchus region - one of the two negatives used in the image on my Images page. It too had some light scratches (put there by the French photo lab).

Nikon LS-1000 Super Coolscan	Polaroid Sprintscan 35 Plus

Very similar comments apply as above. The banding in the Coolscan image is even more visible. Again, the Coolscan image is so overcooked that nothing in Photoshop will redeem it. Was this an unfair test? I don't think so - theSprintscan image is fully useable, and this was one of a pair of scans that formed the digitally double-stacked image of the Antares region on my Images page.

Was there something wrong with the Coolscan? I really don't think so (and Nikon tech support didn't think so either) - it automatically produced beautiful scans of any regular negative material. Here is a small section of a negative (taken with a cheap snapshot camera) and scanned with the Coolscan - exposure was perfect and no scanning defects were visible at all. This image was not touched by Photoshop other than to crop and resize it - it is straight output from the Coolscan using entirely automatic settings.