Scitex S-3 scanner lens tests

[Lou Jost]

I have been playing with the scanner lenses which came from several high-end Scitex 342L scanners. These scanners cost > $50000 when new, so we might expect the lenses to be good. Each scanner comes with three specialized lenses: a 110mm f/5.0 lens to scan 35mm film, an 89mm f/5.0 lens to scan 6 cm wide film, and a 67mm f/4.9 lens to scan 4 inch x 5 inch film. The scanners also have other lenses which I have not tested. I intend to keep one set of lenses and sell the other two sets.

Edit: There is more info on Scitex scanners by Robert O'Toole here:
http://www.photomacrography.net/forum/v ... ght=scitex

These lenses are presumably optimized for the film sizes they are meant to scan. I'll post some results from them near their designed magnification, and reversed at the reciprocal of the designed magnification.

The lenses have rather small apertures, so I would not expect them to be particularly good on very small sensors. So I test them on a FF sensor (Pentax K-1 with Nikon mount instead of Pentax mount). These lenses have 39mm threads on one side and no threads on the other. They can be mounted in either direction using stacks of step-up and step-down rings.

Today I'll start with the 110mm lens. These are DMAP stacks of jpgs with no post-stack processing unless noted.

First, m=0.7, full image:


And a 100% crop near edge:


Now m=1.4, full image:


100% crop near edge:


Same crop, slightly sharpened:

[Lou Jost]

Now the 89mm lens, FOV 6 cm, full image:


and 100% crop near edge:

[Lou Jost]

Here's the 67mm lens, FOV 4 inches, full image:


100% crop:

[ray_parkhurst]

Very nice, Lou.

In my past testing I've missed some level of CA due to testing on Cu coins. I still catch the CA present in specular highlights, and have generally judged lenses on this criteria. Recently I used an Ag coin (essentially equal RGB) and found that my S2 and S3 lenses show a fair amount of LongCA. I intend to re-check them all but thought I'd mention this in case you have time to verify yours.

[Lou Jost]

Ray, I've found that it is important to use them at their designed magnification and orientation. Near the designed values, the lenses are very clean, no LoCA. There are some faint reddish haloes around bright white highlights against black backgrounds, especially on the 67mm lens when used far outside its design range. I will add those results. But generally I am impressed with the corrections of the 110mm and 89mm lenses when used as designed.

[ray_parkhurst]

Ray, I've found that it is important to use them at their designed magnification and orientation. Near the designed values, the lenses are very clean, no LoCA. There are some faint reddish haloes around bright white highlights against black backgrounds, especially on the 67mm lens when used far outside its design range. I will add those results. But generally I am impressed with the corrections of the 110mm and 89mm lenses when used as designed.

OK, so to summarize:

110mm
Forward mounted for m=0.7
Reverse mounted for m=1.4

89mm
Forward mounted for m=0.6 (36/60)
Reverse mounted for m=1.7 (results not shown)

67mm
Forward mounted for m=0.35 (36/102)
Reverse mounted for m=2.8 (results not shown)

Is this correct?

[Lou Jost]

Here's a severe test of the 89mm at 1.4x where it is very good (note this is only a very partial stack centered near the bottom):

Full image:


100% crop:


On the same subject the 67mm at 4x (reversed) shows those slight reddish haloes around the brightest scales on black backgrounds:

Here they are, first the whole image:


Here a 100% crop:

[Lou Jost]

Ray, the best info I have found on the designed magnifications of these lenses is this:

Scitex S-3 4.9/67 for scanning 4x5". Optical magnification 0.215 (4.65 reversed).

Scitex S-3 5.0/89 for scanning 6x6cm, 6x7cm, 6x9cm. Optical magnification 0.376 ( 2.65x reversed).

Scitex S-3 5.0/110 for scanning 35mm. Optical magnification 0.726 (1.4x reversed).

The 67mm and 89mm lenses should be oriented with their threads facing the sensor when m <1.

The 110mm does best when the threads face the subject for m<1.

[Lou Jost]

I also compared these to some other finite lenses. I found that at 2x, testing the 67mm and 89mm against the DiMage 5400, the Dimage was better.

At 4x, among the 67mm, the DiMage, and the Lomo 3.7x, the Lomo was best.

[ray_parkhurst]

I also compared these to some other finite lenses. I found that at 2x, testing the 67mm and 89mm against the DiMage 5400, the Dimage was better.

At 4x, among the 67mm, the DiMage, and the Lomo 3.7x, the Lomo was best.

I must agree that the S2 and S3 lenses, while decent (and ranking high in my shootouts), are not on my go-to list. One advantage they have over the DiMage and Lomo 3.7x is in working distance, allowing more flexible lighting. Plus of course you can use them forward orientation at much lower mags.

One Scitex lens that I'm fond of is the 108LFOV, but it is also no longer on the go-to list, having been supplanted by the 85MV. Before the 85MV the 108LFOV was my favorite lens for full-coin shots.

Edited to add:

Here's a quick calc of the Feff at their optimum mags forward and reverse:

67/4.9
Forward mag=0.21, Feff=5.9
Reverse mag=4.7, Feff=28

89/5.0
Forward mag=0.38, Feff=6.9
Reverse mag=2.7, Feff=18.5

110/5
Forward mag=0.7, Feff=8.5
Reverse mag=1.4, Feff=12

Thinking about using it on a sensor with ~4.3um pixels, the 67 looks OK at very low mags, but since its performance falls off away from the optima then its range is very limited. In reverse the small aperture makes it a poor choice at the conjugate magnification. The 89 looks OK at its low mag optimum, and not terrible at the conjugate. Would be interesting to see how it does at 2x. The 110 is OK at low mag optimum and still decent in reverse at high mag optimum.

Further edit: I included the 67/4.9 in my shootout at 2.4x, and shot it initially in forward, then in reverse later. At that time I had no info on the optimum ranges of the lens. I was also not looking much at CAs, just sharpness. It did OK reversed, but was not a strong performer. Had I known the 89/5 was optimized around this mag in reverse, I would have included it instead.

[Lou Jost]

Ray, while these Scitex lenses are not as good as the Makro Varon or Printing Nikkors, they are also not nearly as expensive. The 110mm is very clean when used as directed, essentially apochromatic. And I do like their working distance relative to the Dimage or Lomo. A set of them would be a convenient way to get a wide range of magnifications with apo or near-apo quality.

That said, anyone who has a Makro Varon or PN certainly doesn't need these.

[Lou Jost]

Here is the 89mm reversed, at 2.65x (FOV 13.2mm), the reciprocal of its designed magnification:

Full image:


100% crop:


Fairly clean chromatically.

[ray_parkhurst]

Ray, while these Scitex lenses are not as good as the Makro Varon or Printing Nikkors, they are also not nearly as expensive. The 110mm is very clean when used as directed, essentially apochromatic. And I do like their working distance relative to the Dimage or Lomo. A set of them would be a convenient way to get a wide range of magnifications with apo or near-apo quality.

That said, anyone who has a Makro Varon or PN certainly doesn't need these.

Lou...like you, I own quite a number of these lenses, but this thread has shown me I've been using them incorrectly. My application interest is always coin photography, where mag range is from ~0.35 up to ~1 on APS-C. However, I've had many requests for shooting full encapsulated ("slabbed") coins, which necessitates >80mm FOV (m=0.25). Sometimes folks want to shoot large medals, whole mint sets, and proof sets, which can be fairly large. I'm now realizing that the 67/4.9 may be a good lens for these larger items. The big question will be how well it works for Dollars without CAs. Some folks only collect Dollars and larger items.

I should also test the 89mm across the full coin range to see if it can be used seamlessly. I hesitate recommending lenses that need to be reversed yet don't have a good facility for doing this, so I would hesitate with the 110mm unless I can figure out a good way to reverse it. This also limits usefulness of the 89mm for reversed use at higher mags, though I'd have a hard time justifying the long extensions required in any case.

[Lou Jost]

Ray, the 67mm is the least impressive of the three lenses, in my experience.

These lenses are easy to mount in any direction by making a "cocoon" of step-up and step-down rings for them. I use this technique with many lenses that have no threads. With careful choices of rings it is perfect. These Scitex lenses are nicely held by 49mm rings. All the Scitex lenses are exactly the same size so one cocoon fits all. By using a 55-52 step-down ring as a female-female coupler, you can have 52mm threads on both ends of the cocoon for easy reversing.

[Lou Jost]

The 110mm Scitex S-3 lens is particularly good. It is apochromatic at the designed magnifications and very sharp all the way out to the corners. I used this lens in the photos I have posted here:
http://www.photomacrography.net/forum/v ... 337#237337