[jimmypage]

I'll be acquiring a Mamiya 6 later this summer, and "scanning" my negatives with my K-3ii seems like a vastly superior option over any dedicated scanner I can afford. However, macro lenses are suggested. I have two that technically qualify: a Sigma 28mm EG DG f1.7 and a Pentax-A 100mm f4 macro. I have a Ricoh GR II that would have been perfect but the electronics on it just bit the dust. I have a number of other good lenses in various focal ranges between 35 and 200mm. If there is a way to put my Takumar 35mm f3.5 to work for this it seems like a good performer in terms of edge to edge sharpness, but I wonder if I can effectively make do by framing the negatives such that they don't fill the whole sensor from top to bottom. 24mp seems like overkill. The only problem is that I'm already losing 20% of my resolution due to the fact that I'll be shooting square negatives on a sensor with 4x6 aspect ratio.

---

Last edited by jimmypage; 05-02-2021 at 09:40 PM.

[AstroDave]

Go with the longest focal-length macro that you have. The closer you are to the negatives, the more likely that field focal curvature might affect you.

[stevebrot]

I have two that technically qualify: a Sigma 28mm EG DG f1.7 and a Pentax-A 100mm f4 macro.

Those are two very different lenses when discussing reproduction. The task is to fit the image to the short axis of the K-3ii's sensor and you will not be needing much magnification. By calculation you will only need about 0.28x (1:3.6). I would use your 100mm macro, mostly to take advantage of its flat field. Focus distance will be about 600mm to the sensor plane with the 100mm, so suitable setup may be an issue.


Steve

[Kobayashi.K]

I have constructed several DSL scanners over the years (for 35mm film), each new version with improvements gathered by using them in practice. Here are some points to consider when choosing a lens.

1 - First, use a macro lens, and not any other standard lens, they are designed to have virtually no field curvature over the complete range, and they are sharp. Before use, test at which aperture value the lens has its maximum sharpness, usually around f/8.

2 - Select a macro lens with the shortest focal length possible, this has the following advantages:

2A - The distance between negative and lens is short, so your setup is compact and less susceptible to vibrations and mechanical instability, which causes unsharpness. A short lens with a resulting short distance between lens and negative does not mean more field curvature, the lens is working in the range it is designed for and curvature is corrected (example: DA 35mm macro limited).
2B - It is unlikely you see field curvature with a (short) macro lens, but anyway it can be corrected in post-processing easily.
2C - When you have a short setup the camera and the film holder can be mounted on a single rail. When the lens is long you are forced to make a setup with a tripod or a copy stand but they are extremely susceptible for vibrations (persons walking in the room, traffic passing by). You can't get maximum sharpness with a tripod or a copy stand.
2D - The area between negative and lens must be shielded from light from the environment and from the light source by which the negative is lighted to avoid flare, which reduces contrast. The best way to do is to mount a bellows between lens and negative holder, but you can also make a sliding box construction. Using a bellows or other means is difficult with a long lens and it complicates the construction.
2E - Having a compact setup with both the camera and the negative holder mounted on a rail allows you to construct a small unit on a base-plate which can be transported easily and stored in a case to avoid attracting dust.

[jumbleview]

I tried to scan with 35 Macro limited and 100 Macro limited. Scan with 35mm was OK but I run into problem which took me some time to solve.

Overexposed Left Edge on the My MX - PentaxForums.com

Film holder separator reflects some light to the lens. It happens for short lens, but not for long. Since then I scan with 100.

[Viking42]

The only problem is that I'm already losing 20% of my resolution due to the fact that I'll be shooting square negatives on a sensor with 4x6 aspect ratio.

If you want to get maximum resolution, you might consider shooting two slightly overlapping images of a single negative, then stitching the two shots in software such as Microsoft Image Composite Editor (ICE). It's a free download and I can attest that it works great for stitching neg scans - very easy to use and quick. I have done several such composites and have been very happy with the outcome.

As for lens, I agree with a dedicated macro lens due to their sharpness and field flatness. The latter is very important, as you don't want to be adding any distortion to the image, in particular if you use the above stitching method.

By the way, I'm a bit puzzled as to why you think 24mp is overkill? You will be buying a camera with some of the highest resolving lenses in the entire film medium format realm, so wouldn't you want the best possible resolution on the back end for your scans?

[jimmypage]

It's not that I think 24mp goes beyond the resolving power of the lens, but that if I have a pic that I think can benefit from DSLR scanning beyond 16mp, I will have it drum scanned. There is already no way for me to scan at 24mp due to difference in aspect ratio.

I figured with the 100mm lens I would need a tripod mounted rail but I am concerned about stability and haven't done enough macro work to know what to expect. Wouldn't it make sense to use pixel shift for such a project, and wouldn't pixel shift detect any movement that could affect sharpness? Or would the effect on pixel shift be undetectable at the level of movement we're talking about? I only ask because I have shot 100mm lenses on tripods using pixel shift without any detectable errors, although shutter speeds were in the 1/125-500 range

---

Last edited by jimmypage; 05-03-2021 at 08:18 AM.

[Not a Number]

You may want to check out this video (link below) for some ideas:

Youtu.be

Youtu.be

The scanning masks for holding the negatives are sold here:
DigitaLIZA 120 Scanning Mask · Lomography Shop

[Viking42]

It's not that I think 24mp goes beyond the resolving power of the lens, but that if I have a pic that I think can benefit from DSLR scanning beyond 16mp, I will have it drum scanned. There is already no way for me to scan at 24mp due to difference in aspect ratio.

Just to clarify, in case I muddled my explanation in my first post, I was proposing that you shoot slightly more than half your neg with the first shot (call it Frame 1, being the top half the neg, for example), then shift the neg and shoot the other half with a second shot (call it Frame 2, the bottom half), leaving a bit of overlap between the two images so that the stitching software has something to latch on to in both. Convert both to TIFF, import into Microsoft ICE (or equivalent), and stitch them together to make a single image with significantly higher resolution than a single shot. This way you can use the full width of your K3 sensor for both shots.

So if you do this two-shot and stitch method you will far exceed 24MP -- probably 36 to 40MP depending on how much overlap you have. And then proportionally more again if you use pixel shift for both of those. I think you may actually get as good or better results than a drum scan using this method, if done well. It is remarkable how good DSLR scanning can be.

Hope I explained it better this time, but if you're interested in more details just let me know.

[Kobayashi.K]

Here are some examples what resolution (dpi) you can achieve with various cameras.
To get a reasonable resolution for scanning 6x6 film with one single frame you definitively need pixel shift.
With a 16 or 24 Mp camera you must shoot several frames and use stitching, which complicates your setup enormously.
Personally I think 4000 dpi is the minimum resolution to reproduce the grain on your film.


Pentax K5-II - Scanning medium format 6x6.
===============
Film format 60x60 mm
1" = 25.4 mm
60 mm = 60/25.4 = 2.36"
Pentax K5-II sensor: 4928x3264 pixels (16 Mp)
Resolution: 3264/2.36 = 1380 dpi.


Pentax K-70 - Scanning medium format 6x6.
===============
Film format 60x60 mm
1" = 25.4 mm
60 mm = 60/25.4 = 2.36"
Pentax K-70 sensor: 6000x4000 pixels (24 Mp)
Resolution: 4000/2.36 = 1695 dpi.

The K-70 has pixel shift but the result is downsampled to the original size so no direct calculation is possible. It will probably be about 2000 dpi effectively.


Olympus E-M1X - Scanning medium format 6x6.
===============
Pixel shift options:
25 Mp (5760 x 4320), manual
50 Mp (8160 x 6120), manual
80 Mp (10400x7792), tripod

Hi Res manual 50 Mp, 8160x6120 px (3:2 or 4:3).
Film format 60x60 mm
1" = 25.4 mm
60mm = 2.36"
Resolution: 6120/2.36 = 2590 dpi

Hi Res tripod mode 80 Mp, 10400x7792 px (3:2 or 4:3).
Film format 60x60 mm
1" = 25.4 mm
60mm = 2.36"
Resolution: 7792/2.36 = 3300 dpi

p.s. The Panasonic G9 mentioned in the video (post #8) is also an M4/3 camera, with a high res pixel shift mode 10368x7776 (80 Mp) like the Olympus EM1X, so a scan of a MF 6x6 negative gives you about 3300 dpi resolution.

[jumbleview]

Here are some examples what resolution (dpi) you can achieve with various cameras.
...

All calculations does make a sense under assumption that lens so good it out-resolves sensor.

[Kobayashi.K]

All calculations does make a sense under assumption that lens so good it out-resolves sensor.

That's why you need a very good macro lens and care for the preconditions mentioned in my post #4. And there are many more things that can go terribly wrong.

The gentleman in the video (post #8) did a good job for assembling a scanner with standard parts. But this scanner has also two major flaws: it is not stable mechanically, with the camera sticking out on a thin arm, and the area between lens and film is not shielded, so the light of the LED panel and the lighting in the environment are causing flare - it is the most heard complaint of DSLR scanner users that the result has no contrast.

Even building a DSLR scanner for 35mm film is not a trivial thing, and building one for medium format is very challenging.

[jimmypage]

All calculations does make a sense under assumption that lens so good it out-resolves sensor.

I want to give my sincere gratitude to those who have responded so far. Great info. But my most immediate concern is whether I need to look for another lens.

I will not have the Mamiya until July. I have plenty of time to plan this out and assemble a macro studio so that by the time I have the camera I will be prepared to quickly digitize the results. What I am trying to figure out is whether I need to budget for a different lens, because whatever studio I build will be built around that lens. It sounds to me like people have had success with the Pentax-A 100m f4, but if shorter focal length is preferable then I would want to know what is the best value lens at the shortest acceptable focal length. I would like my macro studio to be as small as possible, so if I can get away with 28mm that would be my preference. My Sigma will do 0.34x magnification at 20cm, but it's not a sharp lens. I'll pay a lot for a lens if I have to, but only if I'm getting value for money. I'm concerned about flatfield focus on any DSLR and I don't know how close to "perfect" I can get.

I feel like my Ricoh GR II had it, and it would be been fun to build a macro studio around that little lens and see what it could do. If there were a cheap point and shoot with a killer macro lens that that shot 14 bit RAW I would give it a go. I'm sure I can get a lens that will suit me for < cost of a new GR III (don't plan on shooting at f8 on a used one because the sensor is a dust magnet).

[stevebrot]

2B - It is unlikely you see field curvature with a (short) macro lens, but anyway it can be corrected in post-processing easily.

How does that work?

I can easily correct distortion, but field curvature?


Steve

[stevebrot]

My Sigma will do 0.34x magnification at 20cm, but it's not a sharp lens.

There is your answer. A 50mm macro lens should give you about 300mm distance subject to sensor.


Steve