Constructing the definitive DSLR scanning setup. Communal effort!

[DrTebi]

For what it's worth:

http://www.mfphotography.ca/michael...-guide-to-scanning-film-with-a-digital-camera

That link doesn't seem to go where you intended it to go...

 

[DrTebi]

It's gotten a bit quiet around here...

I have sinned and recently bought a non-rangefinder camera, a Kowa Six MM. Did a first test roll, with Ilford FP4 Plus and scanned all images with my DSLR setup (Pentax K-1 with Pentax 50mm DA Macro).

I have done all post processing in Darktable, and finally in GIMP. Images are resized to fit a 4K screen. Note that I have adjusted for the vignetting of the Pentax lens for every shot with a "Flatfield" shot of the light source... it does make a difference.

Please take a look, tell me what you think!
https://www.flickr.com/photos/drtebi/albums/72157698823837781


Cheers,
DrTebi

 

[Jockos]

@Jockos: Have you already started to build the film holder?
There is one thing that I just remembered, which I wanted to do different on mine.

On the top plates, the ones that mount the film, I wanted to add an edge, or recess, so that one could easily fit masks into it. It would have to be some 5mm wide or so, since the distance between the two holders is not always the same. Then just make masks that are slightly larger than the smallest width you would cover (e.g., for 120 film ca. 64mm wide).

I find it, from some tests I have done, not so important to have masks for 35mm film, since you sensor covers exactly the negative.
But when you take a single shot of a 6x7 for example, your sensor will cover some areas where no film is. This will throw off your exposure. Here a mask will make sure that the exposure reading is limited to the negative/positive itself, not the light that comes through the sides...

I hope that all makes sense...

If I had a mill, I would have put an edge into my pieces already... I don't really feel to comfortable trying it on my drill press...

Hi!
I've been away for a while and did not see this. The materials are at the shop, but it'll take a while before it's done I'm afraid..
Unfortunately I was not able to get a go at the router, so they will only do the holes. It sounds like a good idea to make slots for masks though.

 

[SURF]

As I was saying, I am just interested in how much detail exists, and how far one should take it—24MP, 36MP, 50MP? Stitching 4 or 6 shots of 35mm?

5000 pixels on a short side of 35mm is enough for most films. That makes one think that 5Kx7,5K=37.5Mp is enough. But... DSLR resolution is 70% vertical and 70% horizontal. 0.7x0.7=0.49 (0.5). -----> 75Mp DSLR=37.5Mp true scanner. So... 75Mp will be nice to have.

 

[SURF]

First a shot of the whole apparatus:

I like that! Great for modern films. Not sure if it will work with old curly films.

 

[SURF]

I tried the Sony A7 on the BEOON with:

- Micro-Nikkor 55mm f/3.5 (my old standard)
- Micro-Elmarit-R 60mm f/2.8 (one of the finest macro lenses in this focal length ever made)
- Summicron-R 50mm f/2 (superb, flat field lens: almost a reference lens)
- Schneider Kreuznach Componon S 50Mm F/2.8 enlarging lens
- Rodenstock APO-Rodagon-N 50mm f/2.8 enlarging lens
- Voigtländer Color Skopar 50mm f/2.5
- Summicron-M 50mm f/2
- Elmar 50mm f/3.5

I see the culprit is a rather short focus. 75-110mm lenses can change this. But BEOON can be too small for them...

 

[DrTebi]

5000 pixels on a short side of 35mm is enough for most films. That makes one think that 5Kx7,5K=37.5Mp is enough. But... DSLR resolution is 70% vertical and 70% horizontal. 0.7x0.7=0.49 (0.5). -----> 75Mp DSLR=37.5Mp true scanner. So... 75Mp will be nice to have.

A Foveon sensor would throw this calculation off again 🙂


I have been experimenting with a Sigma SD-1M for a few days now. It's quite impressive how there is hardly any difference in detail between the SD-1M and a 36MP Bayer sensor image (scanned slide image, that is). I like the contrast and clarity of the Foveon better. The SD-1M is a bit of a beast that needs to be tamed though... and there is always a small red hue issue, which is easily fixed in post-processing though.

 

[DrTebi]

I like that! Great for modern films. Not sure if it will work with old curly films.

It should work with curled films... that is pretty much the purpose of the whole "jig".

Takes a bit of fiddling to get curled film into the holder, but once it's in, it's flat.

 

[DrTebi]

I see the culprit is a rather short focus. 75-110mm lenses can change this. But BEOON can be too small for them...

Can you elaborate?

 

[SURF]

Can you elaborate?

Since introduction of Sony mirrorless we know that digital sensors covered with glass struggle with wide angle lenses in the corners.

So the better solution can be using APSC with 75-80mm on BEOON.

Sure 55mm lens is not too much wide angle, but you got the idea when we are looking for the best possible quality in the corners of the frame.

 

[SURF]

A Foveon sensor would throw this calculation off again 🙂


I have been experimenting with a Sigma SD-1M for a few days now. It's quite impressive how there is hardly any difference in detail between the SD-1M and a 36MP Bayer sensor image (scanned slide image, that is). I like the contrast and clarity of the Foveon better. The SD-1M is a bit of a beast that needs to be tamed though... and there is always a small red hue issue, which is easily fixed in post-processing though.

Can you show 100% crops of that comparison? It is interesting.

 

[DrTebi]

Since introduction of Sony mirrorless we know that digital sensors covered with glass struggle with wide angle lenses in the corners.

So the better solution can be using APSC with 75-80mm on BEOON.

Sure 55mm lens is not too much wide angle, but you got the idea when we are looking for the best possible quality in the corners of the frame.

That's interesting.

The SD-1M is also APSC, and my Macro lens is a 70mm. I can definitely see much better corner sharpness, and also less vignetting when comparing to my scans with the Nikon D810 with the 60mm Micro or the Pentax K-1 with the 50mm Macro.

It seems almost counter-productive, but for film scanning, an APSC sensor actually seems to be better suited than a full-frame sensor. APS-C "crops out" the best of whatever macro lens...

I haven't seen any 105mm Macro DSLR-Scans of a full-frame camera yet... so you say those should turn out better? Why is that?

 

[DrTebi]

Can you show 100% crops of that comparison? It is interesting.

I can do that... question is, up-size the Sigma image, or down-size the Nikon one?..

 

[DrTebi]

So here I down-sized the Nikon D810 image to fit next to the smaller Sigma crop...

This comparison is far from fair... the light source was different in both shots, and so was the diffuser. Obviously, exposure hasn't been adjusted equally, either.

Nevertheless, this gives some kind of idea I hope. I'll try to provide another example later.

left: Sigma SD-1M with 70mm Macro, right: Nikon D810 with 55mm Nikkor-Micro

 

[DrTebi]

Here the same thing the other way around. The Sigma image is scaled up to fit the size of the Nikon crop.


left: Nikon D810 with 55mm Nikkor-Micro, right: Sigma SD-1M with 70mm Macro

 

[DrTebi]

Here is a better and more delicious example. The Nikon crop is in its original 100%, the Sigma crop is scaled up to match.


The Nikon shot was lit with "Flexlight UltraBright™ High CRI (93+)" LEDs in a custom light box.
The Sigma shot was lit with a light box constructed around the "Solux True Daylight 4700K 50W" halogen bulb.


left: Nikon D810 with 55mm Nikkor-Micro, right: Sigma SD-1M with 70mm EX DG Macro

 

[SURF]

Here is a better and more delicious example.

Delicious as it can be.

I see much more resolution on the left side. Do you know grain trick? I use it to compare grain on flat parts of the images.

 

[SURF]

I haven't seen any 105mm Macro DSLR-Scans of a full-frame camera yet... so you say those should turn out better? Why is that?

I play with what I have and it is APSC and a bunch of scanner lenses. I have tried lenses in 105-110mm range too. Some are not great, some are, so it more depends on the lens.

For 35mm I settled on Pentax bellows and 75mm lens. It's compact and using slide/roll adapter one can scan rolls really fast. Not the best film holder, but OK. If I want to go deeper I attach those bellows to moving carriage that can roll (manually) on a sandwich of two big scanner AN glasses. (I have some). Light table is below the glasses.

Thus it's possible to scan all kinds of film, though I'm not fully happy with stitching and prefer one shot scans.

Thanks for the samples a lot!

 

[DrTebi]

Delicious as it can be.

I see much more resolution on the left side. Do you know grain trick? I use it to compare grain on flat parts of the images.

Hmmm... "much more", I don't see much more. Maybe a hair more. Literally.
What is the "grain trick"?

 

[DrTebi]

I play with what I have and it is APSC and a bunch of scanner lenses. I have tried lenses in 105-110mm range too. Some are not great, some are, so it more depends on the lens.

From my experience, it appears to be an advantage to use APS-C vs. full-frame. Because you will get much less vignetting. It's easy to test—just take a picture of your light source only, properly exposed. Then open it in GIMP or PS and check the RGB or HSL values in the corner vs. the center.


My full-frame scans had much more vignetting that the APS-C scans. It's not noticeable at first, until you have corrected the vignetting...

For 35mm I settled on Pentax bellows and 75mm lens. It's compact and using slide/roll adapter one can scan rolls really fast. Not the best film holder, but OK. If I want to go deeper I attach those bellows to moving carriage that can roll (manually) on a sandwich of two big scanner AN glasses. (I have some). Light table is below the glasses.

Thus it's possible to scan all kinds of film, though I'm not fully happy with stitching and prefer one shot scans.

Thanks for the samples a lot!

What is your problem with the stitching? What program are you using? I can recommend Image Composite Editor, a free stitching program from Microsoft. It's scary how well it works in all automatic mode...
Just make sure you shoot all images with the same exposure, e.g. manually adjust for the brightest section, then leave those shutter-speed and aperture settings for all shots.