E-M5 Mk II Sensor Shift 40 MP mode for film scanning

[coelacanth]

So there was a thread earlier about using Pentax K-1 "pixel shifting" high-res shooting mode for film scanning so and that reminded me my E-M5 Mk II got sensor shifting 40 mp mode that I completely forgotten because I was using the OM-D for underwater photography where sensor shift image is not gonna end well at all.

So I did my usual Leitz BEOON copying stand + USB light box method. Usually I use X-E1 and Elmar 50/2.8 LTM, but This time I swapped the camera with E-M5II. The rest is just about the same. Took the shot in sensor shifting mode, did basic adjustments in Lightroom.

I copy & pated the adjustments, but the overall brightness was slightly off between two images, probably due to resolution difference affecting the LR's calculation. Regular res image was slightly darker and I adjusted that to reasonably match the high-res output. Dust removal was also manually done per image and I did a quicky so they might not be matching.

The original photo was taken with M4-P, Summilux 50mm Pre-ASPH E46, Tri-X shot at ISO 1250 - 1600 ish, developed in Diafine so it's not really a super clean picture to begin with. I really don't shoot fine grained low ISO films, but at least you can see the grains here. 😀

This is regular 16 MP mode, resulting in about 14.2 MP output due to 3:2 ratio crop.

This is sensor shifting 40 MP mode, resulting in about 35.5 MP output.

In a small browser viewing size, you might not see the difference, or probably Flickr's resampling might be fooling the actual quality.

Here are a couple of cropped comps. They are still resized for viewing.

L: regular output, R: high-res output

This is not a scientific lab test, but for us film shooters, the trend of sensor shifting hi-res mode might have some nice advantage for film scanning.

 

[coelacanth]

Here is a comp of high-res mode (left) and regular mode (right) at 100% size.

Sorry for flipping the positions from the OP. I'm too lazy to fix that now.

Now, matching the viewing size of the two.

 

[coelacanth]

One last thing is to compare with X-E1's full 16 MP output at matching viewing size.

Left is E-M5II in high-res mode. Right is X-E1 in regular 16 MP mode.

Both E-M5II and X-E1's 16 MP shots were taken in RAW and developed in LR. E-M5II's high-res mode is recorded in Jpeg.

 

[lynnb]

This is very interesting, thanks for posting Sug. Is it possible to show a comparison with, say, a V700 (my current film scanner)?

 

[coelacanth]

I just printed this shot with R3000 on fiber paper, and I am quite happy with the result. When I'm too busy to go down to darkroom, this is the next best thing.

Lynn, unfortunately I don't have V700 or any other flatbed/film scanners. I just didn't like the scanning experience with them, so much so it was making me wanna shoot less film. I ditched the whole conventional scanner approach and went with BEOON. I just do framing, grain check and exposure check like you'd do with darkroom enlarger, and I'll just shoot away to scan the negatives. This put me back on film pretty much full-time for on-ground photography.

 

[DC1030]

I do my film scanning with a GX7 with a Nikkor 3.5/55 macro lens, since my nikon 9000 scanner died. i also thougt about buying an olympus camera with pixel shifting. maybe i get the pen f. woud be a win win situation. nice camera, high res scans...

 

[shawn]

Would be interesting to see this with a color negative. The pixel shifting is really more about trying to increase the color resolution by recording RGB at every pixel location.

Shawn

 

[robert blu]

about buying an olympus camera with pixel shifting. maybe i get the pen f. woud be a win win situation. nice camera, high res scans...

Similar idea here for when my 5000ed will die (hopefully late!) .
The Pen F small camera as a small camera beside my film cameras and tool for digital reproduction of my negs...hmmm, thinking about !
robert

 

[mani]

Really interesting! Thanks for posting and for the trouble you took with all the comparison shots!

Also thinking about the new Pen F as an alternative scanner - I'm not sure Olympus had film photographers in their plans for this camera, but at least the looks will blend in with all our other cameras!

 

[coelacanth]

I do my film scanning with a GX7 with a Nikkor 3.5/55 macro lens, since my nikon 9000 scanner died. i also thougt about buying an olympus camera with pixel shifting. maybe i get the pen f. woud be a win win situation. nice camera, high res scans...

Yeah. If I'm was looking for a new body, Pen F could be it (only off they ever release underwater housing for it). My main use of the camera will be underwater photography and film scanning.

I'll have to see if I can get this setup work for medium format now.

 

[coelacanth]

Would be interesting to see this with a color negative. The pixel shifting is really more about trying to increase the color resolution by recording RGB at every pixel location.

Shawn

I'm pretty much completely black and white guy but love to see someone trying that.

 

[coelacanth]

Similar idea here for when my 5000ed will die (hopefully late!) .
The Pen F small camera as a small camera beside my film cameras and tool for digital reproduction of my negs...hmmm, thinking about !
robert

Yeah higher pixel count of Pen F sounds great for scanning. One thing I don't like about micro 4/3 has been 4:3 native aspect ratio as I prefer 3:2 (or 6x6) for on-ground photography. I actually prefer 4:3 for underwater, and now this wound work great for medium format film scanning if I can make my BEOON work with it. We'll see...

 

[coelacanth]

So I tried 6x6 medium format scanning with this method.

Note: that it's Tri-X at 1600 developed in Diafine so it is far from most fine grained image to begin with. Also the lines/marks you may see in the sky are just my ****ty drying job (I wanted to test the magazine really quick so I used wiper on it) and not part of the scanning process.

Since BEOON with Micro 4/3 can't cover the entire height of 6x6 on its own, I made a booster out of cardboard box. With that I could fill pretty much the entire height of the frame.

For 6x6, E-M5II could give you about 24 MP output.


Lord of Bees by Suguru Nishioka, on Flickr

Here is a 100% view crop of the scan.

 

[coelacanth]

OK one more post and I'll shut up. 🙂

Probably a bit more test-shot-like shot.

Same setup as above:

Hasselblad 500c, Planar 80mm C, Tri-X at 1250 - 1600, Diafine. Scanned with E-M5II, Elmar 50/2.8 via BEOON + DIY booster.
Width of this photo is the entire height of E-M5II sensor.

100% viewing crop.

 

[robert blu]

This is all interesting, I like the way you reproduce the real grain of the film, would you post a photo of your set up please? Curious to see it 🙂 I like the idea but I'm a little worried because of practical issues, like space and how to keep the film flat (scanning holder maybe?).
Thanks, robert

 

[mcfingon]

As a comparison, Sug, here is a 1987 Hasselblad FP4 and ID-11 scan done on a Nikon D3300 at 24 megapixels, first full-frame then cropped to 100%
Black Duck Rally 1987

100% crop

 

[coelacanth]

This is all interesting, I like the way you reproduce the real grain of the film, would you post a photo of your set up please? Curious to see it 🙂 I like the idea but I'm a little worried because of practical issues, like space and how to keep the film flat (scanning holder maybe?).
Thanks, robert

Robert,

This is my setup.

• BEOON the film copying kit. I'm super happy I got this thing. It also came with a grain checker that you'd put in place of camera body to focus. If you are using M8 or M9 (or I guess Type 262), you'd be using that to focus.

• USB powered LED lightbox/panel. You don't even need one this big, but I use this for other stuff as well.

• DIY 35mm film holder. I made this so have I max control of framing, but for 35mm, the holders that came with BEOON would work fine.

• DIY 120 booster made of Amazon audio cable box. I just made this to elevate BEOON beyond its regular max extended position to cover medium format frame with Micro 4/3.

• I just used masking tape to keep the 120 film down on the lightbox for the test, but I'm gonna now make a similar holder to the one I made for 35mm.

• And of course, E-M5 Mk II. I used the high-res mode with 1 sec delay to eliminate camera shake.

• Elmar 50/2.8 LTM. Any LTM or M 50mm should do the job.

 

[coelacanth]

mcfingon,

Nice! I should try some fine grained negatives soon, but I am usually a 2-stop-push, lazy-boy developer kinda guy... 🙂

 

[quejai]

This is impressive!

Unfortunately I think the limit is going to be the lenses, getting a flat focal plane on both the filmstrip and the image sensor such that the corners are still as good as the center. Also, if the original lens used for the film photograph had poor corner performance, using another lens with poor corner performance could further the resolution losses.

An area that I think is commonly underappreciated is illumination. Since the scanning lens records a cone of light from each point on the film, ideally you'd want a lighting system that can provide that cone of light, and nothing more, otherwise contrast could be reduced. One way to achieve this is to have the diffusion light source at a greater distance to narrow the cone of light provided by the light source.

Further, blue light has twice the resolution potential as red light in terms of wavelength; so if you're scanning B&W film, try to get a monochrome blue light source, ideally LED, if you're after resolution. However, green having double the number of pixels in a bayer array kind of cancels this out. I'd be interested to see the effect of blue illumination on one of these sensor-shaking cameras. A monochrome light source for B&W would also eliminate chromatic abberation issues.

 

[mcfingon]

This is impressive!

Unfortunately I think the limit is going to be the lenses, getting a flat focal plane on both the filmstrip and the image sensor such that the corners are still as good as the center. Also, if the original lens used for the film photograph had poor corner performance, using another lens with poor corner performance could further the resolution losses.

An area that I think is commonly underappreciated is illumination. Since the scanning lens records a cone of light from each point on the film, ideally you'd want a lighting system that can provide that cone of light, and nothing more, otherwise contrast could be reduced. One way to achieve this is to have the diffusion light source at a greater distance to narrow the cone of light provided by the light source.

Further, blue light has twice the resolution potential as red light in terms of wavelength; so if you're scanning B&W film, try to get a monochrome blue light source, ideally LED, if you're after resolution. However, green having double the number of pixels in a bayer array kind of cancels this out. I'd be interested to see the effect of blue illumination on one of these sensor-shaking cameras. A monochrome light source for B&W would also eliminate chromatic abberation issues.

Good explanation, quejai. I noticed more crispness in my scans when I swapped put the normal enlarger globe in my copy setup and replaced it with an LED. The LED has a more even field of illumination and is brighter, as well as being cooler-running.