Slide holding for Scanning - Ohnar and Other Options

Cheers. I will have another try next time I have everything set up. I will us an actual razor blade and make sure it is in focus. I don't think there is much mileage in analysing the etch image further because the metal edges will be less than perfect and my focus was off because the metal fret has very little rigidity.

Having read a little more about MTF I'm not sure that it's addressing the same issue. If I understand it correctly, it is looking for the boundary where the contrast is wrong by 50%. My haloes are probably a good deal less than 50% but they are visible and I'd love to reduce them. Can that software use a different contrast value?

In terms of sharpness I think that I'm already near the limits of what most of my sldes can deliver - not the film, just the limitations of my cameras and my ability to use them at the time the photos were taken.

My best guess is that the halo is a variation of ghosting flare but that is just my guess. Most discussions about flare concentrate on the effect of the Sun or other concentrated light sources rather than a diffused backlight.

I tried the Razor test and MTF Mapper using the 80mm Rodagon lens this evening. I don't have any fancy ways to do fine focus adjustments.

I first focussed manually using my usual steps of setting the lens to F4 and using tethered live view to focus on a slide image which includes some wire mesh detail. Then did a series of shots at progressively narrower F stops (with the exposure time increased to suit). I did not find a huge variation - 0.06 to 0.09 with the higher values generally in the middle although there were some ups and downs.

Then I set the lens to the middle mark between F8 and F11 (which is where I usually use it). I wound the focus forwards by three ridges on the bellows focus wheel grip (that's about as accurate a measurement as I can do) and then took photos (8 in total), moving the bellows back by one ridge each time. I don't think I went forward far enough but the result was quite a wide plateau reading 0.09 or 0.10 which fell off to 0.05 by the last photo.

I tried to save the results but just got empty directories.

I've no idea what 0.09 or 0.10 means. Higher seems to be better. I do think this has confirmed that I'm using the lens in the best range and that there is some tolerance for focus when I stop down to the middle mark.

I may test the EF 40mm on the 21mm extension at some point but it's a bit of a pain to swap the rig between a bellows mounted and a camera mounted lens.

Bernard Delley wrote:

so you managed to install mtf_mapper and did get results. I assume these numbers were read directly from the mtf_mapper output.

I pre-process the raw file and crop to my chosen region of interest, either center or further out portion of edge. I call the resulting file pix.ppm

linux options (in version 5.15) to produce full MTF output file: edge_sfr_values.txt

mtf_mapper pix.ppm . -q -b --bayer green

Thanks.

I tried to use it on the Canon raw files but that produced an error. Instead I turned off most of the optional modules in Darktable (but kept white balance on) and then exported as a PNG with the original image size.

I used the MTF Mapper GUI and opened each image with manual edge finding. I'm not sure if I did right or wrong but trying it on crops of the etch fret image produced very different numbers. It occurred to me that cropping and scaling could affect the answer so not being sure of exactly how it works I decided to do neither for the time being.

I was also unsure about wether the accuracy of my manual edge marking would affect the result - the width of the marked line seems pretty narrow especially when compared to the width of the more blurry images of the edge.

SterlingBjorndahl wrote:

Halina3000 wrote:

switching between lenses doesn't seem to improve things but maybe there is an optical improvement that I have not come across.

You could try using greater magnification to take partial, overlapping images of your target and stitch them together in post-processing, sort of like we commonly do to create a panorama. If part of the cause is because of flare inside the glass, this should reduce the flare-to-image ratio (I'm not an expert so I just made that term up) so that at equivalent magnifications the flare should be less noticeable.

What is your output target? Print? Web? For a "reality check" make sure you're not zooming in more than what your viewer/customer will use. Printing an 8x10 may help you determine whether this problem is serious or not.

Sorry missed your message earlier. Output target... could be web or print depending on the photo. Some are family shapshots that will be shared with other family members online, others are images where I hope to get a decent quality print. Mainly I'm trying to make a decent stab at digitising my slides where previous attempts have failed... partly because of poor quality results and partly because of the time spent waiting for scans.

I'm really not set up for overlapped multi frame captures... basically I want to get the best result that I can achieve from a single DSLR frame.

Having thought about the halo/flare issue some more I've come up with a multi part sanity check...

1. Problems with the real slide should be the same regardless of capture optics. Problems with the capture optics should change with the optics... not that I have a vast range of different optics to swap.

2. Both of the above can be detected by pushing the contrast with curves. This shows that there is a real issue but it is only perhaps 25% of the scale that I'd first thought. It doesnt vary with the two lenses that I've been using so could be ghosting from the skylight filter on the 35mm SLR or the mediocre optics of my first camera.

3. If a fringe remains the same size when the image is zoomed then it's not a problem with the image. This seems to be the more prevalent issue. Now that I've seen it by peering at photos I am also seeing it on other screens with web pages etc.

4. If it's my eyes or brain and not the screen then masking the light side of the boundary of my test SVG with some card should make the halo vanish... which it does. I see the same effect with each eye if I try one eye at a time. Using the card to mask the photo on screen also has the same effect.

So there you are - it seems that the issue is maybe 25% real halo and 75% down to me, the observer.

I don't think this needs to delay my slide digitising project further but I can't say that I am altogether happy with the 75% part of the conclusion.

I've been on the lookout for a decent manual focus macro lens for a while and have now acquired a nice example of the Nikkor 55mm Micro at a fair price. I want the lens for taking new photos with the Canon (and it does that very well) but also wanted to do some comparative tests with the other lenses in the slide digitising setup.

After comparing several results and convincing myself that one image was sharper than another I've now concluded that the variation between images from the same lens is more than any difference in sharpness between slide images from the Nikkor 55mm and Rodagon 80mm. Choosing the best images from each lens I can find nothing to choose between them, either in the centre or at the edges.

I did the razor test at different apertures. The best number I got was 0.09 at f4 which is marginally lower than the best number from the Rodagon but having convinced myself that lack of rigidity was affecting some shots I am not reading too much into this.

The rig would need more work to allow better alignment, better rigidity and so on -  the Nikkor lens and extension tube hanging off the front of the Canon don't help. I shall continue with the Rodagon for digitising my slides because the bellows supports both the front flange of the camera and the rear of the lens.

A couple of 100% crops from the left hand edge of a test slide... not very inspiring subject matter I'm afraid.

Nikkor Micro 55mm f2.8 Image @f8

Rodagon 80mm f4 Image @f8-11