Yes, it will be interesting to hear how @therealpiccard got on with the Rayfact, and how it was mounted to the camera. However I don’t think that the 50mm Componon-S has a particularly good reputetion for 1:1 camera copying (though you don’t say what format you are comparing). I’ve not found any 50mm enlarger lenses that excel at that. The 80mm Componon-S is very good though.
Since flatbed scanners are line scanners it would be unusual to see a diiference in the corners but those Epsons, even the later ones, only give about 2400 ppi despite their inflated claims. The pixel resolution that you get with a digital camera will in a large part be limited by the pixel count of the sensor but for 35mm a 24MP sensor will easily exceed that, for 120 they will be more on a par.
My copy of Componon-S 50mm/2.8 seems to have an edge over Epson in the middle of the 35mm frame, where as corners are visibly weaker, even at less than 100% magnification.
Tested with a 100ASA color transparency and my feeling is, that usually there should be not many details more to resolve. We are probably talking about resolving grain, not image details. So for practical reasons, it seems to me that the level of Epson resolution was chosen for a reason.
I recall using a Nikon Coolscan scanner parallel to Epson, back in time, and wheras Nikon scans seemed to be overall sharper (more resolved), there were massive problems with edge sharpness. That was so annoying (to me), that I eventually got rid of the Nikon.
Taking all that into account, my question would be - are the usual, modern macro lenses really capable of resolving grain, into the edges of the image, and on a higher level that an Epson flatbed/Nikon Coolscan would do?
Robert O’Toole was talking about using the 50mm Componon-S f2.8 as an inexpensive general purpose macro lens, reversed, at magnifications between 1x & 4x and not for demanding flat field copying, he didn’t use it for any of his his comparisons with silicon wafers and says there that ”the only weakness with this lens is that the corner sharpness is not as strong as the center “.
I do suspect that there must have been something wrong with your Nikon Coolscan, although you don’t say which model I’ve never heard of anyone complaining that the edges of the scan were unsharp. As the film goes through the scanner longitudinally this would have meant that both the long edges of your scan would be unsharp.
There is certainly a valid debate regarding how much resolution is enough. The standard resolution for 35mm film scanners was 4000 ppi, some very expensive makes gave more than that. In terms of the pixel dimensions of the final scan that equates to a 21MP camera sensor.
I ‘only’ use a 24MP camera and I’m happy with that, I also have an old 16MP Fuji that gives results that I would also be happy with that but there is more detail in the 24MP. Still various people have demonstrated the advantages of using higher MP sensors and there are clear benefits but these do generally involve looking at how the grain or dye clouds are resolved, still important if you want to print very large (and don’t mind a lot of retouching).
Putting aside the rather low actual resolution of Epson flatbeds for 35mm copying they are also very slow compared to using a camera and I suspect that modern digital sensors will display a higher dynamic range though I haven’t tested this.
However finding a lens that is sharp right into the corners for ‘camera scanning’ is a challenge, many fail to do that but the 80mm Componon-S is very good and often inexpensive, and it is well suited to the lesser magnification of 0.67x for 35mm on to APS-C. I eventually found a 75mm Apo-Rodagon 2x which is ever so slightly better but hard to see on actual film, Sigma 70mm & 105MM ART macros also get a lot of praise.
Here’s a useful comparison from @VladS of the Sigma 70mm ART and the 75mm. Apo-Rodagon 2x on a full frame 30MP Canon.
thank you so much for this detailed response, very appreciated!
Yes, I am aware of context in which Robert O’Toole talks about Componon-S, but this info is better than nothing. There is a lot of misleading information on the net - what a surprise - and with time and experience I’ve become very sceptical about what people and/or commercial web sites are saying and promoting.
Maybe I should give some Rodagon a try, as you say. Sigmas get a lot of praise, but when you look closely at test images, they don’t confirm that.
I tried digitalising my old negatives ca. 20 years ago, using a 10MP APS-C DSRL and a micro-nikkor 55mm/3.5. There was virtually no info on the net at that time, I was not sure about the whole process and the results, and always felt there could be more in to it. So I abandoned the project - until now.
Speaking of 35mm Coolscans, they were ok (and slow!), but afaik the problem laid probably in the film holder. Most negatives are a bit curly, and Coolscans couldn’t compensate for that, their depth of field was too thin. I remember, there was a guy in the UK selling custom made film holders (very expensive) for a LS-8000 (different model, but the working principle was the same) to overcome this.
With a Coolscan you could set the AF-point anywhere in the image, if you chose a point near the edge the center would get visibly unsharp, and vice versa. Not so good for images with important details near the the edge of the frame
I bought an adapter in Latvia, as recommended by Robert O’Toole. However, it didn’t fit the Sony camera very well. So I designed my own adapter and printed it out using an FDM printer (PETG filament). I painted the adapter with table varnish – I don’t know the correct name for the varnish in English. I covered the metal parts with velvet. You can download the adapter from “Onshape.” I attached the original metric fine thread to the adapter with three screws.
Actually during the day it occurred to me that it might have been film curl that was causing that problerm with the Coolscan. That is also a factor with ‘camera scanning’ of course though you can at least either buy or improvise a holder that minimises it, or even focus stack.
Robert O’Toolle didn’t really rate the Apo-Rodagon’s for some reason, he talks about them in another rather obscure thread in another forum. Note that the 50mm Apo-Rodagon N is not appropriate, that is definitely an enlarging lens, really sharp at the centre but not good towards the corners. They have to be the ‘D’ versions, the 1x for 35mm copied to full frame and the 2x for 35mm on to APS-C, or 120 on to either.
Same here , as an afterthought, may be it was the lens in Coolscan as well. Looking at an image in the landscape orientation, these probs with unsharp edges were to the extreme left and right. Not so much up and down, but left and right. But then again, who cares if the clouds are sharp, maybe I simply did not notice it then.
Speaking of lenses, it seems to me now it should be either the D Rodagons - thank you for the hint - or the modern Sigma ART lenses. Other choices are probably too expensive or too exotic.
Or, maybe I’ll give up and come back to the old trusty micro nikkor. My old negatives are not that sharp anyway, and keeping evertyhing flat and parallel will probably be much more decisive than a super de luxe macro lens.
Maybe not widely known, it seems that Rodenstock doesn’t sell their industrial lenses, like the Rodagons-D, by themselves. They are even not mentioned at their website (or at least I couldn’t find them there).
You can find some info on such offerings elsewhere in the net, at some hi-tech company, owned by private equity.
Anyway, and that’s interesting, they say that Apo-Rodagon-D lenses are good for cameras with sensor pitch of 7 microns and above. Well, my camera has only 24MP and a pixel pitch of 6 microns.
How does this translate into real world usage, I don’t know, but it seems that digital sensors are once again ahead of traditional optics made of glass.
I guess you mean Linos, I wasn’t actually recommending buying a new one, I think there would be better ways to spend your money, probably the Sigma 70 or 105 ART route but there are other contenders that come up by searching this and other forums. The 75mm 2x seem to be available from the Far East as ‘ex’ machine vision lenses. A bit of a gamble of course. You would have to judge whether the results in that test I linked to above on the 30MP Canon offered a significant benefit over what you have already. Actually I think at 24MP the 80mm Schneider Componon-S would do you fine but that means a bellows, preferably with a built in focus slide like the Canon FL or Nikon PB-4, but then so would the Apo-Rodagon.
sure, the Linos That’s the problem, there are lots of advices here and there, but what would bring me a real world advantage, hard to say. As I still own the Epson Photo flatbed scanner, along with belows, enlargers and similar stuff - I have plenty of options. My starting point is the level of quality delivered by the Epson though.
Schneider Componon-S is better in the middle of the frame (better resolution and contrast) than Epson, but it degrades quickly towards the edges, exactly like other people found out. It would be nice to have the level of quality like Schneider’s center, but kept towards the corners.
Leaning towards the Far East gamble at the moment, or, as a workaround, to treat the Schneider as an APS-C lens, with loss of MPixels in the resulting scans.
Pragmatically reasonable approach. Nevertheless, a “better” lens will still improve the scans. Current (non-exotic) lenses with high ratings seem to come from Sigma and Venusoptics (Laowa)
I use an old Canon EF100mm macro and a 7Artisan 65mm macro II. Both good enough (I’m not printing posters for pixel peeping) and the 7Artisan is much easier to manually focus because it has a long throw. The canon lens is fiddly - and neither AF nor focus peeking are precise enough.
Speaking of a reasonable approach, I would suggest the following link. I am kinda reluctant to invest another batch of $$, only to find out that at the end of a day nothing really has changed.
It would be nice to have a great extra scan lens for those few once-in-lifetime frames, though.
@Richard1Karash’s tests of lenses for film copying are so useful, together with his real world examples. The links at the bottom of that 35MMC article don’t work any more, I think they were pointing to these forum posts:
Here is a quick & very dirty test. A 24 MP camera is “hanging” attached to a filter thread of an enlarger lens, which in turn is attached to an enlarger in the usual way, meaning, the same as if you would print on a paper (not reversed).
The slide is Fuji RD 100 ASA, and the lens is Schneider Componon-S 50/2.8. The picture on the left is taken at the 5,6 aperture, the picture on the right at the 11. Screenshots are at 100%. First middle of the frame, second frame edge.
Surprisingly, I see very little image degradation due to diffraction. The original slide is sharp edge to edge.
Maybe I am doing something wrong and/or the lens is out of its scope, like a rangefinder lens attached via an adapter to a mirroless cam?
The appearance of “sharp” is subjective and changes with contrast … and noise. Leaving some luminance noise can make an image look sharper. When the advanced NR features were introduced, I applied DeepPrime and added back some grain with FilmPack. Looks like our brain is doing the sharpening in such a case…
Is this test designed to illustrate the effects of diffraction, or the effect of reversing an enlarger lens on a digital camera, or the feasibility of mounting a digital camera body to an enlarger in this way?
An enlarger lens is generally specified for magnifications of, say, ‘2x to 20x’ which handily covers most sizes of prints that anyone would need to make. It doesn’t mean that it will be optimised for either 2x or 20x, so somewhere in between. For film copying with a digital camera you take the reciprocal of this if you mount it to the camera in the normal way (i.e not by the filter ring, reversed) so 1:2 to 1:20. Alternatively if you reverse it then the magnification range reverts to 2x to 20x for film copying and this is why something like a 50mm enlarger lens can be very successful reversed for copying cine film on to an APS-C or 4/3 sensor. Alternatively 75/80mm enlarger lenses, with a larger image circle, can be very good for copying 120 film in their normal orientation.
1:1 is out of its stated range so although by all means try it I’ve not found that reversing an enlarger lens helps to get round the weaker corner definition for 1:1 film copying.
When using Vlad’s 35mm Test Target on full frame or APS-C I’ve found that as a general guide diffraction begins to set in at around f5.6/8 for APS/C and f8/11 for full frame.
Here is Robert O’Toole’s test of many enlarger lenses at 2.8x, at the bottom he shows the orientation of each of the lenses used:
…and other lens compromises usually profit from closing the aperture. All those effects combined provide a sweet spot … which can be understood as a place in which none of the properties are at their best or worst.
Lenses that are tailor made for a specific imaging ratio are best used as described. They are usually sold new for four digit prices and possibly lower than that when used. Whether such lenses are necessary or not can be a very personal decision.
- and with time and experience I’ve become very sceptical about what people and/or commercial web sites are saying and promoting.
