Iāve been looking for that Rodagon, as it is reputed to have nice flat field for scanning. It is optimized for 1:1, so should do very well on 35mm, but I am not sure about pulling back for larger formats. Itās probably fine, but worst case scenario, you can always grab a cheaper lens that does very well at 1:2 like a 55 Micro Nikkor.
Other than @koenji75 , my approach is to get the optimal lens for the large negativeā¦
Assuming that the lenses you took the original photos with are/were good lenses with comparable ppi resolution, the 67 negatives have a higher pixel density when āscannedā with the camera. Therefore, Iād select a lens that is optimised for about 0,5. Industrial lenses tend to be well documented and you can check specs for resolution figures at other ratios.
@brothernature , think of printing: for bigger prints, the MF negatives provided more detail than 24x36 negatives. You didnāt get a 67 camera for 4x6 prints after all. Now, if your scanning gear is optimised for 67, itās easily good enough for 24x36, even when captured at 50%.
Format mismatch loss: Sensor image ratio is 2x3 or 6x9, 67 is 6x7, hence 2/9 of pixels are lost - more if you also capture bigger than the exposed area.
BTW: Welcome to the forum and the world of camera scanning and NLP converting.
Thatās a fascinating prospect you have there, going from the finest medium format scanner (aside from drum scanners) to ācamera scanningā with the 61MP Sony full frame sensor in your A7RIV.
Your Nikon Coolscan 9000 ED is of course a line scanner, but a top end one that has a 10,000 pixel tri-linear CCD. This meant that for 120 film it used 8964 of those pixels to stretch across the 56.9 mm width of the film leading to 4000 ppi for medium format, and contemporary tests showed that it actually came very close to that in actual measured resolution. That means for a 6x7 (56 mm x 67 mm frame) youād get a 92MP image (!).
The Sony sensor āonlyā has 6336 pixels on its short side so for 6x7 in a single shot thatās a maximum of 48MP. Still plenty but for anything more youāll need to stitch so then your 75mm Apo-Rodagon D 1x will be closer to its comfort zone otherwise at 0.42x itās liable not to be the best. The good news is that with your bellows setup there are lots of others to try like the Schneider 80mm Componon-S or, even better, the 75mm Apo-Rodagon D ā2xā. In reality the differences between those two is not going to be huge but I think youāll see differences with the ā1xā.
For 35mm at 1:1 the ā1xā is going to be great and your Sony gives you 60MP instead of 22MP from your 9000 but that maybe more than you need.
Actually I think the ā2xā may be the best bet for you, and they are easier to get hold of. As I said it is fine for one shot copying of 120 film at 0.42x but its recommended range extends to 0.8x. If you donāt fill the frame of your Sony, but instead the 35mm negative fills 80% of the linear size of your sensor then youāll still get 35MP from 35mm.
The Nikon PB-4 bellows are great but if you donāt need the front standard shift and swing then the Canon FL bellows is just as well made, if not better.
āCamera scannningā will be much faster of course and these lenses work best at around f5.6/f8 so a little more depth of focus than your 100mm f2.8 ED scanner lens.
Thatās a fascinating prospect you have there, going from the finest medium format scanner
Haha, have you used one? They can be a real pain in the butt. Iāve repaired this one twice, and Iām afraid it has a minimum life span left. I shoot about 85% 35mm, so a 50MP scan for 6x7 is acceptable for now, and Iāll still have the Coolscan if I need to go higher.
Iād like to get a camera scanning setup started because it a) seems so much nicer as a workflow b) can eventually upgrade cameras (Iām guessing in a few years there will be a used Sony body that can match the Coolscan in # of pixels)
Thank you for your help! I have seen that article by @Mark_Segal Itās really helpful. I need to spend some more time with it.
No, I havenāt used one so I might have had a different opinion if I had I guess but technically they seem superb, people seem to still ask very large sums of money for them if they are working properly.
Mark Segal goes into stitching and the easiest way to proceed with that is if the long side of your sensor is across the short side of the film, so 3 overlapping exposures for 6x7. Your Sony is 9504 pixels on the long side so really youāre right up there with the Coolscan resolution if and when you needed to do it. Also the magnification of 0.65x is right in the middle of the ā2xā recommended range. On the other hand youāve also got pixel shift to try.
You probably also know that a lot of people go for the Sigma ART 70mm & 105mm on the Sony.
Clearly the way you support your camera is critical, a very solid stand, remote firing and probably tethering. I hope you report back when youāre up and running. Certainly itās future proof unlike these old scanners (I have an Imacon). Good luck with it.
Hi Digitizer - Iām a bit unclear where you mention that 6x7 negatives have higher pixel density when scanned with the camera - higher than what exactly? Are you referring back to his default option of the Nikon Coolscan 9000? Itās sensor resolution is 4000 pixels per inch (and effective resolution typically less), whereas the Sony A7RIV sensor is 6336 pixels per inch and depending on the set-up can pretty much deliver it, so if thatās what you meant, yes, true. I think the format mismatch doesnāt matter because the ālost pixelsā donāt exist in the original photo anyhow, insofar as the 2/9ths donāt have any data in the negative given its 7 long dimension. Heāll get 6336 pixels covering the 6 dimension, or 1056 pixels per cm, and heāll consume 7392 pixels for the 7 dimension, leaving (9504-7392) = 2112 pixels empty on the long dimension, which is inconsequential given the size of the media. I think the key thing he needs to be mindful of in creating his camera-scanning set-up is that the image circle of the MACRO lens he buys must be large enough to comfortably accommodate the 6x7 media size, in which case, as you say, it will be fully adequate for also photographing 35mm negs.
More (absolute number of) pixels per image hight reduced to smaller hight = more pixels/hight.
Or, if 135 film captures have 2400 pixels (short side), 67 captures on 120 film get 5200 pixels (short side) with the same emulsion and with lenses that were actually able to deliver these figures.
As for pixels/inch, 67 format requires a lens that can deliver over a larger area and can therefore deliver for the smaller negativesā¦but all of it is theoretical if target output is postcard sized.
I suppose that the OP thinks of 120 film when writing 5x7 - and not of 5 by 7 large format.
In analog times, films were meant to be able to print decently up to a certain size. Bigger prints required bigger negatives in order to avoid grain to be too visible from normal viewing distance.
OK, now Iām really puzzled, so bear with me as I think through step-by-step. Letās take this back to basics. Letās imagine the O/Pās capture set-up. The camera is a Sony a7riv, the same I am using, 61 MP, and if we look at it in Landscape orientation, 6336 pixels on the height and 9504 pixels on the width. Letās assume the lens he is using to capture the negatives can deliver the sensorās resolution, so no loss of resolution from the lens. His 6x7 film image is 5.6 cm high x 7 cm wide. If heās got his set-up consistently oriented, he fits the 5.6 cm film dimension to the 6336 pixel height of his sensor. If he makes an exact fit, he is getting 6336/5.6 = ~ 1131 p/cm resolution. Therefore on the width, applying the same resolution to 7 cm, he is getting 1131 p/cm x 7 cm = 7917 pixels used, out of 9504 available. So he is not getting the benefit of a remaining 9504-7917 pixels = 1587 pixels simply because of the difference of aspect ratio between his film format and his sensor format. Right? So for purposes of comparison just below, let us retain in mind these pixel dimensions from the Sony camera scan: 7917 wide x 6336 high total pixels.
Now letās compare this with his Nikon Coolscan 9000, which at best delivers resolution of about 4000 ppi = 1574 p/cm (given 2.54 cm/inch); if he scans the film with this scanner, he is scanning-in pixels = 1574 p/cm x 7 cm wide by 1574 p/cm x 5.6 cm high = 11018 pixels wide x 8818 pixels high. Have I made any mistakes yet?
So the end result is that the Nikon scanner gives him more pixels and therefore capability to make bigger enlargements than possible with his Sony a7riv, assuming the same ppi printing resolution for both options. The reason for this outcome is that the total pixel count of the camera is fixed regardless of the media size, whereas up to the maximum media size the scanner is built to ingest, the total pixel count is variable depending on the size of the media, and respecting the limits of the scanner resolution.
When it comes to inkjet printing from a digital file, of course the size of the original media ceases to matter - it is only the number of pixels in the image file produced from the scanning set-up that matters. Give me more pixels, all else equal Iāll give you a bigger print.
But the fact remains that the camera sensor does have higher sensor resolution than the scanner sensor (about 6336 ppi Sony versus 4000 ppi Nikon), therefore for prints of the same size, the result from the camera file could be better resolved than the result from the scanner file - if we can see the difference and depending on lenses qualities etc.
The room might be to big for the cat though, and what happens when the elephant has left the room?
Actually you make a good point if I understand you correctly but I would say that if 35mm is the top priority then there are very few lenses that perform very well at 1:1 on to full frame. Still I do think that a case could be made that 61MP is too many for 35mm so using a lesser magnification with your lens chosen for medium format will then mean that the corners of 35mm will clean up, I even suggested this myself so that the 75mm ā2xā Apo-Rodagon would be at its optimum.
If we are to take everything we are told from the Magnum videos, like this one (and I donāt see why we shouldnāt) then they use a Fuji GFX 120mm Macro which Iāve read elsewhere is not great for film copying at 1:1 since it is designed to go to 1:2. However they use their 100MP GFX to include the whole slide including the mount and a bit to spare so that the actual image area is a little less than 24MP by my āback of the envelopeā calculation. The slide is lit with carefully balanced polarised lighting so that only a single exposure is required. That will still give a 20" print on their Fujifilm printers and indeed the prints from that archive on display in London at the āFujifilm House of Photographyā last year were around that size, certainly not bigger than that, and the limiting factor was the grain. It probably didnāt help that they were next to giant prints from the GFX taken by Magnum photographers Olivia Arthur and Mark Power.
Iām only looking at the camera scanning part of things.
I do NOT compare camera scans to scanner scans.
If a scanning lens is optimised for 100% (FF reproduction with a FF sensor)
it will be softer at 50% which means that quality is sacrificed for MF.
If a scanning lens is optimised for 50% (MF reproduction with a FF sensor)
it can also reproduce FF, if scanned at the same lens and distance setting
(it can also scan at other settings with softer output)
For best results one must have two lenses, one for 100% and one for 50%.
As weāre less rich or willing to invest and therefore have to live with compromises,
we have to decide what weāre willing to sacrifice in exchange to lower cost.
For me, the goal is reproduction quality understood as āaccepted level of grain visibilityā. Therefore, Iād have to go for an āoptimised for 50%ā lens. But as I scan with a APS-C type sensor, things are different still. And with all the trials I made, I came to conclude that technical quality should be 2nd priority to āgetting that old stuff into todayās worldā.
Thatās how I imagined it would be, thatās true of my Imacon so can I get 6300 ppi from 35mm, 3200 ppi from medium format, 2050 ppi from 5" x 4". Thatās because the lens can change its position with respect to the film but that canāt happen in the 9000 so 35mm is definitely the poor relation at only 22 MP and just using 4000 RGB pixels from the 10,000 pixel array.
Yes, APS-C for me also as 24MP is fine for 35mm and lots of enlarger lenses are good to very good but the 75MM ā2xā f4.5 Apo-Rodagon is just out ahead. 24MP is disappointing for medium format though so ālong side/short sideā stitching is needed depending upon the application but the same lenses can be used to very good effect.
For that lens ā2xā means 0.5x for camera scanning, a little confusing. For ā2xā it has to be reversed.
Could well be, what you say about optimized lenses. And at the prices for high quality lenses I too would not necessarily like to buy two of them; but I suppose it also depends on the lens. With my Schneider APO Macro-Digitar Iām not noticing differences of grain rendition between 35mm and 6x9, but thatās perhaps because the lens is designed for a wide range of magnification ratios, and used on a bellows the media can fill the frame regardless of media size for anything between 35mm and at least 4 x 5 inches (I havenāt tested beyond that as I donāt have bigger originals). Unfortunately these kind of lenses are hard to find these days.
And yes, I agree with your sentiment about the end-result of all this.
Having such a lens on a bellows with micrometer movement would really be something. It would probably smoke the scans I take with a 7Artisan macro lens I got for about 150 ā¬. Itās a gem of a lens and has decent MTF and a very long focus throw.
A while ago, I decided to stick to what I have and do with it. All I have to do now is to scan those negativesā¦which is really boring.
