> but no LF lens performs at 429 lp/mm at f/3.5.
Absolutely correct.... and if you notice, I tend to only use the apertures the LF lenses are optimized for, at those apertures, lenses are much closer to diffraction limited. For LF lenses, the reason you see f3.5 - f5.6 is strictly for bright viewing of the GG, NOT for usage. The ground glass resolution is so low, that even the worst aberated lens will produce a nice clean bright image on the gg. As you stop down on LF lenses, specially after f22, the resolution threshold for the lens to meet becomes much less, hence why they become diffraction limited much easier vs. at wider apt. lenses.... that doesn't make them deliver more aerial resolution than lower f stops, but it delivers a higher % of max. diffraction limited value.
In some cases though, the lens is only optimized in a very tight f stop range....this is why its valuable to read the MTF curves for a lens, as it gives you an idea of where its sweet spots are at.... then, you can more accurately predict the recorded resolution. If you have no MTF curves, a simple film test will advise you.... Anyway, as you can see, this is where you have to apply a little common sense to the math.
But lets not forget the obvious, these issues only degrade the MP's issues you provided above, not improve them... driving the "real" MP's even lower than calculated values, hence why I always "max. possible". In all fairness, most modern LF lenses will reach their diffraction lmtd values, as well as MF lenses, specially Mamiya 7 lenses.
> Bryan Guyer says that maximum film-resolution increases as f-stops are increased:
ONLY if the lens is highly aberrated at the lower f stops, which is sometimes the case, specially in LF. But many people confuse this with some type of phenomena....instead, the lens was NEVER designed to be used at those lower f stops, they are highly aberrated and only exist for ground glass viewing. This is not the case with Digitar or very high end 35mm lenses....
> Best lens resolution is achieved when shooting wide open....
this is not always true....its a result of people getting carried away with 1500 / fstop. This can be tested quite easily by placing a ~ 3mm focal length astronomy eyepiece behind a lens, and view targets, you can easily see the aerial resolution of the lens.... when on film resolution is lower at wider f stops, its due to the aerial resolution being less, due to a highly aberrated aerial resolution at those wide apertures. This varies greatly with the lens type and format size....
Also, keep in mind, I think the lens tests you refer to are from Perez's site? If so, these are for B&W film, NOT color film....you are aware of that? I trust Chris's work, he has been testing lenses for years. But if you re run the numbers for color film, you will not see this level of recorded resolution. I did not check check your math on your examples, but it seems you have the hang of this. If you shoot color, you may want to reverse engineer the calculations for color film, 50 - 60 lp/mm.... I am glad someone has embraced these fundamentals !
Your are correct though, many lenses are optimized quite well for one focus distance and one f stop. This is where a vintage lens can still perform well. It was always easier to design a lens to perform ONE task very well.... the complexity of lens design comes into play when the designer wants a lens to do many tasks, and expect great performance in all areas, i.e. different f stops, different focus distances, varying fl's for zooms, etc. This is why you see some 35mm lenses with 15 - 20 elements these days, often with a few aspherical elements as well.... LF lenses are much more limited in their application, hence why you need so many of them :-(
>* Measured aerial resolution for the Kodak Ektar 207/7.7 at f/11 was 103 averaged from 0 to 10 degrees from axis, as done by Larry and Linda Whatley; but 64 lpmm as done by Chris Perez (ctr + mdl).
At f11 103 lp/mm is close to diffraction limited for the aerial resolution.... are you referring to Perez test results of 64 lp/mm on film, or aerial rez? I think Perez only shows on film resolutions, right? If so, 64 lp/mm to film, using 120 lp/mm MTF for the film, and the numbers work out perfect again... i.e. 1/R = 64 lp/mm....
> By contrast, a Phase One P45 39-megapixel P45 back is 7216 x 5412 pixels. That corresponds to a 57" x 43" print at 5 lpmm. Of course, this is a favourable estimate, as it assumes any lens on the P45, no matter the f-stop, will actually resolve the full 39 megapixels
NO! Not sure if you were writing this, or quoting someone else..... see my other posts on 1/R.... at best, you can resolve about 1/2 the native pixel count.... Digital capture is NOT immune to 1/R. The other thread has this in more detail, I can't keep re-writing this stuff....
> The equivalent megapixels of a 4x5 range from 5 to 83 megapixels.
Just keep in mind, you have color neg. on the low side, and high resolving Tmax B&W on the other side of the equation.... then you have GREAT lenses at low f stops, such as the SSXL's, 110, 150mm, and you have vintage lenses that were never even designed to be used for enlargements....they imaged directly onto paper inserted in the camera, before the days of film. This is why the numbers can be all over the map with LF, but its a simple formula, just build a model and its all quite predictable. But knowing a bit about the lens is critical, as many of your examples are for vintage lenses.... often with NO MTF data, so testing is your only means of trying to back track the lenses performance characteristics....
> If you need to stop down to f/64, you might as well use a DSLR; enormous hit.
You are really getting the hang of this..... Kudos.... One tip I mentioned previously.... lens tilt "can" be a miracle focal cure....but only when very little tilt is used.... as the MTF of a lens is based on parallel subject plane and image plane. When you tilt the lens, the two planes are no longer parallel.... now, the lenses aberations increase greatly... the more you tilt, the worse it becomes. As a general rule, if you can keep tilt under 8 degrees, and save 2 f stops (or more) by tilting, you will surely improve IQ....
> Given that the maximum resolution one gets from 4x5 under optimal conditions is 80 megapixels (maybe 150 with black-and-white), how deep do you think one should scan the film? I.e., to get the full resolution out of it, how much does one have to over-scan it?
This is where the quality of the scanner comes into play. Because now you are re-imaging an existing image. Also, the answer is highly dependent on the spatial frequency recorded on the film. The lower the spatial frequency, the more efficient the scanner, remember, the scanner too is confronted with 1/R.... as the process of scanning also involves a subject plane, a lens, and an image plane.
I will try to answer your question with two extremes.... with very low spatial frequency on the film.... and a very good scanner (high end flat bed or a drum), you can probably scan at 1.2x the highest frequency on the film. At higher frequencies, maybe up to 1.5x....
lower end scanners, you can double these values.
This is the real value of high end scanners - smaller files to work with. This can all be easily tested, by shooting targets, and scanning the film targets, and seeing how deep you must scan to resolve what you know is recorded on the film.... which you can clearly see with a good loupe and bright light box. Be sure the loupe has sufficient magnification, so it does not become the limiting factor.... Often in the 20 - 80x range. Make sense?
Reply With Quote
Originally Posted by dh003i
Bookmarks