Home · Register · Join Upload & Sell


  Previous versions of Peter Figen's message #16436443 « Digitizing Film »


Peter Figen
Upload & Sell: On
Re: Digitizing Film

coralnut wrote:
Well, I guess I'm lucky enough to be one of those guys who knows what he doesn't know. That is to say, I don't know everything that I don't know, but I do know that there are specific gaps in my knowledge and I'm trying to fill them in. I'm trying to avoid being doomed to endure a lengthy digitization project, only to find out that I need to start over with new equipment to get it right. I'm willing to do what it takes to get good results the first time, so I'm still in the mode of asking a lot of questions to try to avoid making misteaks.

Yeah, you only know what you know until you learn what you didn't know and then you go, oh crap, I wish I knew it then.

My preliminary tests are being done on a full-frame DSLR with a "high end" (for the format) 1:1 macro lens (Nikon 200mm/f4 Micro). I like to think of two points of weakness in any digitizing camera system: pixel resolution and lens performance.

A good starting place.

There's no question that results would be better with a dedicated 1:1 reproduction lens that is optimized for transparency duplication and internegative preparation, with the reference standard being the Rodenstock Apo-Rodagon-D. Granted, it's not as flexible as the Rodenstocks that are designed for product photography, but being optimized strictly for reproduction, the Apo-Rodagon-D seems like the better choice. It certainly seems more economical than a Digaron-S that has other capabilities beyond negative reproduction.

The problem I've found with most enlarging lenses is that used for copying film, they're just not that well optimized for that distance and while they might be great enlarging lenses, they're not really the best for digital scanning. And looking up a 100mm f/4 Digaron-S, it's only optimized from one tenth life size to infinity, from Rodenstock's website. It'd be great for copying 8x10 film.

The lens that really rocks this whole application (and there are some others as well) is the Rodenstock 105mm f/5.6 Digaron APO Macro FLOAT lens where you can dial in the floating element anywhere from one third life size to three times life size. It's the being able to set that floating element to the exact reproduction ratio that takes this lens above and beyond. The Contax 645 120mm f/4 Macro with its automatically set floating element is in the same league but limited to a 1:1 ratio and where those two lenses, the Contax and the Float lens overlap, it's very very hard to tell the difference between the two, but for me, shooting with the GFX, I need to get to just under 1.4:1 ratio to fill the GFX frame with a 35mm piece of film, so the Rodenstock is preferred. And to address something further down, that Rodenstock lens cannot be mounted directly to the GFX or any other camera as there is no integral focusing mount in the lens. You have to mount it to some sort of focusing bellows system like a Cambo or Novoflex. I've been using the Novoflex tilt-shift bellows for a variety of macro images as well as the digital scanning. I'm sure there are other solutions as well.

Looking at recording cameras, Peter makes the point that pixel dimensions (small pixel sensors) are important in maximizing the resolution of your captures. He's using the latest era 102MP medium format cameras.

One of the things that has kept me from taking the plunge into digital medium format has been that prior to the development of the ~100MP cameras by Fuji and Hasselblad, there just wasn't a convincing reason for me to move from Full Frame (FF) to Medium Format (MF). Let's compare some numbers:

In the FF realm, the current generation of Nikon 45.7MP bodies have 8256 horizontal pixels covering 35.9mm of 35mm film frame, resulting in 230 pix/mm at a pixel size of 4.35um (5200 dpi). Crunching numbers, that leads to a Diffraction Limited Aperture (DLA) of 7.1.

The Sony a7IV has 61.2MP, 9504 horizontal pixels covering 35.9mm, resulting in 265 pix/mm at a size of 3.78um (6700 dpi). Calculated DLA is 6.2.

The Fuji GFX50R has 51.4MP, 8256 horizontal pixels covering 43.8mm, 188 pix/mm at a size of 5.31um (4800 dpi). Calculated DLA is 8.7.

Looking at these numbers, I don't see a compelling reason to move from a Nikon or Sony sensor to the Fuji 50MP system ... unless you can mount a significantly better lens onto the system.

In the world of ~100MP both MF Hasselblad and Fuji have come out with 102MP cameras. The Fuji GFX100 has 11,648 horizontal pixels covering 43.8mm, 266 pix/mm at a size of 3.76 um/pix (6750 dpi). Calculated DLA is 6.2. These numbers seem very close to the Sony a7IV.

Well, that's because it's the same basic chip. When you shoot in 35mm crop mode on the GFX you get 61mp with the exact same pixel dimensions as the Sony.

The change in DLA that goes hand-in-hand with MP count makes it clear that lens quality is becoming more important. As MP increases f/8 is no longer "optimal" and we have to change our reference to something more like f/5.6; if your sensor dictates a DLA in the range of 6.2 to 7.4 then it's important to have a lens that is optimized by f/5.6.

Don't forget that the apertures you've listed for the diffraction onset are the number prior to factoring in bellows factor, so that your 200mm f/4 Nikkor is already an effective f/8 when it's wide open at 1:1 ratio and you might have to stop down a stop just to pull the corners in.

Simply looking at the numbers (I will be embarrassed if my math is wrong), I'm trying to wrap my mind around exactly where the big step-up in quality occurs with MF vs FF. (Sony certainly has been trying to close that gap.)

I used to wonder the same thing until I did my first shot on the GFX with that crazy ass Rodenstock lens and then I knew, although I probably didn't know exactly why but I think it's a combination of a larger chip that might afford smoother gradations because theyr'e happening over a larger area in conjunction with the availability of some very superb lenses like the Rodenstock and Contax I've already mentioned but for images where you only need half life sized, the Canon 135mm f/s-e is out of this world as is the Zeiss 100mm f/2 Milvus. And you can always adapt any 35mm to the back of the Novoflex, which is what I did prior to getting the Fuji.

Is the biggest advantage that MF offers over FF attributable to the lenses? I'm wondering how the Rodenstock lenses mount on the Fuji bodies and if it's even feasible/worthwhile to adapt them to FF bodies like the Sony or Nikon Z.

After doing all of this math, it's evident that none of the cameras will beat the 8000 dpi drum scan. The drum scanner still reigns supreme.

For as long as we can keep those scanners going with parts and service starting to more scarce and more expensive at the same time, there will be a time when they become giant door stops. The setup I'm using with the GFX and the Rodenstock Float lens is very very close to 8000 on the drum scanner but without the flexibility of being able to set the scanning aperture independently from the scanning resolution, which I know is a mouthful that probably doesn't make any sense now but we can get in to the technical explanation later.

Jan 06, 2024 at 07:56 PM

  Previous versions of Peter Figen's message #16436443 « Digitizing Film »


This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.