[Rondec]

I thought the only thing had to do with if you need to stop down to achieve a certain depth of field. So, 50mm f4 and 1/100 second at iso 200 will give the same exposure on a K3, K1, and 645z, but obviously the images will be very different. Depth of field will be really shallow on medium format and you'll need to stop down more meaning either you increase your iso or end up using a tripod.

[k1man]

While reading about larger camera format, this article The Light Eater: The Biggest Challenge of Large Format Portraiture | Fstoppers caught my attention. I though of getting some clarification and feedback from people having experience with medium format: how practical is the use of a medium format format compared to 24x36 ? Is it truly difficult to use for common use cases? Thank for sharing your thought from experience. I'd be very interested.

Hi biz, I used 2 645s and a 6x7ii exclusively for about 20 years and no the light loss is not daunting you learn how to use what you have, dunno if this makes sense, a few too many beers, you can see a few of my pics on FLICKR stevensonian1, my passion was corrida not everybodies cup of tea but all shot with 645s 1st generation, Ian

---------- Post added 03-18-19 at 03:28 AM ----------

It has been called "ideal" because the aspect ratio translates to an 8x10" enlargement with minimal crop.


Steve

Yes even though many shots were taken with Blads all were cropped to fit a magazine cover, ideal format 6x7, many regrets selling all of mine, Ian

---------- Post added 03-18-19 at 04:03 AM ----------

I shoot a LOT of medium format film. More than my small format digital. It's practical enough for a lot of things. It's practicality all depends on a person's tolerance level of what becomes inconvenient and if they're so sensitive they need the touchy-feely aspect of a camera to give them inspiration to take pictures or not.

I loved my medium format cameras as you most probably do, but getting 120 or 220 film was next to impossible here so I went digital, Ian

[stevebrot]

I've noticed that the hyper-focal distance increased roughly by the square of the focal length. In practice, we see that compact cameras can easily produce all in focus images while they struggle with making bokek. It becomes easier to produce decent amounts of out of focus with apsc or 24x36 cameras, while the lens aperture should be closed to get all in focus images. Now I assume that formats larger than 24x36 require even higher f numbers with increasing effects of diffraction. That leads me to think that if I want all in focus images without drop of resolution (diffraction), the larger the "sensor" the wider angle the lens should be [that was my point]. The other possibility to improve sharpness with larger format without stopping down the lens is to tilt the lens to reduce the gap between the plane of focus at the scene and the sensor plane (Scheimpflug principle, camera movements: what I learned).

You may be over-thinking this a little, but you are on the right track. A key thing to remember is that the degree of enlargement required for display at normal viewing distance drives the DOF equation with absolute aperture being other variable; similarly, factors such as as diffraction and lens sharpness become less of a concern as the format size increases. A good example might be an 8x10" contact print vs. the same subject and framing from a 35mm negative.

On a related note, it is not well-appreciated that medium and large format lenses are often not particularly sharp. They don't have to be. More important is contrast and freedom from distortion and flatness of field.


Steve

[tuco]

...
On a related note, it is not well-appreciated that medium and large format lenses are often not particularly sharp. They don't have to be. More important is contrast and freedom from distortion and flatness of field.


Steve

At a museum, I saw a 20x24 print dated 1915 from an 8x10 view camera of a street scene. And I could see buttons and small detail on nearby people's clothing. While the film's rendering of tonality was rather poor by today's standards, the resolution was amazing. 1915!

[photoptimist]

Thanks for sharing this practical aspect, I had no idea about that (learning here), I appreciate.


I've noticed that the hyper-focal distance increased roughly by the square of the focal length. In practice, we see that compact cameras can easily produce all in focus images while they struggle with making bokek. It becomes easier to produce decent amounts of out of focus with apsc or 24x36 cameras, while the lens aperture should be closed to get all in focus images. Now I assume that formats larger than 24x36 require even higher f numbers with increasing effects of diffraction. That leads me to think that if I want all in focus images without drop of resolution (diffraction), the larger the "sensor" the wider angle the lens should be [that was my point]. The other possibility to improve sharpness with larger format without stopping down the lens is to tilt the lens to reduce the gap between the plane of focus at the scene and the sensor plane (Scheimpflug principle, camera movements: what I learned).

What you say is true but neglects the relationship between camera format and acceptable levels of blur from diffraction and DoF.

The physical diameter of the diffraction blur is proportional to the f-number. But obviously a blur of a given size will have a much larger impact on the print resolution of a small-sensor image than on a large-sensor one. Thus, larger format cameras can stop down to narrower apertures than can small format cameras. Thus, if f/11 on a 24x36 camera represents the narrowest aperture before diffraction impacts the print's resolution, then a 4x5 camera could be operated at perhaps f/32 to f/45 before diffraction impacts print resolution.

The hyperfocal distance does depend on f^2 (and f is proportionally larger for large format cameras) but hyperfocal distance also has an inverse dependence on the aperture and the circle of confusion. Again, a large format image can have a proportionally larger circle of confusion before it become noticeable in a print of a given size. And the larger format lens can be operated at a proportionally narrower aperture (proportionally larger numerical aperture) than could a smaller format lens.

The upshot is that a wide angle lens operated at the limit of diffraction will have the same hyperfocal distance across all formats. A 24 mm lens @ f/11 on a 24x36mm camera has the same hyperfocal distance and diffraction behavior as a 4" = 100mm lens @ f/45 on a 4x5 camera.

[biz-engineer]

The physical diameter of the diffraction blur is proportional to the f-number. But obviously a blur of a given size will have a much larger impact on the print resolution of a small-sensor image than on a large-sensor one. Thus, larger format cameras can stop down to narrower apertures than can small format cameras. Thus, if f/11 on a 24x36 camera represents the narrowest aperture before diffraction impacts the print's resolution, then a 4x5 camera could be operated at perhaps f/32 to f/45 before diffraction impacts print resolution.

There is a difference between the sharp rolls-off of nyquist pixels, and the soft effect of diffraction (slow reduction of detail contrast), won't show the same way on a print. With digital, we've used to zoom in to 100% expecting to see sharpness, I've come to realize that pixel-peeping is not necessarily a correct way to evaluate image quality.

---------- Post added 19-03-19 at 00:13 ----------

The upshot is that a wide angle lens operated at the limit of diffraction will have the same hyperfocal distance across all formats. A 24 mm lens @ f/11 on a 24x36mm camera has the same hyperfocal distance and diffraction behavior as a 4" = 100mm lens @ f/45 on a 4x5 camera.

My problem is that if that was true, we wouldn't see a difference of image quality of small formats vs large formats, but practically, images from the larger formats generally look better. So where do I get it wrong with the theory?

[photoptimist]

There is a difference between the sharp rolls-off of nyquist pixels, and the soft effect of diffraction (slow reduction of detail contrast), won't show the same way on a print. With digital, we've used to zoom in to 100% expecting to see sharpness, I've come to realize that pixel-peeping is not necessarily a correct way to evaluate image quality.

---------- Post added 19-03-19 at 00:13 ----------

My problem is that if that was true, we wouldn't see a difference of image quality of small formats vs large formats, but practically, images from the larger formats generally look better. So where do I get it wrong with the theory?

Although a 35mm format image @f/11 has similar diffraction and DoF blur as a 4x5 @ f/45, the 4x5 used at f/11 will have much higher resolution than a 35mm format image @f/11 and probably be better than a 35mm format image @f/2.8. And the 4x5 camera wide open @f/5.6 will probably be sharper (but have the same shallow DoF) as a 35mm camera wide open at f/1.4.

The format quality issue is also accentuated by film grain issues. A 4x5 image shot on ISO 100 film will look a lot better (smoother tonality) than the same shot taken with a 35mm camera using the same ISO 100 film. Or, in the digital world, a 36 MPix sensor on 4x5 format (19 µm pixels) would have better DR and IQ than a 36 MPix sensor on 35mm format (4.8 µm pixels).

Thus, the quality differences come from having a much larger photosensitive surface and a higher numerical aperture. The design of a decent f/5.6 lens for LF is much easier than a decent f/1.4 lens for 35mm format.

P.S. Getting that better tonality and quality does require more total light which was the point of that article.

[biz-engineer]

Thus, the quality differences come from having a much larger photosensitive surface and a higher numerical aperture. The design of a decent f/5.6 lens for LF is much easier than a decent f/1.4 lens for 35mm format.

Ok. The overall increase of image quality comes at the cost of usability (tripod, critical focus etc..), the larger the format the more time is takes to get the setup right.

[photoptimist]

Ok. The overall increase of image quality comes at the cost of usability (tripod, critical focus etc..), the larger the format the more time is takes to get the setup right.

Exactly!

And there are some that argue that large format cameras don't really offer substantially higher photographic quality than do small format cameras but that large format camera photographers are both more experienced and take more time to get great images.

It's really a combination of both gear and user. A 150 mm f/5.6 lens on 4x5 almost certainly out-resolves (at the print level) a 40 or 50 mm f/1.4 lens on 35 mm format. I've seen work done with a 8x20 banquet camera with insane levels of detail in huge prints. No 35 mm camera could touch it although multi-shot panorama methods could replicate the results.

But, of course, the 35mm gear is so much easier to point, focus, and shoot that it gets more images per hour of photography.

[kaseki]

There is a difference between the sharp rolls-off of nyquist pixels, and the soft effect of diffraction (slow reduction of detail contrast), won't show the same way on a print. With digital, we've used to zoom in to 100% expecting to see sharpness, I've come to realize that pixel-peeping is not necessarily a correct way to evaluate image quality.

In a digital camera such as the 645Z, there are R, G, and B detectors (of unequal count). I assume that Pentax uses some scheme to interpolate the RGB values of each "pixel" from the detected charges of the nearby single color detectors. One might expect that the algorithm used would soften the sinc function that would represent the MTF if only detector Nyquist effects were considered. Does anyone reading this have a reference for the actual MTF that some digital camera achieves in its "raw" output imagery vs. the Nyquist assumption?

[photoptimist]

In a digital camera such as the 645Z, there are R, G, and B detectors (of unequal count). I assume that Pentax uses some scheme to interpolate the RGB values of each "pixel" from the detected charges of the nearby single color detectors. One might expect that the algorithm used would soften the sinc function that would represent the MTF if only detector Nyquist effects were considered. Does anyone reading this have a reference for the actual MTF that some digital camera achieves in its "raw" output imagery vs. the Nyquist assumption?

It's easy enough to compute in a simulation although adding in the effects of diffraction and the optical low-pass filter in many camera makes it harder.

But that MTF underestimates the what the user will see because any decent RAW developer is going to make clever use of the green channel signal to reconstruct the red and blue channel signals at a resolution higher than the native red and blue channels.

[biz-engineer]

No 35 mm camera could touch it although multi-shot panorama methods could replicate the results.

Panorama isn't quite the same as a larger format, a rectilinear projection stretches edges for perspective correction, resolution isn't the same across the frame, stitching also run into depth of field vs diffraction trade-off.

[photoptimist]

Panorama isn't quite the same as a larger format, a rectilinear projection stretches edges for perspective correction, resolution isn't the same across the frame, stitching also run into depth of field vs diffraction trade-off.

That's technically true but If one has enough image tile data of sufficient resolution, then one can replicate a rectilinear or any other projection.

From what I've seen, the better quality panorama stitchers allow one to pick the projection (rectilinear, cylindrical, Mercator, spherical, or little planet) although only the rectilinear replicates what an LF camera typically provides.

[kaseki]

It's easy enough to compute in a simulation although adding in the effects of diffraction and the optical low-pass filter in many camera makes it harder.

But that MTF underestimates the what the user will see because any decent RAW developer is going to make clever use of the green channel signal to reconstruct the red and blue channel signals at a resolution higher than the native red and blue channels.

I think you have supported my point -- that the clever developer is going to make the apparent MTF after the raw processing algorithms different than the Nyquist MTFs for the three color arrays (with different spatial phases), and possibly more like a film MTF. In any case, someone in the camera design business would have to have performed the simulation accounting for his company's secret algorithm sauce (or at least done the MTF measurements) and written an article for JOSA or some similar publication.

[photoptimist]

I think you have supported my point -- that the clever developer is going to make the apparent MTF after the raw processing algorithms different than the Nyquist MTFs for the three color arrays (with different spatial phases), and possibly more like a film MTF. In any case, someone in the camera design business would have to have performed the simulation accounting for his company's secret algorithm sauce (or at least done the MTF measurements) and written an article for JOSA or some similar publication.

Quite true! And yet ...

The clever demosaicing software developer may certainly strive for a more "natural" MTF but they cannot recreate image elements that the camera failed to correctly capture. If the light from a tiny red berry happens to fall only on the green or blue pixels of the sensor, the berry totally disappears in the final image! (Digital vs Film Cameras | Photography | On Landscape). This kind of issue also causes color artifacts in high-detail black-and-white subjects.

It's also worth noting that the film's MTF can have some quirks, too, that are created by the clever film chemistry developer (the developer developer?). The chemistry of some films includes an edge-enhancement effect and the MTF in medium frequencies will be above 100%. Ironically, although human viewers judge these edge-enhancing films to be sharper than their competitors, these films tend actually have poorer high-resolution MTF values.


P.S. It's worth noting that the 645Z is not a 50 MPix camera so much as it's a 12,5 MPix red-light camera added to a 25 MPix green light camera added to 12.5 MPix blue-light camera. And it's also worth noting that film absolutely sucks on quantum efficiency compared to silicon sensors which explains why ISO 1600 film images look like gravel compared to ISO 1600 digital images.