MF sensor area, crop factor and lens AOV.

[Theodoros]

Its being a lot of talk on how one can compare the AOV of an MF lens with respect to FF 35mm, there are two ways of doing this: 1. Compare the diameters of the image circles and come up with one crop factor 2. Compare independently the horizontal and the vertical size differences and come up with two values.
Out of the two ways (both correct), I prefer the second because usually photographers think of the direct with respect to horizon angle that will cover their subject, either when they shoot in "landscape" or in "portrait" mode.
For example, my Contax 645 has a frame that is of 56x41.5mm when used with film, then the difference from 35mm film (24x36mm) is of 73% on the height of the frame and of 55% on the width of it. Hence, my 35mm Zeiss, has an AOV that will present the same AOV as a 22.5mm focal length horizontally (22.5 x 1.55 = 35aprox), but the height of the frame will have the same AOV as a 20mm (20 x 1.73 = 35aprox.)…
If one uses an MF sensor of 36x48mm in size, the above values turn out to be of 50% in height and 33.33% in width, hence my 35mm will now show the equivalent AOV of 23.7mm in height and (about) 26mm in width…

But lets come down to crop factor of sensors with respect to FF MF… To start with, FF sensors still have some area left to equal film area, this is about 5% horizontally and only about 2% vertically, however, sensors of lesser size are declared with "strange" crop factors… for example the 33x44mm sensors are said (from makers) to be of 1.3x crop, which is more than they are!!! They are only f 1.2x when compared to "FF"(MF) sensors (44x1.2=53aprox) and they are of about 1.26x if compared with the full 56mm that film would cover on my Contax…

OTOH, 36x48mm sensors are said to be of 1.1x crop, which is not true either… they are of 1.1x crop when compared with "FF"(MF) sensors that weren't around when the factor was applied and they are of about 1.17x (!!!) when compared to what film width is! That's not fare for 33X44mm sensors, is it? If one thinks about it, there is not much AOV loss between 33x44 and 36x48mm sensors, if we apply the "real" crop factors on my 35mm when used on Contax, with 36x48mm sensor it shows 26mm equivalent 35mm AOV, while it shows same as a 28mm on a 35mm DSLR  if a 33x44mm sensor is used.
 Another point of discussion is what sensor size is ideal to use with digital MF… I use a 37x49mm image sensor and find this ideal… The reason is that I don't have to worry much about corner sharpness at wide apertures and the additional 2mm that I loose from my 35mm lens (23mm vs. 25mm FF DSLR equivalent values) than if I used a FF (MF) sensor are not important to consider. Why makers "lie" on the crop factors of the image areas they use? Is this to their benefit? Is more than 5cm of sensor width important for modern technology? (especially now that WA lenses have gone down to 24mm) …Your thoughts?

[Doug Peterson]

Or you can use a comparison tool that jus tells you the answer:
https://digitaltransitions.com/page/tech-camera-visualizers

"1.1 crop" is in comparison to the actual exposed area of 120 film in a typical 645 camera, not compare to the width of the film - the full width was never exposed.

[Paul2660]

Doug,  great tool a picture is worth a lot of words in this case.  
As a big user of wides an easy decision.

Paul

[synn]

That is indeed a great tool. It has helped me a lot while planning my MF purchase.

When I go back and forth between my Bronica ETRSi viewfinder ("Full frame" 645) and my DF+ with a mask for the Credo 40 (1.3x crop), I do not see a HUGE difference.  However, when I go from either of them to 35mm full frame, the difference is very apparent. The difference between a 35mm full frame viewfinder and an APS-C viewfinder is also much more dignificant that that of the Bronica and the DF+ with 1.3 crop.

I do feel that 44x33 gets an unfair rap from some quarters. Apart from the usage of extreme wide angles; it's quite a viable solution and is much closer to full frame 645 than to 35mm full frame in real life.

[Telecaster]

Not being a big fan of really wide lenses (other people do great work with 'em...me, not so much) I love the FOV I get with my 35mm on the Pentax 645D (44x33mm sensor). If I want wider coverage I'm happy switching to the 55mm and doing a stitched pano...that yields a different look that works better for the way I see things.

-Dave-

[FMueller]

I have a P40+ that I use on a Hasselblad 503 and on a tech cam. I was initially concerned about "crop factor" but found it to be more a theoretical limitation than a practical one. A benefit of the tech cam is the ease of stitching when one feels compelled to go for the wide shot.

I have no experience with the D800 but I do own an A7r with the the Sony 35 and 55. P40+ is significantly better image quality, as expected.

My point is that the 33x44 sensor is not just slightly larger than 35mm "full frame", it is a first rate sensor that mates easily to some of the finest lenses in the world. Hasselblad V series  and lenses can be had very inexpensively and many of the Schneider tech cam lenses have been deemed problematic for the 80mp backs can be found very inexpensive now also. I don't know exactly WHY, but those old crop sensor MFDB's are still far ahead of any Canon, Nikon, Sony DSLR/mirrorless offering.

[Theodoros]

Or you can use a comparison tool that jus tells you the answer:
https://digitaltransitions.com/page/tech-camera-visualizers

"1.1 crop" is in comparison to the actual exposed area of 120 film in a typical 645 camera, not compare to the width of the film - the full width was never exposed.

But Doug, the actual exposed film area on my Contax is 56mm, now 48x1.1=52.8 that's 3.2mm less than 56mm , which means that the real crop factor is 1.16x… But more than that, if we take Dalsa's FF sensors (about  53.8 in width) and consider those as FF "standard" (no crop factor)… then, 48X36 is of 1.12x crop, Kodak's older 48.9X36.7 sensors are of exactly 1.1x crop and 33x44mm is of 1.2x crop, Hence, there is not much difference in size, they all are between 1.1 (for 37x49) and 1.2 (for 33x44) crop factor with respect to Dalsa's FF sensors! OTOH, it is a mystery where this 1.3x crop for 33x44 sensors came from… 1.3X exceeds even film image area.  So, while the makers are optimistic  with 48x36, 49x37 and FF sensors, they are pessimistic  with 33x44mm sensors… why is that?

[torger]

Once Phase One had sensors for full-frame 645 (almost, 56x42 would be more fair full-frame), the 49x37 was a bit too close to 54x41, so to differentiate the entry level from high end more they introduced the 44x33 crop and dropped the 49x37.

I like well-balanced systems. This means that the sensor size is adapted to the size of the lens image circle, and that the resolving power of the lens and sensor is in harmony (ie not "out-resolving" the lens too much, but not a zillion elements 2 kg over-corrected lens either), and the resolution is high but not so high that it's near impossible to make any sane use out of it other than repro, cartography or scenery-from-a-mountain-top type of shots.

For a MF DSLR full-frame is of course what you'd want in a well-balanced system.

For a tech cam with the current lens offers from Schneider and some Rodenstock (often 90mm image circle) I think the 48x36 / 49x37 with ~6um pixels ending in the range 50-60 megapixels is the ultimate format landscape photographer format. A wide-angular response CMOS of this size would be a dream. You could continue with the LF-style design, symmetric small and light. Great look, balanced performance, great value too. And you can have lots of lenses in your backpack as they're small and light.

I just hate the idea of having to stitch because my sensor is cropped so I don't have enough wide angle, or having to perspective correct in post because I don't have enough shift range, or having to distortion correct in post because lens is too much retrofocus. I don't think that is what tech cam photography should be about.

I don't like to crop a lot either, although one solution is to have only a few extreme high resolution lenses and back and then crop to the desired size, I think that takes away much of the enjoyment of photography in the field. For the same purpose I don't like stitching, being able to see and expose the finished image directly on the sensor out in the field is what gives me the most satisfaction. Using a tech camera is for me more about loving the shooting process than image quality, although image quality is important too of course.

[Theodoros]

The thing is that if 60 & 80mp Dalsa sensors are NO crop, then 37x49 is 1.1x crop, 36x48 is 1.12x crop and 33x44mm is 1.2x… not 1.3x as P1 claims.

[EricWHiss]

Or you can use a comparison tool that jus tells you the answer:
https://digitaltransitions.com/page/tech-camera-visualizers

"1.1 crop" is in comparison to the actual exposed area of 120 film in a typical 645 camera, not compare to the width of the film - the full width was never exposed.

Doug,
That is a really great tool!  Thanks for that!
Eric

[Bart_van_der_Wolf]

The thing is that if 60 & 80mp Dalsa sensors are NO crop, then 37x49 is 1.1x crop, 36x48 is 1.12x crop and 33x44mm is 1.2x… not 1.3x as P1 claims.

Who made (the Dalsa) sensor dimensions to be the reference?  The origin is film of nominal 6x4.5 (cm) which is effectively turned out to be a 56 × 41.5 mm exposure area on film (e.g. Pentax and here as introduced in 1984), hence 56/44=1.27 , and  41.5/33 =1.26. To call it 1.3x is close, but it's slightly better in practice.

That your Contax 645 (introduced 15 years after the original Pentax) potentially cheated you out of a few extra millimeters (which helped with film flatness) is not relevant for other cameras.

Besides being just an approximation for focal length adjustments, a crop factor is relatively unimportant (and inaccurate) for the determination of the actual Field of View. The actual sensor dimensions and the focal length will allow to approximate the actual angle of view (at infinity focus). The FOV also changes when focusing closer than infinity, and if you want to be really accurate then the lens design's pupil factor also needs to be incorporated in the equation (see equation 3.).

I have that FOV calculation incorporated in my (DOF) Output Quality Planning tool, to which I've also added tentative sensor dimension data for the upcoming Hasselblad H5D-50c (645 CMOS). The PhaseOne IQ250 was already added earlier into the pre-selectable camera sensor database.

Cheers,
Bart

[Theodoros]

Who made (the Dalsa) sensor dimensions to be the reference?  The origin is film of nominal 6x4.5 (cm) which is effectively turned out to be a 56 × 41.5 mm exposure area on film (e.g. Pentax and here as introduced in 1984), hence 56/44=1.27 , and  41.5/33 =1.26. To call it 1.3x is close, but it's slightly better in practice.

That your Contax 645 (introduced 15 years after the original Pentax) potentially cheated you out of a few extra millimeters (which helped with film flatness) is not relevant for other cameras.

Besides being just an approximation for focal length adjustments, a crop factor is relatively unimportant (and inaccurate) for the determination of the actual Field of View. The actual sensor dimensions and the focal length will allow to approximate the actual angle of view (at infinity focus). The FOV also changes when focusing closer than infinity, and if you want to be really accurate then the lens design's pupil factor also needs to be incorporated in the equation (see equation 3.).

I have that FOV calculation incorporated in my (DOF) Output Quality Planning tool, to which I've also added tentative sensor dimension data for the upcoming Hasselblad H5D-50c (645 CMOS). The PhaseOne IQ250 was already added earlier into the pre-selectable camera sensor database.

Cheers,
Bart

You don't read past posts, do you Bart? My Contax is 41.5x56 as it should, on this neither 60 or 80MP sensors are "FF" (no crop), nor 36x48 or 37x49 are 1.1x… they are more close to 1.2x, but still 33x44 is less than 1.3x… 

[Bart_van_der_Wolf]

You don't read past posts, do you Bart? My Contax is 41.5x56 as it should, on this neither 60 or 80MP sensors are "FF" (no crop), nor 36x48 or 37x49 are 1.1x… they are more close to 1.2x, but still 33x44 is less than 1.3x… 

Hi,

I tried following your logic, but still do not see how you could get an accurate FOV from those cropfactors. If your point is that the crop-factors mentioned are not accurate either, I'd agree, but they are not as relevant anyway, so some liberal rounding is not a biggie.

Cheers,
Bart

[Theodoros]

Hi,

I tried following your logic, but still do not see how you could get an accurate FOV from those cropfactors. If your point is that the crop-factors mentioned are not accurate either, I'd agree, but they are not as relevant anyway, so some liberal rounding is not a biggie.

Cheers,
Bart

I just wonder why makers are so unfair with 33X44mm sensors… they present them as being "tiny" while they are only of 1.1 crop from 37x49 sensors and 1.2 from "FF" sensors…  As I state in the O/P, I like crop sensors better than "FF" ones… (especially 37x49, I think is ideal). Reason is that I don't have to worry about "usual corner issues" when at wide apertures.