The GetDPI Photography Forum

Great to see you here. Join our insightful photographic forum today and start tapping into a huge wealth of photographic knowledge. Completing our simple registration process will allow you to gain access to exclusive content, add your own topics and posts, share your work and connect with other members through your own private inbox! And don’t forget to say hi!

Is there a compelling reason to move to MF?

ErikKaffehr

Well-known member
Hi,

Considering the price of entry, I think it is absolutely a reasonable question to ask.

The obvious advantage of medium format is larger sensor size. Assuming equivalent technology, like Sony CMOS on both, the advantages of larger sensors are twofold: They collect more light and make lesser demands on the lenses. If you take one of those 33x43 mm sensors they have 1.68 times the area of a 24x36 mm sensor. That means they can collect more photons. If you expose a 24x36 Sony sensor fully ETTR at 100 ISO, the 33x44 sensor will give the same noise levels 168 ISO. The full frame sensor will have the same noise levels 270 ISO. This should apply pretty exactly for same generation sensors, say Sony A7rII, IQ 250 and IQ3-100MP.

The other factor is that a larger sensor makes less demands on the lens. The task of the lens is to provide a correct image. Now, if you transfer information from the subject to the sensor some information will be lost in transition. For large detail very little information will be lost. For smaller details more and more information will be lost. Check the image below, at 0 lp/mm the MTF is 100% and it drops to small values at high lp/mm. This applies to any lens ever made on this planet. Now, let's assume that you need 40 lp/mm for a large print from a 24x36 mm sensor. That would give you an MTF of 0.56 with this lens. Moving to a 33x44 mm sensor and assuming same short dimension on that print we would need 40 * 24 / 33 = 29 lp/mm. This would give around MTF = 0.7.





Now, we could assume having a sharper lens on our 24x36 mm camera, like in the image below:


In this case the green curve cuts 40 lp/mm at around 71%. The lenses here are a Zeiss Planar 100/3.5 at f/8 measured on a P45+ and a Sony 90/2.8G at f/5.6 measured on the A7rII. So using the Planar on 33x44 sensor would give about the same image quality as the Sony lens on the A7rII. (*)

Now, I don't have a 33x44 sensor, but I have a P45+ with 37x49mm sensor. If we look at the long dimensions instead the ratio between the two sensors is 1.36, so our sensor would need 29.4 lp/mm on the larger sensor and 40 lp/mm on the smaller one. So, both sensors would give the same result.

How does this look in practice?

An actual pixel crop from the P45+ looks like this:


While the Sony A7rII gives this results both processed same way and with pretty conservative (low) sharpening. Theory say they should be very close, but you are the judge:


Now, in normal processing we also have sharpening and sharpening increases noise. All factors identical the larger sensors would have a better signal noise ratio. If we assume that Signal Noise Ratio is around 240 on the A7r, the IQ 250 would have an SNR of 311, that would allow a bit more sharpening before noise is obtrusive. Older sensors like the P45+ have lesser full well capacity. FWC is not that difficult to measure, but the SNR 18% plots from DxO-mark give some indications:

The figure below shows that the A7rII has a bit higher SNR than the P45+. So sensor development has compensated for the size advantage of the P45+. DxO has not reported on latest generation MFD. The latest ones are the IQ-180 and the Leica S2. The Leica sensor is similar to P45+ sensor I happen to have. The IQ-180 is a bit better. The P45+ sensor is slightly larger, which would give it some advantage in noise and the Leica sensor probably has micro lenses that increase light capture efficiency, shifting the curve to the right, giving higher ISO.
Screen Shot 2016-03-24 at 21.31.39.jpg

The IQ-180 has better SNR ratio at base ISO than the A7rII
Screen Shot 2016-03-24 at 21.35.03.jpg

I guess that this post demonstrates a few things:

  • There is a significant advantage of size for the larger formats
  • New and improved technology may compensate for that advantage to some extent
  • It would be very hard to keep up with a high resolution full frame back with the best lenses, so the guys who say that full frame 645 rules are probably right.

Best regards
Erik









I didn't have any preordained beliefs when I asked the question.

My criteria is image quality - the image quality difference has to be obvious and apparent. I didn't see a very apparent quality difference (other than resolution, which is obvious and apparent), so given that 35mm formats are hovering around 50mp, I wanted to find out what is it about 50-60mp MF backs that attracts folks more than a 40-50mp 35mm format.

My second criteria is the unique ability of the MF format to do things that other formats cannot. In this case, it became apparent that the movements and lenses offered by technical cameras seemed to offer a unique application of the format. However, I found out later that tech cameras can accommodate 35mm format, therefore that unique application no longer existed. I wanted to find out more about that. Then it turned out that Canon TSE lenses were used on tech cameras and med format backs. That blurred the lines even more, so the case for unique application could no longer be made at all. That really confused me because from my way of thinking, MF made no practical sense at all. In fact, from what I could see, there was nothing within MF (other than resolution of over 60mp) that could possibly justify such an expense. And that's pretty much when poop started flying...
 
Last edited:

Abstraction

Well-known member
Thank you for the best possible explanation that I've read. This gives a very good baseline of theoretical expectations and a better appreciation of what to look for in comparable images.
 

jerome_m

Member
Newer lenses are better, let's check the Hasselblad HC 35/3.5:
View attachment 117342

The Otus 28/1.4 on the other hand is really great:
View attachment 117343

Here in Sweden both cost around 44000 SEK around 5000$US.

But, for the A7rII Zeiss also offers the Batis 25/2.0, that is pretty good, looking at Zeiss and Hasselblad lenses is nice because they post similar MTF data.
View attachment 117344

I would guess that a 40+ MP 24x36mm camera using either the OTUS 28/1.4 or the Batis 25/2 would outperform a 40-50MP Hasselblad using the HC 35/3.5 except for the sweet spot of the lenses, based on the MTF data. With the Batis 25/2 you could probably buy the A7rII and the lens at about the same price you pay for the HC 35/3.5.
I had the curiosity of making a simple back of the enveloppe calculation. These curves are given for a maximum of 40 cycles/mm. With cameras like the Sony A7RII or the Canon 5DR, the sensor resolves around 100-120 cycles/mm. Maybe we need curves computed at a different resolution?

Bonus question: where do the following lenses MTF transmission sit on the following curve for the two cameras cited?

 

Abstraction

Well-known member
Erik,

Would I be correct in assuming that a 35mm camera with around 20mp and 6 micron pixel pitch would have the same noise at base ISO as a MF back with about 50mp and the same 6 micron pixel pitch? If that's the case, then a 100mp back will have the same noise at base ISO as the 35mm at around 40mp given that both will have about a 4.5 micron pixel pitch. Am I correct? If I am, then my understanding is that the 100mp back will tax the lenses the same way as a 40mp 35mm camera.

So, the 100mp back will have the same noise and tax the lens the same way as a 40mp 35mm camera. That also leads me to assume that in terms of noise and thereby tonality, the lower resolution backs/cameras should theoretically be better than the higher resolution ones.
 

ErikKaffehr

Well-known member
Hi,

Yes, I am aware of it, don't know why. On the other hand, there is good data on the BClaff site: http://www.photonstophotos.net/Charts/PDR.htm

Keep in mind that BClaff uses nominal ISO while DxO-mark uses saturated based measured ISO. Vendors can have different ISO ratings.

Best regards
Erik


You may find this of interest Eric. Apparently DxO has reported on the 645Z and has since removed the ranking...?


http://www.pentaxforums.com/article...ax-645z-ranked-highest-by-dxomark.html?src=hs
 

Paratom

Well-known member
If we forget other reasons like large bright viewfinder (which I find quite important), synch time, ... and concentrate just on IQ, here are the reasons why I own and use the Leica S in regards of IQ:

-better skin tone and color
-smoother transition from focused/sharp area to background
-smoother color transition
-more 3d look (might have to do with the transition)
-more (micro)detail
-very good DR

-How much is from sensor quality? How much from sensor size? how much from lenses? How much from profiles? I dont know, but I just love the IQ.
 

hcubell

Well-known member
Erik,

Would I be correct in assuming that a 35mm camera with around 20mp and 6 micron pixel pitch would have the same noise at base ISO as a MF back with about 50mp and the same 6 micron pixel pitch? If that's the case, then a 100mp back will have the same noise at base ISO as the 35mm at around 40mp given that both will have about a 4.5 micron pixel pitch. Am I correct? If I am, then my understanding is that the 100mp back will tax the lenses the same way as a 40mp 35mm camera.

So, the 100mp back will have the same noise and tax the lens the same way as a 40mp 35mm camera. That also leads me to assume that in terms of noise and thereby tonality, the lower resolution backs/cameras should theoretically be better than the higher resolution ones.
This looks good. I think you are getting closer to determining whether medium format digital is right for you.
 

jerome_m

Member
I did not realise it at first, but it seems something went amiss with my previous samples. Apparently, the forum software does not like png files. Let me try again. First, the image from imaging-resource showing where the crops where taken:full_image.jpg

- - - Updated - - -

Now the center crop. You may need to clic to see the original resolution.

center1.jpg

- - - Updated - - -

And now, the side crop.

side1.jpg
 
It seems like there are certain people that choose MFD for highly technical reasons. They maximize DoF and sharpness and want the least possible sensor noise. Those people may indeed be satisfied with newer 35mm offerings from Sony or Nikon. Heck even m43 is capable of some amazing results at base ISO depending on the subject being photographed. All of this is totally fine, horses for courses as they say. I don't pretend to speak for all users.

However there is another camp that is more into the intangibles and I fall firmly in that camp. It's easier to explain in terms of film cameras but for example, I frequently shoot my Rolleiflex 2.8 Planar even though I KNOW a Hasselblad with a modern 80/2.8 has a better lens. Indeed the 100/3.5 is FAR better than the old Planar on my Rollei. But... my Rollei is extremely fast to focus, I enjoy using it, people have interesting reactions to it which changes their reaction to being photographed in general, and the lens renders in a way that I find beautiful. It's got just the right balance of low contrast, un-harsh sharpness, and interesting bokeh for my portrait work, and at f5.6+ it is a fantastic landscape camera. An MTF chart, or a granularity chart (we're talking film here) would not answer those questions for me. I had to be a Hasselblad user for years and borrow a Rollei to come to this conclusion.

For MFD I shoot with the 645Z. I am aware that a D810, even a D750 has high enough resolution to actually satisfy my print and even cropping requirements. My 58/1.4G has less DoF than my Pentax lenses wide open, in fact I got it because it renders very "medium format-like" for portrait shooting. But I STILL think it's worth it to shoot a 645Z. Why? Because I personally see something in the files that make them stand out for me. The compression of a 75mm but the angle-of-view of a slightly long normal is a look I really like. Shooting 4:3 instead of having to crop, and adjust crop, of every single image, is a huge time saver. (why digital cameras are locked into few aspect ratios is completely beyond me, the dominance of 3:2 cannot be explained rationally). The differences are extremely subtle. It's that extra focus fall off, that extra highlight headroom, a bit more tonal scale in the skin. A graph wouldn't explain it, nor would a web jpg. Some might argue that my clients won't know the difference either, and they're probably right! But I'm the one that has to live with this portfolio. A lot of people understate the relationship that you have to your own work. I personally want my work to satisfy me just as much as it satisfies my clients. I like being delighted when I start working on an image. It's the reason why people buy Leica cameras, or prefer Canon color of Nikon.

So if we parse everything down to the absolute minimum requirements of what is necessary to do the work, well I'd be shooting a D7100 and a series of zoom lenses, and editing on a Windows computer, but I sure as hell don't want to do that! I want my equipment and my pictures to make me really happy, and they do! That is my compelling reason to move to MFD. I hope these arguments make sense.
 

ErikKaffehr

Well-known member
Corner drop off

Hi,

Regarding corner drop off, let's look at some real images, the images below were shot on a Hasselblad Distagon 40/4FLE on my P45+ at f/11 and with a Canon 16-35/4 zoom at 24 mm and f/8.

Many 24-35 lenses drop dramatically in MTF at the extreme corners, in some cases MFD lenses behave better in the corners. The 16-24/4 suffers from astigmatism at 24 mm f/8 according to MTF measured at lens rentals. I don't know if Lensrentals used a 2 mm glass corresponding to the cover glass on the sensor in their test.
Screen Shot 2016-03-22 at 06.59.13.jpg
The Hasselblad lens look to have both a bit of field curvature and astigmatism:
Screen Shot 2016-03-21 at 03.31.21.jpg
Zeiss has a Batis 25/2.0 made for the Sony A7rII, according to the MTF data from Zeiss, that lens has no drop off in sharpness at all. But I didn't buy the Batis, as I want to have the flexibility of the zoom.
Screen Shot 2016-03-25 at 17.34.09.jpg

Are the new Schneider lenses for Phase One better than my Distagon? Probably! Schneider's presentation of MTF data is less clear than Zeiss or Hasselblad's and they use different frequencies 15/30/60 lp/mm vs 10/20/40 lp/mm. That means they are quite a bit better than what they look like.
Screen Shot 2016-03-25 at 17.45.31.jpg
Another option I could have used would be the Canon 24/3.5 TSE LII I also own.

So, what I see in these images? I would say the two are quite close in three of the four areas. In the extreme corner the Hasselblad lens wins. So, if corners are more important than the subject, the Distagon/P45+ combo is clearly to prefer. Otherwise it is a wash…

Distagon 40/4 f/11 to the left, Canon 16-35/4 at 25 mm f/8 to the right, moving out from center to corner from top down:
 

Paul2660

Well-known member
Eric

If I a looking at this correctly, the Distagon is doing a great job in the corners.

Thanks for the pics

Paul C
 

Abstraction

Well-known member
I did not realise it at first, but it seems something went amiss with my previous samples. Apparently, the forum software does not like png files. Let me try again. First, the image from imaging-resource showing where the crops where taken:View attachment 117438

- - - Updated - - -

Now the center crop. You may need to clic to see the original resolution.

View attachment 117439

- - - Updated - - -

And now, the side crop.

View attachment 117440
I think the left is MF and the right is 35mm. The corners on the right are a bit softer. Overall, I would say that it's still a wash. They're close enough.
 

jerome_m

Member
Re: Corner drop off

Regarding corner drop off, let's look at some real images
Were the other images any less real?


Many 24-35 lenses drop dramatically in MTF at the extreme corners, in some cases MFD lenses behave better in the corners.
Not only in the corner. The 2/3 zone is critical, especially for wide-angle and zoom lenses.


Zeiss has a Batis 25/2.0 made for the Sony A7rII, according to the MTF data from Zeiss, that lens has no drop off in sharpness at all. But I didn't buy the Batis, as I want to have the flexibility of the zoom.
View attachment 117443
But you are still using MTF curves with values of 10, 20 and 40 cycles/mm, which would be adequate for 12 Mpixels sensors. For the A7RII, we would need curves at 100 cycles/mm, wouldn't we?

- - - Updated - - -

Hasn't this been beat to death in numerous threads?

What's better, a cat or a dog? Graphs, charts, and internet images welcome.
Your wish is my command:

 
Top