"Format equivalence" and all that

[enricosavazzi]

I am quite sure no photographer has been able to completely escape from discussions of "format equivalence" and the many more-or-less unfounded, more-or-less erroneous, statements on which film format and sensor format is "better" for a given application. Luckily there have been a few attempts to underpin these comparisons with an objective, physics-based conceptual framework. If one is serious about understanding format equivalence, one way is to start with these trustworthy attempts instead of the many hearsay and make-believe discussions that one can find with a random googling, and the following is a recent-enough example to still be relevant today:

 

D. Andrew Rowlands, “Equivalence theory for cross-format photographic image quality comparisons,” Opt. Eng. 57(11), 110801 (2018), doi: 10.1117/1.OE.57.11.110801

 

Luckily, this paper is also freely available for reading online and PDF download for archival storage:

https://www.spiedigitallibrary.org/journals/optical-engineering/volume-57/issue-11/110801/Equivalence-theory-for-cross-format-photographic-image-quality-comparisons/10.1117/1.OE.57.11.110801.full

 

 

[colinbm]

@enricosavazzi My brain is hurting....

[dabateman]

This explanation isn't bad and maybe easier to understand:

http://www.josephjamesphotography.com/equivalence/

 

Also I do see some errors in the paper. A huge astrix must be next to table one and two indicating sensor generation.

My Nikon Df at ISO3200 is noisier than my Olympus EM5MK2 at ISO800. The pictures aren't equivalent.

Sensor generation matters.

 

Also the Kodak SLR/n camera shouldn't be used at ISO values greater than ISO160. Everything else was noisy and the Olympus E3 is better, which isn't saying much.

There are also numerous Canon 135 format sensors with worse dynamic range than m43rds cameras.  So this point can't be ignored. 

 

[ulf]

2 hours ago, dabateman said:

This explanation isn't bad and maybe easier to understand:

http://www.josephjamesphotography.com/equivalence/

 

Also I do see some errors in the paper. A huge astrix must be next to table one and two indicating sensor generation.

My Nikon Df at ISO3200 is noisier than my Olympus EM5MK2 at ISO800. The pictures aren't equivalent.

Sensor generation matters.

 

Also the Kodak SLR/n camera shouldn't be used at ISO values greater than ISO160. Everything else was noisy and the Olympus E3 is better, which isn't saying much.

There are also numerous Canon 135 format sensors with worse dynamic range than m43rds cameras.  So this point can't be ignored. 

You are right David, the sensor generation of sensors is important.

That adds yet another dimension in the multidimensional equivalence, just as sensors from different sources seams to have different performance.

[enricosavazzi]

@colinbm It's actually no rocket science, it is based on the same algebra and simple formulas used all the time e.g. in close-up imaging and photomacrography. It takes some effort to initially penetrate the mathematical formalism (which one could perhaps call "mathematical obscurantism") but after that it is easier going. It has been said that mathematical/scientific obscurantism is all about preventing the laymen (and sometimes your scientist colleagues) from seeing that your ideas as a scientist are in reality quite simple.

 

@dabateman, @ulf granted, it is not a complete coverage of all variables. There is also e.g. a lot of post-processing going on in firmware between reading the sensor data off the sensor and the supposedly "raw" data recorded in the image file, perhaps even more difference here between successive camera generations than the intrinsic difference in noise between successive generations of sensors.

 

 

[dabateman]

56 minutes ago, enricosavazzi said:

@colinbm It's actually no rocket science, it is based on the same algebra and simple formulas used all the time e.g. in close-up imaging and photomacrography. It takes some effort to initially penetrate the mathematical formalism (which one could perhaps call "mathematical obscurantism") but after that it is easier going. It has been said that mathematical/scientific obscurantism is all about preventing the laymen (and sometimes your scientist colleagues) from seeing that your ideas as a scientist are in reality quite simple.

 

@dabateman, @ulf granted, it is not a complete coverage of all variables. There is also e.g. a lot of post-processing going on in firmware between reading the sensor data off the sensor and the supposedly "raw" data recorded in the image file, perhaps even more difference here between successive camera generations than the intrinsic difference in noise between successive generations of sensors.

Yes absolutely. 

When I tested my Panasonic 42.5mm  f/1.2 lens on my GM5 vs on one of my Olympus cameras,  including the EM1MK1,  which uses the same Panasonic sensor,  the GM5 is better.  I think Panasonic know their lenses and program in some kind of control for diffraction and noise control into the raw. 

For this reason I may buy a used G9 one day, as I have many Panasonic lenses. 

 

[ulf]

1 hour ago, enricosavazzi said:

@dabateman, @ulf granted, it is not a complete coverage of all variables. There is also e.g. a lot of post-processing going on in firmware between reading the sensor data off the sensor and the supposedly "raw" data recorded in the image file, perhaps even more difference here between successive camera generations than the intrinsic difference in noise between successive generations of sensors.

More modern sensors also have improved dynamic ranges with more bits to work with and less noise from the readout electronics.

There is an incremental improvement closing in to the theoretical limit of conversion efficiency from photons to electric charges in the pixel wells.

Part of it is the improved semiconductor efficiency generating less heat during capture and conversion.

That is in a way inline with improved efficiency of more modern CPUs

[Andy Perrin]

3 hours ago, enricosavazzi said:

It has been said that mathematical/scientific obscurantism is all about preventing the laymen (and sometimes your scientist colleagues) from seeing that your ideas as a scientist are in reality quite simple.

I think that's (mostly) not true. I think people who say that are just making excuses for why they failed to learn math. If you are in a field like physics or engineering, mathematical models come about because the physical principles themselves are described mathematically. You can sometimes translate them in words, but the words don't let you design anything or build anything. That kind of understanding is very shallow and unsatisfactory for a scientist, but sometimes it's the best one can do for a layman explanation.

 

Also, why would anyone want to PREVENT laymen, who control the funds ultimately, from understanding the work?

--

 

Imagine if people started saying "It has been said that publishing sheet music is all about preventing laymen (and sometimes musician colleagues) from seeing that your new tune is really quite easy to play." Wouldn't we laugh at that nut!

[enricosavazzi]

8 hours ago, Andy Perrin said:

...

Also, why would anyone want to PREVENT laymen, who control the funds ultimately, from understanding the work?

...

There are two aspects (possibly three) to it. One, as I said, is an attempt to conceal that many of a scientist's ideas are as simple and obvious (once explained) as those of the next person. Simple ideas are hardly worth paying good money for. A second aspect is that the scientist is supposed to know the previous work of his/her colleagues, and must show this with abundant, relevant references to literature as well as by demonstrating a command of the terminology and methods used in the specific field. A non-scientific but necessary third aspect is the need to impress the people holding the money purse, many of whom are not scientists but career politicians (and in some cases failed scientists who turned to politics as an easier way to make a living). As a former scientist, I can attest that for most of my career the need for continued financing was never far from my thoughts (although by far the least of my actual motivations to carry out research).

 

As to why it may be effective to cloak one's ideas with obscure language, there are a few studies of the phenomenon of scientific obscurantism, albeit not many. Perhaps this subject is still a bit too sensitive to allow a self-critical evaluation by many scientists. But one of the motors is the instinctive awe in seeing someone else doing things one does not quite understand and does not know how to imitate. It is not restricted to science, we see it all the time in the behavior of wizards and witch doctors of the past, as well as the greatest and most successful fraudsters of our time.

 

Some references about obscurantism in science. The most common, and most accepted, approach is using a semi-serious approach, as done in most discussions of this subject:

https://www.jstor.org/stable/2015011

https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&cad=rja&uact=8&ved=2ahUKEwins_bcv7mEAxWtFRAIHX5xDSEQFnoECDsQAQ&url=https%3A%2F%2Ftidsskrift.dk%2Flev%2Farticle%2Fdownload%2F136279%2F181209%2F295585&usg=AOvVaw1UHVIX2VcLEJM8evu-Evxi&opi=89978449

https://acsjournals.onlinelibrary.wiley.com/doi/pdfdirect/10.3322/canjclin.20.6.360

[Andy Perrin]

There are undoubtedly a few bad scientists who use math like some kind of witch doctor to hide behind. But it’s not a very effective curtain because the majority of real scientists DO know their math (if they are in a relevant field of science). Such papers tend to get mocked, like that case of the person who re-invented integrals a bunch of years ago:

https://academia.stackexchange.com/questions/9602/rediscovery-of-calculus-in-1994-what-should-have-happened-to-that-paper
 

I still have no idea what you mean by using math to obscure simple ideas. How are you going to design a building with no math? The concepts can certainly be explained in ordinary language (Mario Salvadori’s “Why Buildings Stand Up” is a fantastic example of this) but you can’t build anything from Mario’s book.
 

Also the general idea that scientists are just wizards standing behind a curtain of humbug is dangerously prevalent in our culture right now. There are some fakers out there for sure, but pushing the idea that math is just for hiding behind is false. 

[Lou Jost]

9 hours ago, enricosavazzi said:

There are two aspects (possibly three) to it. One, as I said, is an attempt to conceal that many of a scientist's ideas are as simple and obvious (once explained) as those of the next person. Simple ideas are hardly worth paying good money for. A second aspect is that the scientist is supposed to know the previous work of his/her colleagues, and must show this with abundant, relevant references to literature as well as by demonstrating a command of the terminology and methods used in the specific field. A non-scientific but necessary third aspect is the need to impress the people holding the money purse, many of whom are not scientists but career politicians (and in some cases failed scientists who turned to politics as an easier way to make a living). As a former scientist, I can attest that for most of my career the need for continued financing was never far from my thoughts (although by far the least of my actual motivations to carry out research).

 

As to why it may be effective to cloak one's ideas with obscure language, there are a few studies of the phenomenon of scientific obscurantism, albeit not many. Perhaps this subject is still a bit too sensitive to allow a self-critical evaluation by many scientists. But one of the motors is the instinctive awe in seeing someone else doing things one does not quite understand and does not know how to imitate. It is not restricted to science, we see it all the time in the behavior of wizards and witch doctors of the past, as well as the greatest and most successful fraudsters of our time.

 

Some references about obscurantism in science. The most common, and most accepted, approach is using a semi-serious approach, as done in most discussions of this subject:

https://www.jstor.org/stable/2015011

https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&cad=rja&uact=8&ved=2ahUKEwins_bcv7mEAxWtFRAIHX5xDSEQFnoECDsQAQ&url=https%3A%2F%2Ftidsskrift.dk%2Flev%2Farticle%2Fdownload%2F136279%2F181209%2F295585&usg=AOvVaw1UHVIX2VcLEJM8evu-Evxi&opi=89978449

https://acsjournals.onlinelibrary.wiley.com/doi/pdfdirect/10.3322/canjclin.20.6.360

 I'm with Andy on this. Math is absolutely necessary to really understand most complex processes. Sure, a good scientist does have an intuitive feel for a process and can roughly describe it in words, but this is not enough for most tasks. Enrico, I suspect you were just adding a humorous note, and not making a general assessment of math in science. Nearly every field of science depends on math; even qualitative fields still usually need to apply mathematical statistical reasoning to judge the meaning of their results.

[photoni]

Dear @enricosavazzi Today, talking about format equivalence is questionable.
Are you looking for a better engraving? a different depth of field? a different color depth (from 8 to 16 bit)?

For 25 years I have taken photos with the large 4x5 inch format with dedicated lenses and Ektachrome 64 slides; today with my standard Nikon Z7 and Nikkor Z 105 I have much more.
Probably the most advanced sensor today is that of the Hasselblad CFV 100 etc... and the Fuji GFX100, but I read that Samsung will make a smartphone with 200 Mpx, the difference is predictable, it is ridiculous in terms of engraving, color depth etc.; but in "aesthetic" terms it changes little, change the depth of field, which is almost infinite with the smartphone and can be modified as desired with the 33x44 mm medium format.

You don't need mathematical formulas to define the aesthetics, if you take a photo with film or wet collodion in 8x10 inch format with a Petzval 300mm f4 lens ... or with a 24x36mm mirrorles with a New Petzval 85mm f2.2 lens or better yet a 50 f1 ,2 ... it changes little if we compare them on the PC monitor, it changes a lot if we pick up a silver print or a wet plate and an inckjet print.

[Lou Jost]

3 hours ago, photoni said:

Dear @enricosavazzi Today, talking about format equivalence is questionable.
Are you looking for a better engraving? a different depth of field? a different color depth (from 8 to 16 bit)?

For 25 years I have taken photos with the large 4x5 inch format with dedicated lenses and Ektachrome 64 slides; today with my standard Nikon Z7 and Nikkor Z 105 I have much more.
Probably the most advanced sensor today is that of the Hasselblad CFV 100 etc... and the Fuji GFX100, but I read that Samsung will make a smartphone with 200 Mpx, the difference is predictable, it is ridiculous in terms of engraving, color depth etc.; but in "aesthetic" terms it changes little, change the depth of field, which is almost infinite with the smartphone and can be modified as desired with the 33x44 mm medium format.

You don't need mathematical formulas to define the aesthetics, if you take a photo with film or wet collodion in 8x10 inch format with a Petzval 300mm f4 lens ... or with a 24x36mm mirrorles with a New Petzval 85mm f2.2 lens or better yet a 50 f1 ,2 ... it changes little if we compare them on the PC monitor, it changes a lot if we pick up a silver print or a wet plate and an inckjet print.

Format equivalence matters more than ever these days. I have been using 4"x5", 6cm x 7cm, and 35mm for forty years and now I use every format from 1/2" to 4"x 5". It is more important than ever to understand the effect of format and lens size. Not just gut feelings but actual quantitative understanding of the ways in which the formats are equivalent, and the ways in which they are not.

[enricosavazzi]

On 2/20/2024 at 6:31 PM, Lou Jost said:

[...] Enrico, I suspect you were just adding a humorous note, and not making a general assessment of math in science. [...]

Absolutely, keeping the discussion in humorous, or at least semi-serious, terms makes it a lot easier for scientists to be self-critical. Any non-humorous discussion would very likely be interpreted as a personal attack, and either prompt a personal attack in return, or be ignored.

[photoni]

15 minutes ago, enricosavazzi said:

Absolutely, keeping the discussion in humorous, or at least semi-serious, terms makes it a lot easier for scientists to be self-critical. Any non-humorous discussion would very likely be interpreted as a personal attack, and either prompt a personal attack in return, or be ignored.

I totally agree @enricosavazzi
I also see something ridiculous, I who am a curious fan have more messages than you who are an icon of mine

[photoni]

I found this table of equivalences length and brightness of the lenses in proportion to the film - sensor format

.

I have an old Petzval lens (noname) made around 1860, it is long 300mm f 4.5
according to this table if it is used on an 18x24 cm format it is equivalent to a 43mm f0.6 lens on the Leica format

.

[Andrea B.]

You all will probably laugh at my simple take on "format equivalence". Does the lens cover my sensor, yes or no? If yes, then I will learn how to get the best photo with it that I possibly can based on characteristics of the given sensor and the given lens. 

 

The subject of format equivalence is interesting theoretically. (And yes I understand the math. Or at least I once did. "Use it or lose it". LOL!!!) 

 

I downloaded the Rowlands PDF. Thanks Enrico for posting that. 

I had the link for the James write-up. But thanks David for posting that.

 

 

[Andrea B.]

Rowlands:  Nowadays many photographers use several cameras based on different format sizes and need a framework for translating camera settings from one format to another.

 

Has any one of us ever "translated camera settings from one format to another"?

Well, I haven't. YMMV, of course.  I would venture to guess that most of us just figure out new, optimal settings and then take a few test photos to check whether adjustments are needed. I am not trying to diss Rowland's work. I get that he is listing justifications for his presentation. And that is indeed one of them. But it does seem to be more of a theoretical take rather than a practical approach.

 

Rowlands: Equivalent photos are produced using the same amount of light.

This is probably the more important statement. Especially in situations where getting enough light might be a problem. Such as in UV photography, huh??  

[Lou Jost]

1 hour ago, Andrea B. said:

You all will probably laugh at my simple take on "format equivalence". Does the lens cover my sensor, yes or no? If yes, then I will learn how to get the best photo with it that I possibly can based on characteristics of the given sensor and the given lens. 

 

The subject of format equivalence is interesting theoretically. ...

 

 

I think the real first question we ask is not "Does the lens cover my sensor?" but rather "Which sensor (and lens) should I use for this photo?" The subject of format equivalence is an everyday practical issue.

[Andrea B.]

For some, maybe.

 

For me, at this point, not a consideration 'cause I only use full frame cams/lens. I do have leftover old 4/3 and smaller sensor conversions (D200, D300, Lumix GH-1), but no longer use them. (The IQ was pretty bad.) And medium format is too heavy & expensive (...for me, personally.) 

 

Thinking thru all my UV-capable and UV-dedicated lenses, I seem to have exactly one UV-dedicated lens which will not cover full frame. So I just switch the camera to DX mode.

Yes, fewer pixels in DX mode, of course, etc. So it might indeed be better to switch to a DX format camera with more pixels? Hard to say whether that would make any significant difference unless one is printing large.

 

It is still fun and interesting to read about, to try to understand, and to discuss this stuff. 

 

 

[Andrea B.]

...that case of the person who re-invented integrals a bunch of years ago

 

OMG!! That is totally hilarious. 

The fact that Tai's paper passed "peer review" certainly tells you something about those "peers". 

geez!! unbelievable!!

[lukaszgryglicki]

For me this is simple now - I would use the largest digital sensor that is there.

I wish I will have 6x7 or 6x9 size digital sensor some time in the future - I'll make use of my Pentax 67 lenses on it and will be happy.

I can be 6x9 cm size and resolution can be even 25 Mpix - I don't need more.

 

[dabateman]

9 hours ago, lukaszgryglicki said:

For me this is simple now - I would use the largest digital sensor that is there.

I wish I will have 6x7 or 6x9 size digital sensor some time in the future - I'll make use of my Pentax 67 lenses on it and will be happy.

I can be 6x9 cm size and resolution can be even 25 Mpix - I don't need more.

 

Well for $26000 you can by a 6.7 Mpixels 4.7 x 5.5 inch sensor, the LS45.

http://largesense.com/index.php/products/4x5-large-format-digital-back-ls45

 

They recommend a 5x7 camera.

That might almost be affordable and is large. The 911 is larger for the real full frame sensor,  but its very expensive. 

[Lou Jost]

26 minutes ago, dabateman said:

Well for $26000 you can by a 6.7 Mpixels 4.7 x 5.5 inch sensor, the LS45.

http://largesense.com/index.php/products/4x5-large-format-digital-back-ls45

 

They recommend a 5x7 camera.

That might almost be affordable and is large. The 911 is larger for the real full frame sensor,  but its very expensive. 

This shows why understanding format equivalence is so important. There are huge economic issues with always using the largest format.

[photoni]

we must not forget that you don't just need a larger sensor, you also need lenses that have an equivalent engraving... if you don't want to take dreamlike photos :D
.
I remember that when I used the 4x5" Ektachrome only some lenses (for example the aporonnar 240 f9 with intermediate apertures 22-32 enhanced the engraving) others had a softness and terrible chromatic aberration (schneider symmar 240 f5.6)
.
old Leitz lenses on the new M11 are not recommended
I read that to take advantage of the quality of the sensor you need the new 50 apo which costs the same as a GFX100

.

the Largesense LS45 6.7 Mpixel 4.7 x 5.5 inch sensor, could be good for UV, BG38 filter is recommended to be used.
The price isn't high but I think 6.7 Mpx is ridiculous
they are much worse than the 55 polaroid negative film that Ansel Adams used.