[John Sadowsky Participant Posts: 169]

Micro 4/3

on: October 22, 2019 at 10:07 am

Kevin Raber wrote:

I still believe folks buy these cameras for different reasons. Olympus is very popular with serious photographers that take lots of trips or want something light for hiking etc. If you aren’t worried about big prints the Olympus is a feature-heavy well-built camera system.

Absolutely!  I’m not discounting the value of smaller formats.  They certainly have their niche.  I hope I are not coming across as being belligerent – that’s not my intent.

Equivalence theory (comparing across different formats at equivalent image quality) turns out to be a more complex topic than many people think.  For those looking for some serious torture, here is a link to a technical article on it:

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?SSO=1

 

JSS

[John Sadowsky Participant Posts: 169]

Re: Micro 4/3

Reply #1 on: October 21, 2019 at 10:34 pm

No sir.  Same shutter speed and aperture on different sized sensors are not equivalent exposures.  An equivalent exposure should deliver the same total amount of light to the sensor; that is, the same number of photons.  It is the sensor’s job to convert that light to a digital image.  It is a simple fact of geometry that the light gathering capability of the camera is determined by lens diameter – not f-number.  So when you say the same shutter speed and same f-number is an equivalent exposure for two different sensor sizes, what you are actually saying is that the smaller sensor deserves fewer photons.  Modern digital photography is largely limited by photon noise.  Fewer photons is not an equivalent exposure.

You have to compose the equivalence question as follows:  what does it take to take to capture the same image, with the same focus properties and the same image noise across the image.

Go to Wikipedia and look up the formulas for hyperfocal distance and DoF.  They are functions of N and c = the circle of confusion.  You have to scale c by the crop factor, because images should be compared on the bases of the same circle of confusion relative to the sensor size.  Eliminate N using N = f/D where D is diameter.   What you will find is that to get the same hyperfocal distance, and the same DoF, requires equal diameter lenses.  With equal diameter, equal exposure time then yields that both sensors are provided the same number of photons.

I’ve not talked about resolution.  For equal sized photosites (which is more or less constrained by semiconductor physics), the larger sensor has better resolution.  But we can’t compare images unless we agree to have the same circle-of-confusion relative to sensor dimensions.  The larger sensor have more Mpix, which only means they can take higher resolution images than the crop sensor.  But to compare we have to set a common resolution (circle-of-confusion relative to sensor dimension) that both cameras can shoot.  That can be done by resizing to, say, 8 Mpix (the “print” standard) then comparing on the basis of equal SNR.  That exactly what equal diameter and equal exposure times does.

I didn’t invent this.  It is a common misconception that smaller sensors make everything smaller and cheaper.  But when it comes to lenses – its just not true.  Equal diameter lenses are roughly the same size and weight regardless of the sensor size, and the limited data I looked at indicates that they are of comparable cost.  There may be other factors that make one format more cost effective than another.  (For one thing, if you want to realize the larger potential resolution of the larger sensor, you have to have a lens that can meet that resolution.)  That’s why I originally asked the question about cost comparison.  However, the bottom line is that the advantage of larger sensors is that they are capable of taking higher resolution images than crop sensors.

JSS

[John Sadowsky]

Re: Micro 4/3

Reply #2 on: October 21, 2019 at 5:58 pm

OK, I did some lens comparison using B&H pricing.  I searched for mirrorless only, in the telephoto categories – primes only.  The problem is that there isn’t a very big selection.  In order to compare prices of lenses with different focal lengths, for each lens I calculated it diameter = f/N, and then the area.  My hunch is that lens area would be the most significant cost determining factor.

There were only a few lenses in the FF category.  The top end are

Sony GM 400 mm f2.8 at $0.59/mm2
Sony GM 600 mm f4, at $0.58/mm2

(mm2 = square mm – can’t do a superscript here.)  Both these lenses have about the same diameter (143 and 150 mm respectively).

There where several 135mm FF lenses

Rokinon f2 at $0.11/mm2
Sigma f1.8 at $0.29/mm2
Sony GM f1.8 at $0.37/mm2
Zeiss Batis f2.8 at $0.51/mm2

For micron 4/3, the longest lenses were

Olympus 300 mm f4 at $0.57/mm2
Panasonic 200 mm f2.8 at $075/mm2

The Olympus 300 mm is price comparable to the Sony GM 400 and 600 in the FF format; that is, in terms of price per lens area.  However, the diameter of that lens is only 75mm, compared to the Sony GMs having a ~150 mm diameter.  That’s the difference between $12,000 lens (the Sony FFs) and a $2,500 lens (the Olympus 300 mm).  The Sonys are faster lenses.

The Panasonic seems to be  a price outlier.

At shorter focal lengths Rokinon 135 mm f2 was the cheapest at $0.13/mm2, while the Olympus 75 mm f1.8 is the high end as $0.66/mm2.  Other lenses landed in the $0.30-$0.50/mm2 range.

Conclusions

The sample size of this experiment is small, but I think the data supports my price-to-lens area ratio approach.   We expect Sony GM, Zeiss and Olympus to be expensive, because those are higher quality brands.

The Olympus – Sony comparison seems to show that micro 4/3 does not have an inherent lens cost advantage over FF.  The Olympus lens is a lot less expensive, but it is also a much slower lens.

JSS

• This reply was modified 4 years, 6 months ago by John Sadowsky.

[John Sadowsky Participant Posts: 169]

Re: Micro 4/3

Reply #3 on: October 21, 2019 at 3:45 pm

I should have been more precise.  I’m interested in doing some widelife photography, which is what I wanted a long lens for.  I would like to have something in the 300-600 mm range for FF.  All else equal, is it more cost effective at micro 4/3?

I differ with you on equivalence comparison.  When comparing across different sensor sizes, it is equal lens diameter, not f-number, that results in the same focus and exposure properties.  For the same exposure time and DoF (or hyperfocal distance), shooting at f2.8 in FF is equivalent to f1.4 for micro 4/3.  So that 70-200 comparison both at f2.8, of course the micro 4/3 lens is cheaper, but it is a much smaller lens and not as fast.

A better way to express the question is as follows.  For equivalent lenses, where equivalence means equal diameter and a focal lengths scaled by the crop factor, do micro 4/3 lenses a cost advantage?  I might have to go to B&H and start doing that comparison – enquiring minds want to know.

JSS

[John Sadowsky Participant Posts: 169]

Re: What is dynamic range?

Reply #4 on: October 18, 2019 at 7:08 pm

Here is a little interesting factoid.  When Sony says their cameras have a 15 stop dynamic range, you have to read the fine print!  That’s for images that have been resized down to 8 Mpix.  Downsizing it like averaging.  It reduces noise and hence increases dynamic range.  The DR of a Sony A7R III at full 42 Mpix resolution is 13.6 stops (see DP Review), not 15 stops.

That’s “engineering DR” as described in a previous response, which is also an optimistic standard.  It is based on a bottom end signal-to-noise ratio (SNR) = 1.  Most photographers would agree that SNR = 1 is an unacceptable noise level.  At the other extreme, SNR = 10 defines the bottom end of “photographic DR,” however, in some instances we can work with that level of noise.  The bottom line is that DR should be taken as a figure-of-merit for comparison of different cameras.  There are ambiguities in the precise definitions, so the actual number has to be taken with a grain of salt.  The DR of the human eye is only 6.5 stops.  The wider DR of our cameras is useful because we can recover information “lost in the shadows” or “lost by underexposure” in post processing.

Except for the very lowest stops of that dynamic range, the noise in images from digital cameras is dominated by photon noise – the granularity of light itself.  For 14 bit ADCs, ADC noise and distortion don’t really play a significant role at all.  (I personally am not convinced that 16 bits buys anything.). The biggest factor in determining the DR of a camera turns out to be the sensor’s Full Well Capacity (FWC).  The photo diodes in the sensor are photon counters, and the FWC is the maximum number of photons they can count before it hitting a ceiling at the top end of DR.

JSS

[Author]

Posts