Olympus Outlook

[Cruzan80]

My point was that it is useless to compare to a standard, unless the standard is something you desire.

 

[jnewell]

The full-frame lens has a larger imaging circle than the u4/3 lens - so even at the same f-stop it is gathering 4x the total light (heck, the lens is 4x the size, so what do you expect ;-).

...and spreading that light over a much bigger surface area. The amount of light delivered per square unit measure will be the same.

 

[TransientEye]

...and spreading that light over a much bigger surface area. The amount of light delivered per square unit measure will be the same.

Yes - the flux per unit area is the same given the same f-stop. However, a full-frame lens + sensor is collecting the light from from 4x the area, so the total light captured is 4x larger.

As some people are fond of repeating, a 50mm f1.8 lens is a 50mm f1.8 lens, regardless of format. The difference is that lenses designed for full-frame will normally create an image with 4x the area of a u4/3 lens, which is why the glass is bigger and the field-of-view wider for the same focal length.

 

[barry]

Yes - the flux per unit area is the same given the same f-stop. However, a full-frame lens + sensor is collecting the light from from 4x the area, so the total light captured is 4x larger.

Another way to look at is is that the sensor 'pixels' are larger on a full-frame sensor, as long as it's less than 48MP.
Then, each pixel can capture more photons.

Barry

 

[jnewell]

Yes - the flux per unit area is the same given the same f-stop. However, a full-frame lens + sensor is collecting the light from from 4x the area, so the total light captured is 4x larger.

As some people are fond of repeating, a 50mm f1.8 lens is a 50mm f1.8 lens, regardless of format. The difference is that lenses designed for full-frame will normally create an image with 4x the area of a u4/3 lens, which is why the glass is bigger and the field-of-view wider for the same focal length.

OK, we're on the same page, err, frame. I thought you had a different perspective on the point that you make in the first sentence of the second paragraph.

 

[tomO2013]

This is where I don't necessarily agree.
The total light captured as an equivalent system whole, does not necessarily follow a linear scale as you move up to a bigger sensor i.e. the generalization that is being made here is that a 50mm lens on a sensor 4X larger than m43 correlates to 4 times the amount of light that is captured by that sensor. This is misleading and is not correct (I need to say here that I am not saying that you are misleading and I appreciate you taking the time out of your day to give your opinion on this here, but the common belief itself is misleading IMHO). My problem stems from the use of the word 'capture'.

Light is not stored in a bucket where the bigger sensor is in effect a four times bigger bucket.
The sensor is more akin to a sponge, where more important than sensor size (sponge in this analogy ) is efficiency of absorption of that sponge. It doesn't matter a damn if a 35mm sensor is bigger, has bigger pixels, has light falling over a bigger surface area if less of that information is captured accurately.

http://admiringlight.com/blog/full-frame-equivalence-and-why-it-doesnt-matter/
A quote from this excellent article (BTW I don't agree that smaller sensors are always more light efficient than larger ones. Again here the answer I believe is 'it depends', but generally speaking relative to their pixel pitch, smaller sensors tend to perform much better.
The premise as a whole I agree with...

This is why, often, it’s said that full frame sensors will have two stops better ISO performance over a Micro 4/3 sensor.
Noise comparison - OM-D E-M5, Canon 5D and 5D Mark II, Courtesy, DxO, Click to Visit DxO Labs Comparison Tool
Noise comparison – OM-D E-M5, Canon 5D and 5D Mark II, Courtesy, DxO, Click to Visit DxO Labs Comparison Tool
However, this doesn’t work completely linearly in the real world, as smaller sensors are more light efficient than larger ones, and it is also dependent on sensor technology being identical. If you look at the DxO Mark sensor comparison, you will see that if you compare the Olympus OM-D E-M5 with the Nikon D600 and the Canon 5D Mark III (both of which are full frame sensors), they measure roughly 1 2/3 stops better in ISO performance. Not two like you’d expect. You may be thinking ‘that’s close enough, but it speaks to the further point:
Compare different sensor generations and everything breaks down, though advocates of this equivalence never use the equivalence when comparing full frame sensors of different generations. The OM-D E-M5 is only about a half a stop behind the Canon 5D and just over one stop behind the 5D Mark II…a camera that was current only 9 months ago. (See chart above).
My whole point here is that the total light argument implies a direct 2x or 4x improvement in image quality with ISO. However, this is really only true for approximating how one sensor design may scale with size, but is terrible as a blanket equivalence based on sensor size alone due to the changing of technologies over time and different sensor construction even among cameras of the same generation.

I'll follow this up with two interesting links from techradar on that are interesting from both a dynamic range and SnR perspectives:
EM5 dynamic range at base ISO's for TIFFs is extremely competitive with 35mm (converted RAW). There isn't 2 stops worth of a difference :
http://www.techradar.com/reviews/ca...pus-om-d-e-m5-1075717/review/5#articleContent
5d mark iii (much cleaner high ISO , but much lower dynamic range compared to the EM5 in lower ISO's )
http://www.techradar.com/reviews/ca...n-5d-mark-iii-1074186/review/6#articleContent

Finally just one last quote from Michael Reichmann of luminous landscape from his XT1 review here http://www.luminous-landscape.com/reviews/cameras/fuji_x_t1_further_thoughts.shtml:

Yes - the flux per unit area is the same given the same f-stop. However, a full-frame lens + sensor is collecting the light from from 4x the area, so the total light captured is 4x larger.

As some people are fond of repeating, a 50mm f1.8 lens is a 50mm f1.8 lens, regardless of format. The difference is that lenses designed for full-frame will normally create an image with 4x the area of a u4/3 lens, which is why the glass is bigger and the field-of-view wider for the same focal length.

 

[TransientEye]

This is where I don't necessarily agree.
The total light captured as an equivalent system whole, does not necessarily follow a linear scale as you move up to a bigger sensor i.e. the generalization that is being made here is that a 50mm lens on a sensor 4X larger than m43 correlates to 4 times the amount of light that is captured by that sensor. This is misleading and is not correct (I need to say here that I am not saying that you are misleading and I appreciate you taking the time out of your day to give your opinion on this here, but the common belief itself is misleading IMHO). My problem stems from the use of the word 'capture'.

It is correct *if* the sensors are similarly efficient - it's just optics.

Of course, there are real-world differences that mean one brand/size system may be more efficient than another, but usually these are just transient as newer designs emerge to given a different brand/format a minor advantage. Currently, Olympus and Panasonic have some of the best sensor technology available - but two years ago their 12 Mpixel sensors were not at all competitive. Canon's current sensors are lagging in performance, but Nikon's D800 (Sony again) has roughly similar efficiency and read-noise when compared to Olympus (e.g. see http://www.sensorgen.info).

Current sensors are pretty amazing and are starting to hit the limits of physics. The days of getting huge improvements year-on-year in sensor RAW performance are long gone, and manufacturers are resorting to tricks to give illusion of better performance (the weak CFA in the 5D Mark III and the weak IR and UV filters in current u4/3 16Mpixel sensors - the cause of the "purple flare" problem).

So I do not agree that somehow one size of sensor is magically (non-linearly?) better than another...

Also, I think that Michael Reichmann is incorrect in generalising that sensor size does not matter for image quality. It is true that in good light you simply will not see a difference - we have some sets architectural images from a mix of a Canon G12 and a Canon 5D, and most people can not tell the images from the smaller camera. But the sensor size does matter if you want to achieve a wider-angle perspective with shallow depth of field, or if you want to shoot moving subjects in low-light.

I do agree that lens choice matters more, but that is a completely different discussion.

I never understood why equivalence was so controversial. It is not a question of whether one format is "better" than another, because "better" is always subjective. It is only one of many characteristics of a camera system - the majority of which are likely more important for most people. These are tools, not football teams...

 

[orfeo]

A large format camera or a chamber camera is gathering way more light than your tiny 135mm camera you fools... But that doesn't mean it's more suited for low light shooting! Same goes for mu43 ... Its beat 135 camera because it got more dof at fast aperture!

 

[TransientEye]

Same goes for mu43 ... Its beat 135 camera because it got more dof at fast aperture!

Yes, but the u4/3 image is noisier as the sensor captures only 1/4 of the light. As a result, you can stop down the larger camera to get the same DOF and get exactly the same image.

The only inherent difference sensor size makes is that a larger sensor gives the *option* to get lower noise at the expense of shallower DOF (and a much larger camera/lens). If your system has appropriate DOF for your needs, then issues such as lens selection, ergonomics, AF and IS are much more important (and these all very subjective and personal preferences).

 

[OzRay]

You've just gotta love forums.

 

[orfeo]

Yeah and the 135 sensor is noisier than large format sensor too.. IM perfectly fine with mu43 low light noise!

 

[TransientEye]

You've just gotta love forums.

Yes, sorry! Sometimes my inner astronomer gets carried away...

 

[alex66]

Yes, but the u4/3 image is noisier as the sensor captures only 1/4 of the light. As a result, you can stop down the larger camera to get the same DOF and get exactly the same image.

The only inherent difference sensor size makes is that a larger sensor gives the *option* to get lower noise at the expense of shallower DOF (and a much larger camera/lens). If your system has appropriate DOF for your needs, then issues such as lens selection, ergonomics, AF and IS are much more important (and these all very subjective and personal preferences).

Sorry but I do not get this, how does a larger sensor that is exposed correctly at say 1/60 f8 gather more light than one that is smaller and gives me a correctly exposed image at the same settings? Going to this bucket gathering rain water argument I keep hearing that again I am amiss on, when rain ager is measured (tend to use a beaker) you don't measure volume or area but height of the water captured. It does not matter what size the beaker is and if you were measuring volume you would have to use the same sized beakers anyhow or the results would be a nonsense. The light measuerement is ev per unit area, this is the same as measuring the height of rainfall. Now a smaller sensor will give a greater amount of depth of field per angle of view than a larger one, If I was wanting this razor thin DOF I would not waste my time with a so called FF its time for medium or large format.

 

[barry]

Sorry but I do not get this, how does a larger sensor that is exposed correctly at say 1/60 f8 gather more light than one that is smaller and gives me a correctly exposed image at the same settings? Going to this bucket gathering rain water argument I keep hearing that again I am amiss on, when rain ager is measured (tend to use a beaker) you don't measure volume or area but height of the water captured. It does not matter what size the beaker is and if you were measuring volume you would have to use the same sized beakers anyhow or the results would be a nonsense. The light measuerement is ev per unit area, this is the same as measuring the height of rainfall.

Hi,

1. The beakers are NOT the same size.
2. Volume DOES matter

Let's compare a 16MP 4/3 sensor to a 16.2 FF sensor (such as in the Nikon D4S).
The sensor 'pixels' are ~4x larger (in area) on the full-frame sensor.
Thus, each pixel can capture ~4x more photons (VOLUME, not depth).

This makes a difference wrt high ISO / sensitivity / SNR, etc.

To use your water analogy, imagine 2 buckets filled with small beakers or graduated cylinders or whatever.
The larger bucket has larger beakers.
Each beaker in the larger bucket will catch more water (in VOLUME, not in depth).

Googling finds lots of sites discussing this, here's a nice one with pictures of photon 'rain':
http://www.clarkvision.com/articles/does.pixel.size.matter/

Anyways, I'm sticking with my 4/3 cameras as the sensor performance is fine for my uses.

Barry

 

[alex66]

Hi,

1. The beakers are NOT the same size.
2. Volume DOES matter

Let's compare a 16MP 4/3 sensor to a 16.2 FF sensor (such as in the Nikon D4S).
The sensor 'pixels' are ~4x larger (in area) on the full-frame sensor.
Thus, each pixel can capture ~4x more photons (VOLUME, not depth).

This makes a difference wrt high ISO / sensitivity / SNR, etc.

To use your water analogy, imagine 2 buckets filled with small beakers or graduated cylinders or whatever.
The larger bucket has larger beakers.
Each beaker in the larger bucket will catch more water (in VOLUME, not in depth).

Googling finds lots of sites discussing this, here's a nice one with pictures of photon 'rain':
http://www.clarkvision.com/articles/does.pixel.size.matter/

Anyways, I'm sticking with my 4/3 cameras as the sensor performance is fine for my uses.

Barry

But measuring light fall or rain fall is done on fall her unit area, so rain is measured in mm and an EV value is measured the same way a larger pixel has an advantage with noise but it does not affect the EV response by being larger but the after effects of this. Besides I was questioning the wrong analogy of using a larger bucket to measure rainfall where the measurement is in mm regarding height the beaker could be 1m squared or 10mm squared it will give the same result.

 

[fortwodriver]

Yeah... it's not a useful calculation. f8 @ 1/125 @ ISO 400 is the same regardless of the sensor/film size.
If that's not the case when you test, it's because your meters in the cameras you used are wrong, or your shutters and apertures are mis-calibrated.

This whole "bigger sensors funnel in more light" thing is a nifty abstract idea - but it's being used in a completely incorrect way. Sensors are calibrated to ISO sensitivity. So a large sensor that seemingly accepts more light, is manufactured to be "sensitive" to whatever base ISO it's designed to mimic.

Everything else above that base sensitivity is a determined by increasing gain at the sensor.

 

[Reflector]

Yes, but the u4/3 image is noisier as the sensor captures only 1/4 of the light. As a result, you can stop down the larger camera to get the same DOF and get exactly the same image.

The only inherent difference sensor size makes is that a larger sensor gives the *option* to get lower noise at the expense of shallower DOF (and a much larger camera/lens). If your system has appropriate DOF for your needs, then issues such as lens selection, ergonomics, AF and IS are much more important (and these all very subjective and personal preferences).

That assumes that every sensor has the same pixel per pixel performance which none of the equivalence trolls will ever admit is untrue... By all means the 5D should be crushing all with its intrinsic superiority by that argument.

 

[dhazeghi]

Yeah and the 135 sensor is noisier than large format sensor too...

Until the arrival of Sony's 50MP 44x33mm CMOS sensor a few months ago, that wasn't true at all. Anything larger than 135 format used older CCD designs from Kodak and Dalsa that performed very poorly at higher ISOs. It appears that the new Sony sensor might change that, but I've yet to see any actual high ISO tests between them.

 

[fortwodriver]

Until the arrival of Sony's 50MP 44x33mm CMOS sensor a few months ago, that wasn't true at all. Anything larger than 135 format used older CCD designs from Kodak and Dalsa that performed very poorly at higher ISOs. It appears that the new Sony sensor might change that, but I've yet to see any actual high ISO tests between them.

Yeah. CCD is great at small sizes... Once you get to gargantuan medium format sizes it gets really hard to keep electrical and read noise down across a large CCD surface.

 

[Fri13]

It is a question of perspective...

The full-frame lens has a larger imaging circle than the u4/3 lens - so even at the same f-stop it is gathering 4x the total light (heck, the lens is 4x the size, so what do you expect ;-).

ISO is defined as the sensitivity required to expose and image for a given f-stop and shutter speed. Hence if I need ISO 200 1/50th on u4/3, the same same scene will need ISO 200 1/50th on full-frame. But because the total light captured is more, the full-frame image will have two stops less noise even though the ISO values are the same...

The total light capture on sensor isn't different. Both sensors gets exactly same amount of light with same lens.

I hate when people talking about small format sensor size benefits in light gathering versus m4/3 and praising its possibility to give less noise. That is just pure BS.

Why? Because they do not understand the digital part in digital cameras! With sensors, A/D processors and all other from hardware related to software related algorithms there is the ruling factor!

The noise level is results of used sensor, A/D processor etc. Each invention of design of sensor layout is patented and designed for each sensor size. When there are different generation sensors and different batteries and different A/D processors and different heat transfer designs in each camera, noise levels are different!

We can take latest m4/3 sensor generation and compare it to 5D Mk I & II D2X & D3 to see how it looks same, even when pixel density is higher and sensor size is smaller.

We can take a 4/3 camera like E-520 what has same size sensor as EM-1 but 10Mpix vs 16Mpix and see clearly that newer sensor technology blows older out of the water in dynamic range (!= exposure range)!

We can even take a Sony A7 model what has APS-C crop mode. What it does is disable surrounding pixels around APS-C sensor size, electronically! Meaning the A7 sensor is suddenly now a 1.5x crop factor sensor.
Now if people would still believe sensor size is responsible for light gathering, it would mean now suddenly the A7 with crop sensor would require a one stop more light and noise level would raise by one stop.

Even if different manufacturers use same sensors in different cameras, they do have different A/D processors feeding electricity to sensor and reading sensor output different ways and all other electronical differences affecting the resulting digital image from analog light manipulated via software with different algorithm.

If sensor size and pixel size would be only factor ruling noise levels, we would never had change to get higher than 1Mpix level sensor on small format and no need to get other electronical components better either.

In few years from now on, m4/3 cameras will gain a next generation or two sensors. They will improve the dynamic range so we get either less noise, more pixels or then both and maybe even wider exposure range from 13 stops to 14-15.

Who knows if someone invents a sensor design what gives 80Mpix to m4/3 sensor size with dynamic range what gives ISO 102400 with quality like today's ISO 200 and 16 stops exposure range so there is no clipping in either end because A/D processor can convert data in 16bit?

If that would be possible, there would be challenge to improve optics meet higher resolution and memory card capacity and other electronical parts what would clearly come a bottleneck for new sensor. Now think about 42 million pixels Nokia pushed to their 808 smartphone 1/1.2" size what is almost 1/4 of m4/3 sensor size.

If we now could take all electronics from Sony A7s and retrofit them to EM-1 with only chage being sensor being sliced to m4/3 sensor size, why suddenly would the amount of light landing on sensor drop or dynamic range be worse? We would really only get a less pixels (~3Mpix = 12Mpix / 4) but otherwise get same exposure range, same dynamic range when using same lens between modified EM-1 and A7s(+ change to angle of field).
Meaning placing those two cameras side by side, we would get exactly same exposure range and same dynamic range for both cameras