Micro Four Thirds cameras are better at Low Light. (busting the Full Frame myth)
The Convention Wisdom is Wrong
The convention wisdom says “Any camera can get the photo in daylight. If you want to shoot in low light, full frame cameras are better than crop sensor cameras.”
This is backwards. Full Frame cameras are better in daylight. Crop sensor cameras tend to prefer low light. A controversial statement I know — let’s examine this in detail.
The larger photosites of Full Frame cameras allows them to gather more photons. When do you need to gather more photons? During the day when there are more photons to gather.
Having larger photosites allows the camera to operate at a lower base ISO. Having a lower base ISO makes them better for bright light situations.
Nikon has a full frame camera with a true base ISO of 64. Most full frame cameras have a base ISO of 100. They have “deeper wells” allowing them to gather more light before saturating.
Crop sensor cameras have smaller photosite wells — they can’t gather as many photons before they saturate. Many crop sensor cameras have a base ISO of 200.
An Example — Daylight
Let’s say you’re shooting daylight portraits.
A typical exposure for f/2.8 may be ISO 100, and 1/4000 sec. (Calculated from the Sunny 16 rule — 5 stops brighter on the aperture and 5 stops darker on the shutter speed).
Now let’s say we want to take the same photo on a Micro Four Thirds sensor. First we need to shoot at f/1.4 to get an equivalent depth of field — two stops brighter. Then the camera’s base ISO is 200 — another stop brighter. We need 3 stops of shutter speed to darken down the photo — 3 stops darker than 1/4000 of a second, but you just can’t do it.
Most cameras top out at either 1/4000 or 1/8000 of a second. Some cameras can shoot at faster shutter speeds thanks to electronic shutters, but that introduces artifacts. My Olympus Pen-F allows me to shoot at 1/16000 with electronic shutter — that’s two stops darker than 1/4000 but not the three required.
Full Frame cameras are better for daylight photography.
An Example — Low Light
Now let’s do the reverse. Let’s say we’re shooting in low light, again on the full frame camera. Still at f/2.8 but now we’re at ISO 1600 and 1/50 sec. 4 stops above base ISO.
The equivalent Micro Four Thirds image is taken at f/1.4 — two stops brighter. We want to keep the same shutter speed of 1/50 sec, so we make up for it with ISO — we’re now two stops lower at ISO 400. This is just 1 stop above base ISO.
Whether or not this results in less shot noise can be debated (see Quantum Efficiency below). But Micro Four Thirds cameras prefer low light to bright light scenarios.
The Inverse Square Law
This becomes easier to understand if you imagine the inverse of a camera — a projector. We tend to think of lenses as things we attach to cameras — but as a thought exercise, let’s imagine it as the other way around — we bring the sensor closer to the lens.
Whether a projector or a lens, an image is being projected onto a flat surface.
The further away the surface, the larger the image and the dimmer the image as well.
The same total light reaches each sensor, but because the light has to cover a larger area, it is dimmer. How much dimmer is governed by the Inverse Square law.
The Inverse Square Law describes how light travels away from a point. By the time it’s traveled twice as far, it’s at 1/4 the intensity because it has to cover 4x the surface area.
We can think of sensor size in the same way.
This 4x difference in light intensity is why a full frame camera needs +2 stops more ISO for the same image relative to a Micro Four Thirds camera.
The inverse square law is captured neatly in the f-stop calculation.
Equivalent Depth of Field equals Equivalent Light Hitting the Sensor
The f-number (or f stop) is defined as focal length / the diameter of the entrance pupil. N = f/D.
A 50mm lens on full frame is 25mm on Micro Four Thirds. To get the equivalent depth of field, you similarly divide aperture by 2.
A 50mm f/4 lens on full frame has an entrance pupil of 12.5mm. (50/12.5 = 4)
A 25mm f/2 lens on Micro Four Thirds has an entrance pupil of — 12.5mm (25/12.5 = 2).
There is the same amount of total light hitting the sensor in both cases (same entrance pupil), but it’s concentrated onto a smaller area, which is why you must decrease your ISO to get the same exposure — ISO doesn’t care how big your sensor is, it cares how intense the light is.
Smaller sensor + same entrance pupil = same total light, but like taking a magnifying glass and using it to start a fire, the smaller the area, the more intensity, therefore the lower your ISO needs to be.
If both sensors are receiving the same amount of total light, then the argument for smaller sensors being worse at low light is “quantum efficiency” — smaller pixels are noisier than larger pixels.
Signal to Noise Ratio (The Quantum Efficiency Argument)
“Larger pixels are better at gathering light, therefore full frame cameras are better at low light.”
It’s true that larger pixels are better at gathering light — if both sensors are the same size.
(Though it is questionable if overall sensor noise decreases because of this — see the DP Review TVvideo “Why lower resolution sensors are NOT better in low light.”)
But as we just demonstrated, crop sensor cameras gather the same amount of light as full full frame cameras.
If each pixel receives the same amount of total light, the argument is now that larger pixels have better “quantum efficiency” — the larger pixels are more efficient at “counting photons.”
This should show up as improved SNR — Signal to Noise Ratio and better Color Fidelity. The camera with the better SNR and Color Fidelity is the one that’s better at “counting” the number of photons that reach it.
To get the same photo on a full frame and Micro Four Thirds camera, the full frame camera must shoot at a higher ISO (in accordance with the inverse square law).
Does the quantum efficiency argument hold if one camera is so much further above is base ISO?
Comparison — SNR
I went to DxO and compared two cameras. I chose the two most recent 20 megapixel cameras from Canon (full frame) and Olympus (m43) on their list.
The Olympus OMD-EM1 mk2 was released in 2016 and is a 20 megapixel Micro Four Thirds.
The Canon EOS-1D-X mk3 was released in 2020 and is a 20 megapixel full frame camera.
https://www.dxomark.com/Cameras/Olympus/OM-D-E-M1-Mark-II---Measurements
https://www.dxomark.com/Cameras/Canon/EOS-1D-X-Mark-III---Measurements
We’re looking for the SNR — Signal to Noise Ratio of the full frame camera at ISO 1600 vs the Micro Four Thirds camera at ISO 400 (same total light reaches each sensor).
At ISO 400 (“measured 352; manufacturer 800) the Olympus produces an SNR of 31.1 dB.
At ISO 1600 (“measured 1149”) the Canon produces an SNR of 30.2 dB.
The Micro Four Thirds camera has an SNR advantage of +0.9 decibels.
Comparison — Color Fidelity
Using the same ISO settings as before…
The Olympus produces 19.7 bits of color sensitivity.
The Canon produces 18.5 bits of color sensitivity.
The Micro Four Thirds camera wins again with +1.2 bits of additional color sensitivity.
I honestly had no idea what the result would be when I chose these two cameras to compare. It’s surprising that the crop sensor camera has both better SNR and color sensitivity.
The Black Swan — A Real World Example
For the 2010 movie Black Swan, cinematographer Matthew Libatique used the APS-C Canon 7D for some of the subway scenes. He chose the 7D over the full frame 5D because the smaller sensor gave him a greater depth of field.
We used a Canon 7D or 1D Mark IV for all the subway scenes; I could just carry a 7D and shoot on the subway all day with a very small crew. … The 7D has more depth of field than the 5D, but I needed that because I didn’t have a follow-focus unit and needed to work really fast. I shot everything documentary-style. I did all the focus pulls by hand, and we’d just look at it on the camera’s monitor. I ended up shooting on a Canon 24mm lens at 1,600 ASA to get as much depth of field as possible at a stop of T81⁄2.
https://theasc.com/ac_magazine/December2010/BlackSwan/page1.html
Caveats and Criticisms
1. A shallower depth of field
Low Light Example — On m43 f/1.4, ISO 400, 1/50th. The equivalent full frame photo is f/2.8, ISO 1600, 1/50th.
On the full frame camera I have the option to shoot at a shallower depth of field to get a lower ISO — go to f/1.4 to shoot at ISO 400 (the m43 equivalent would be f/0.7). Whether the cost to depth of field is worth the improvement to ISO is up to you.
2. Neutral Density
Daylight Example — I can reduce the amount of light reaching the sensor with a neutral density filter on Micro Four Thirds and still have less gear to carry up the mountain than full frame.
I discuss this in my article “Why Full Frame cameras are better than Crop Sensor cameras for light control.”
3. Speed Boosters
Speed boosters can be used on crop sensor cameras to effectively increase the sensor size relative to the absolute aperture, giving you a shallower depth of field and greater light gathering.
Thoughts and Conclusions
There are characteristics of each sensors size that do have implications for which camera you want to use in daylight and which you want to use in low light.
Full Frame cameras are better at making the scene darker. Between the smaller f-stop and lower base ISO, full frame cameras are better at controlling light by making a scene darker. See the “Further Reading” article for a real world example.
Crop Sensor Cameras are better at making the scene brighter. Between the larger f-stop and higher base ISO, crop sensor cameras are better at making scenes brighter.
Are Micro Four Thirds cameras “better” at low light than full frame? Not necessarily, and that’s not quite the argument I set out to prove. I set out to disprove two specific myths. “Any camera can get the shot in daylight.” and “Full frame cameras are better in low light.”
How did this argument fare?
- Full frame cameras are better at daylight than crop sensor cameras. To get the “same shot” on a crop sensor camera you’ll need to add ND.
- At moderate apertures, crop sensor cameras hold up respectably well in low light, as measured by both SNR and Color Fidelity. Where they lose to full frame is that full frame cameras can shoot at a shallower depth of field to gather more light.
I hope I’ve presented enough evidence to at least challenge those traditional assumptions.
Further Reading
Why Full Frame cameras are better than Crop Sensors at Controlling Light for a practical real-world example.
