I was wondering: how is it possible for a smaller sensor to perform better in low-light, if it is capturing much less light than the bigger sensor?
Short answer: We don't know that it is capturing much less light than the bigger sensor.
Smaller sensors capture less light than larger sensor when certain assumptions are made:
- Both cameras have lenses with the same field of view and are framing the same scene at the same scene brightness
- Both cameras are used at the same f-number
- Both cameras are used at the same shutter duration/exposure time ("shutter speed")
- The resulting images from both cameras use the full area of each of their respective sensors.
If any of these things change in the right direction, then the smaller sensor might be allowed to capture more light than the larger sensor.
In the case of a 1.6X APS-C camera and a Micro Four Thirds camera, the ratio of sensor area is approximately 1.5:1 (332mm² to 225mm² is actually 1.4755556:1). If both cameras were used with lenses that provided the same field of view of the same scene using the same f-number and exposure time, the APS-C camera would collect about 50% more light than the µ4/3 camera.
Note that the difference between APC-S and µ4/3 is nowhere near the difference between full frame (FF) and 1.6X APS-C, which is 2.6:1 (864mm²:322mm²). So while a FF sensor is 160% larger (about 1.4 stops) than the area of an APS-C sensor, the APS-C sensor is only 50% larger (about 0.58 stops) than the µ4/3 sensor.
When the two cameras used in the example are used to shoot HD or 4K video, though, neither of them uses the full surface area of their respective sensors in the resulting images. This is because both 4K and HD video use different aspect ratios than either of these cameras' sensors.
- The Panasonic Lumix G7 has a 4592x3448 pixel sensor with an aspect ratio of 4:3 (1.33:1) which is 17.3mm x 13mm for a pixel pitch of 3.77µ
- The Canon EOS M50 has a 6000x4000 pixel sensor with an aspect ratio of 3:2 (1.5:1) which is 22.3mm x 14.9mm for a pixel pitch of 3.72µ
- 4K video is 3840x2160 with an aspect ratio of 16:9 (1.77:1)
- HD video is 1920x1080 with an aspect ratio of 16:9 (1.77:1)
Please note that the size of each camera's individual pixels is almost exactly the same, it's just that the larger sensor of the M50 has more pixels than the smaller sensor of the G7 by roughly the same ratio as their respective areas. Also note that the difference between their widths is more than the difference between their heights.
- 23.2mm is 1.34X as wide 17.3mm
- 14.9mm is 1.15X as tall as 13mm
If both cameras used the full width of their sensors for 4K and HD video, they would not be able to use their full height and produce output at a 16:9 aspect ratio.
- At 16:9 if the G7 used all of its 4592 pixel width, it could only use 2583 of its 3448 lines. That means the area of the sensor being used would only be 17.3mm x 9.73mm for a total area of 168.48mm².
- At 16:9 if the M50 used all of its 6000 pixel width, it could only use 3375 of its 4000 lines. That means the area of the sensor being used would be only 22.3mm x 12.54375mm for an area of 279.73mm².
- The ratio of 279.73mm² used by the M50 to 168.48mm² used by the G7 is 1.66:1, or a slight bit less than 3/4 stops.
The M50 records 4K video (3840 x 2160 pixels) by using the center 3840 x 2160 pixels of its sensor, a 1.5602X (horizontal) crop factor. The 3840 x 2160 pixels in the center of the sensor cover about 14.272mm x 8.046mm. The resulting 114.83mm² area is smaller than the area used by the G7! It turns out that the G7 slightly crops its 4K video as well. It uses about 4130 x 2323 pixels, which computes to 15.56mm x 8.75mm for an area of 136.15mm². That's still about 18.57% larger than the area used by the M50 for 4K video.
So in the case of 4K video, the Panasonic Lumix G7 uses more surface area of its sensor than the Canon EOS M50 does! All other things being equal (which does not appear to be the case here - see below), the G7 would collect more light shooting 4K video than the M50.
In the case of HD video, the EOS M50 seems to use its full width, as does the Panasonic G7. So for full HD video, the 1.66X larger sensor area used by the M50 should collect more light, right?
Not necessarily. Let's look at the example image comparing the two.
- The two images were not taken from the same camera position.
- The background wall to the right of the main subject is darker in the image on the right than it is in the image to the left.
- There's an additional light source and shadow in the background to the left of the main subject in the image on the right.
- The same areas of the table appear to be darker in the image on the right than in the image on the left.
- The lid of the jar appears to be darker in the image on the right than in the image on the left.
- The circle of light below the jar is brighter in the image on the left than in the image on the right.
From these observations we should be able to conclude that the image on the right was exposed brighter than the image on the left. Possible ways this occurred:
- The aperture of the lens on the G7 was set at a lower f-number than the aperture on the M50
- The exposure time for the G7 was longer than the exposure time for the M50
- A combination of both
If the aperture, exposure time ("shutter speed"), or combination of both used by the G7 was more than three-quarters (3/4) stops brighter than those used by the M50, then the 168.48mm² used by the G7 to record FHD video would have collected more light than the 279.73mm² used by the M50 to record FHD video.
All else being equal, an image exposed brighter will show less noise than an image exposed darker. In the case of in-camera video processing, this can be compounded by the different ways each camera does noise reduction.
Unfortunately, we do not know enough specifics to conclude exactly why the sample images differ in the way they do. Things we would need to know:
- The actual focal lengths of each lens used
- The actual aperture (f-number) used by each camera
- The actual exposure time ("shutter speed") used by each camera
- The actual distance each camera was from each of the light sources illuminating the scene
- The difference in brightness, if any, in the scene between when each frame was captured. Where any of the light sources flickering with the phase of AC current? Were the lights inside the jars cheap LEDs that typically flicker?