It is a truism that a large sensor will show less obvious noise in low light than a small sensor. Given equal processing and noise reduction algorithms, a camera with a large sensor and not too densely packed pixels must perform better in low light than a camera with a tiny sensor. That sounds obvious but in practice there is another factor at play in video recording: the lens aperture. Since the exposure for video is constant or virtually so (i.e. 1/50th of a second for 25 frames per second or 1/60th for 30 fps) the aperture of the lens is the determinant of how much light reaches the sensor over that period. More light equals a higher signal:noise ratio and, therefore, less noise. Some people, many online, interpret having to increase the ISO setting, or gain, as the cause of the noise. That is not so. Increasing the ISO setting simple increases the gain of the amplifier to make the image visible; both signal and noise are increased.
The Canon HF G70 has a 1/2.3”* sensor. The maximum aperture varies across the zoom range between f/1.8 and f/2.8. To see how good—or bad—the noise was in low light I compared it first with the main camera on the iPhone 14 Pro (‘main’ f/1.78, said to be size 1/ 1.28"). This test was done in a dim, northerly facing room in daylight. The Exposure Value was 4. In the deepest shadow area there was a little noise and I could not find any difference between the Canon HF G70 and the iPhone. I then compared the HF G70 with my Nikon Z7, which because of the very large number of pixels is known to be noisier than the Z6. I put on a 50mm f/1.8 lens. The Exposure Value was 4.3. Again there was a little noise at f/1.8 (ISO was 800), slightly less perhaps than on the HF G70 and iPhone. However, when I turned the aperture to f/4 (keeping the shutter speed at 1/50 and letting the ISO setting increase to 4000), there was marked noise. In other words, my Nikon Z7 in full-frame mode at f/4 was noisier than the Canon HF G70 and the iPhone 14 Pro at f/1.8. The difference between f/1.8 and f/4 is 1⅓ stops (EVs) but of course a more than doubling of the light hitting the sensor.
In other words we are talking about trade-offs with apertures, sensor size and depth-of-field. One advantage of using a full-frame camera is being able to isolate a subject in a narrow depth-of-field. However, with video, the depth-of-field at f/1.8 is wafer thin. With a 50 mm lens at a distance of 5 metres it is only around 20 cm. Therefore, it is highly likely that I would want to use a smaller aperture. With a smaller sensor, the same depth-of-field can be attained at a larger aperture so what you gain on the swings by using a camera with a larger sensor, in terms of noise, can easily be lost on the roundabouts of using a smaller aperture.
This is all a very long-winded way of saying that I am very pleased with the performance of the Canon HF G70 in low light. My guess is that Canon have got it as good as it can be with the current state of technology. All my trials have been with 4k video. I note that Canon in their marketing of this camera stress ‘Over Sampling HD Processing High quality Full HD. Uses the 4K UHD sensor to deliver superior Full HD images’. I haven’t tested that yet.
While I was doing all this I realised something I had not fully taken in before. When my Sony AX-53 (also a 1/2.3” sensor) broke down in use (cash refunded by Amazon) and I unwisely bought the Sony AX-700 just before covid lockdown (camera also broke down in use and just out of extended warranty but with very few minutes on the clock) that the advantages of the larger (the so-called 1” sensor) in terms of noise and of a slight isolation of any subject at larger apertures, were offset at least partially, by the smaller maximum aperture of the lens. The AX-53 is f/2-f/3.8 over its zoom range; the AX-700 f/2.8-f/4.5. The Canon HF G70 as I noted earlier is f/1.8 and f/2.8 (over a much wider zoom range).
Unfortunately, it is not worth trying to show the videos of the trials; the definition here and on YouTube is just not good enough for any difference to be illustrated. It is also worth pointing out that while I always try to take 4K, I normally put this on a 1080 timeline thus giving me the opportunity to zoom in by cropping without losing definition. I did that with all the samples (where I could spot the noise on playback on my 24” Mac with 4K Retina Display). However, when I did as I usually do, output the video as an MP4 file and looked at it on a television (no, not one of those huge things that fill a wall) I could see no noise, except for the slightest flicker evident in the footage from the Z7 at f/4.
*Desperately needed is some international standard for sensor sizes in cameras. The use of fractions of inches and the like can be utterly confusing. We need the dimensions in mm--a simple statement some manufacturers are very reluctant to state. That, together with the number of pixels should be a standard imposed.
Coming Next: Ergonomics and Setup