Pixel binning is the process of combining the electric charge from adjacent CMOS or CCD sensor pixels into one super-pixel, to reduce noise by increasing the signal-to-noise ratio in digital cameras.
Typically, the binning happens on groups of four pixels that form a quad (see image) but some sensors can merge a block of up to 4×4 pixels (16 pixels) instead of 2×2 (4 pixels). By doing this, the sensor is increasing the relative sensitivity by 4 (signal to noise ratio), but also reducing the (spacial) resolution by 4. The combined pixels are sometimes called “super-pixels.”
Sensors can also perform 2×1 binning (Horizontal binning) or 1×2 binning (Vertical binning), but these binning patterns are not being used in consumer camera application.
What is the purpose of binning?
The reason for being of binning is to increase the signal-to-noise ratio (SNR or noise reduction), a key metric in analog applications (such as image sensing). In modern camera, this is particularly useful to obtain higher brightness in extreme low-light conditions.
Note that some sensors do average binning, rather than additive binning. It is a different way of getting a better signal to noise ratio, so both techniques lead to a similar goal. It’s not always clear which is the best, and it depends on the context.
With less noise from the analog data, the image can be subjected to higher levels of gains/amplifications during the post-processing phase. This will also offer an opportunity to obtain higher quality low-light images.
Olympus has an interactive binning widget you can look at.
Where does camera noise come from?
Binning dramatically helps reduce noise, but where does noise come from to start with? Mainly 3 possible sources:
- Shot noise
- light captured by the sensor and converted to electron charges
- Read noise
- when we read the sensor’s data while converting from analog to digital
- Thermal noise
- electrons released by the sensor itself. Longer exposures create more heat
In general, Read Noise is negligible, so it’s not a real factor for consumer applications. Thermal noise is not an issue for regular photo ops, and we’ve never seen it become an issue even when recording video. Therefore the Shot Noise is the main issue that binning is designed to address.
If you want to learn more, there’s a fascinating discussion on the SharpCap Forums. For a very deep dive, you can read The Microscopy by Mortimer Abramowitz and Michael W. Davidson (PDF link), jump to page 375.
Optical equipment maker Hamamatsu has a slightly more detailed explanation , while industrial camera maker Basler has a bit more information about Horizontal and Vertical binning.
DPReview Forum user “smorgasbord” talk about some common misperception about binning in compact cameras while Military-applications company Adimec shows hows the math of binning work with a simple schematic.
Xiaodan Jin and Keigo Hirakawa have an in-depth paper titled “Analysis and processing of pixel binning for color image sensor” in which they go over several binning techniques and show the outcome.
Does it work?
Yes, as the research suggests, Pixel Binning (aka 4-in-1 pixel as some OEMs call it) does have some tangible benefits. However, it depends how it is implemented. For example, the LG G7 which is using pixel binning on a 16 MP sensor, ends up with 4 MP photos, which feature less details than alternative solutions (a large sensor and aperture). Read our full LG G7 Camera Review.
The Huawei P20 Pro opts for an extremely large sensor (for a phone) and a 40MP pixel resolution, so when binning is used, the final photo resolution is 10 MP, which is a much better trade-off than 4MP. Read our Full Huawei P20 Pro Camera Review.
In theory, Pixel binning is a great solution to have high megapixel count in bright photo conditions, and great low-light sensitivity. In reality, it does bring some benefits, but not a crushing blow to those that don’t use it. What we know, is that more mobile cameras will include this technique in their photo toolbox.