What is “10-Bit 4:2:2”?
Earlier this זquick video videographer Griffin Hammond (executive producer of Indy Mogul) looks at a topic which is often mentioned but much less often explained – 8bit vs. 10bit color and chroma subsampling.
The upcoming Panasonic GH5 which we saw as a prototype in Photokina just a few weeks back will come with a 10bit 4:2:2 – what does that mean exacly? well, Hammond tries to explain in this video and we shall try and help.
In general the 8bit/10bit/12bit etc. has to do with how a camera captures color information. The GH4 for example captures 8bit color or about 16 million color combos while the upcoming 10bit GH5 will have 1 billion color combos. Sound amazing? well yes and no. First – most computer screens can’t show 10bit color so you can shoot but can’t see it (YouTube also will not show it). However if you are a pro color specialist doing color grading – this can help you as well as if you are doing a lot of green screen work.
The second topic Hammond talks about is chroma subsampling. The idea is actually simpler than it might look initially (well, it is complex but the general concept isn’t that much). The color of the image as it is captured by the cameras sensor isn’t sampled and stored per pixel (which is very taxing on the camera) instead researchers discovered a way to sample neighbouring pixels in a way that will best represent the color of each of them.
The difference between 4:4:4, 4:2:2 and 4:2:0 is basically how you choose these neighbouring pixels. You can think of 4:4:4 as RAW where you actually do look at the color of each individual pixel (this isn’t really used much apart from very high end cameras as it is typically considered unnecessary and hard on the hardware). In 4:2:2 every two pixels share information while with 4:2:0 every 4 pixels share information – so a lot more color information is lost. You can see this quite more cleraly in the video below.
A video explaining the basic concept of chroma subsampling