After researching a bit more, I found that 4:4:4 means no subsampling is applied. and 4:2:2 and 4:2:0 are represented as follows. After the JPEG is created, the Y'CbCr color space allows for efficient encoding by applying these transforms to the chrominance channels. What is interesting is that the image can be further compressed through methods of discrete cosine transforms and quantization.
randyen
One question that I was how does the 4:2:2 and the 4:2:0 representations differ in the images they end up coloring? Although I can see how the representations differ in the slides, I have trouble visualizing it for an actual image. Additionally, I wonder if there are any other variants such as a 4:2:1 or anything else which one could utilize to get a desired chroma representation.
Edge7481
I think i'm not understanding the notation. How does 2:2 correspond to half resolution in the horizontal dimension while 2:0 corresponds to half resolution in both dimensions?
oliver-ni
These numbers make no sense to me — same as Edge7481. I read a bit on Wikipedia and it seems like it's J:a:b where a is the number of horizontal samples in a Jx2 section, and b is the number of vertical changes (so a or 0), but that seems really arbitrary and not able to encode that many variants. Why is it like this?
After researching a bit more, I found that 4:4:4 means no subsampling is applied. and 4:2:2 and 4:2:0 are represented as follows. After the JPEG is created, the Y'CbCr color space allows for efficient encoding by applying these transforms to the chrominance channels. What is interesting is that the image can be further compressed through methods of discrete cosine transforms and quantization.
One question that I was how does the 4:2:2 and the 4:2:0 representations differ in the images they end up coloring? Although I can see how the representations differ in the slides, I have trouble visualizing it for an actual image. Additionally, I wonder if there are any other variants such as a 4:2:1 or anything else which one could utilize to get a desired chroma representation.
I think i'm not understanding the notation. How does 2:2 correspond to half resolution in the horizontal dimension while 2:0 corresponds to half resolution in both dimensions?
These numbers make no sense to me — same as Edge7481. I read a bit on Wikipedia and it seems like it's J:a:b where a is the number of horizontal samples in a Jx2 section, and b is the number of vertical changes (so a or 0), but that seems really arbitrary and not able to encode that many variants. Why is it like this?