Yellow_
20th April 2012, 09:54
Is JFIF always full range chroma?
My query revolves around h264 in a container (MOV & mp4) with fullrange flag set 'on' encoded with a BT601 matrix, being treated as JFIF by NLE and yuvj420p by FFmpeg. That is immediate luma squeeze to 16 - 235 to maintain levels when converted to RGB for processing in the NLE.
Although I intend looking at this myself at some point, can anyone suggest problems that may be seen in video decoding full range luma encoded using a BT601 matrix interpreted as JFIF and any tidbits on why JFIF is used. ie:full levels in YCbCr.
Do you ever see, jpeg or JFIF in metadata like we see BT601 BT709 in the header to a stream? JFIF appears to use centre pixel alignment.
Sample here for anyone wanting to read header.
http://www.yellowspace.webspace.virginmedia.com/Gold.aMOV
Other theory, is that the camera firmware didn't offer that choice for flagging that matrix or some custom matrix used, so used the fullrange flag to force JFIF behavior in codec decompression?
My query revolves around h264 in a container (MOV & mp4) with fullrange flag set 'on' encoded with a BT601 matrix, being treated as JFIF by NLE and yuvj420p by FFmpeg. That is immediate luma squeeze to 16 - 235 to maintain levels when converted to RGB for processing in the NLE.
Although I intend looking at this myself at some point, can anyone suggest problems that may be seen in video decoding full range luma encoded using a BT601 matrix interpreted as JFIF and any tidbits on why JFIF is used. ie:full levels in YCbCr.
Do you ever see, jpeg or JFIF in metadata like we see BT601 BT709 in the header to a stream? JFIF appears to use centre pixel alignment.
Sample here for anyone wanting to read header.
http://www.yellowspace.webspace.virginmedia.com/Gold.aMOV
Other theory, is that the camera firmware didn't offer that choice for flagging that matrix or some custom matrix used, so used the fullrange flag to force JFIF behavior in codec decompression?