Blocking artifacts with v1.4.3 and v1.4.4
Posted: Sun Feb 05, 2012 7:32 pm
I have uploaded two .ts file to your ftp site:
test2.ts is the original sample clip.
test2_cut_1.4.4.ts is a sample edit (frames 183 to 427) made with SmartCutter v1.4.4.
The edited file has very bad blocking artifacts at the beginning of the 2nd GOP (frames 19 onwards) when played back on ffdshow tryouts rev3771 and also on a hardware PVR using a Broadcomm H.264 decoder.
It looks like the decoder is using the wrong reference frames to re-construct the image.
I get the same results with v1.4.3 (in fact the cut .ts files are bit identical). I have not been able to try older versions.
The cut-in point (frame 183) is the 2nd frame of a GOP. If I make the cut-in point one frame earlier, i.e. cut-in on the GOP boundary (cut frames 182-427) I still get the artifacts. If, however, I cut from the start of the next GOP (i.e. cut frames 202-427), there are no artifacts. Perhaps that it because frame 202 is an I-frame.
test2.ts is the original sample clip.
test2_cut_1.4.4.ts is a sample edit (frames 183 to 427) made with SmartCutter v1.4.4.
The edited file has very bad blocking artifacts at the beginning of the 2nd GOP (frames 19 onwards) when played back on ffdshow tryouts rev3771 and also on a hardware PVR using a Broadcomm H.264 decoder.
It looks like the decoder is using the wrong reference frames to re-construct the image.
I get the same results with v1.4.3 (in fact the cut .ts files are bit identical). I have not been able to try older versions.
The cut-in point (frame 183) is the 2nd frame of a GOP. If I make the cut-in point one frame earlier, i.e. cut-in on the GOP boundary (cut frames 182-427) I still get the artifacts. If, however, I cut from the start of the next GOP (i.e. cut frames 202-427), there are no artifacts. Perhaps that it because frame 202 is an I-frame.