Re: [Mlt-devel] More experiments with Java bindings - toying with frames
Brought to you by:
ddennedy,
lilo_booter
From: Charles Y. <cha...@pa...> - 2004-08-08 20:13:51
|
On Sun, 2004-08-08 at 21:26, Stepan Roh wrote: > second branch ("image" property) - should return actual format (= > "format" property) for actual image Agreed - the code on the second call to get_image is definitely wrong because it lacks the format on properties. > third branch (test card producer) - 1. has test frame - should set > "format" (I see that there is a hardcoded yuv422 dependency: *width * > *height * 2 which is not good), 2. has no test frame - calls itself again > - falls into the fourth branch The width * height * 2 thing is wrong - the byte size of the image is generally ignored at the moment, but this should be fixed (in fact, all that needs to be done is to obtain the image and size from the properties of the test frame and set that on the requested frame). > fourth branch (blank picture) - generates blank image in desired format so > it should set "format" property > > I think the change I proposed goes towards consistency, but as yuv422 > seems hardcoded on more places it won't change much :-) Heh - well, timescales were short :-) - we worked to a spec which basically required yuv422 throughput - the fact that there's any hooks in for more colour spaces was over and above requirements and hence secondary. Many thanks for your analysis - much appreciated. Cheers, Charlie |