the implementation was following my early commit for Math node
I haven't had a chance to run those through Brecht, but would like to do eventually. (they work fine though)
- Add check for header field in BMP decoder. This is needed to distinguish
whether file is indeed BMP image or not.
Without this check Blender could easily crash when it'll try to load
non-BMP image.
Tested with files from own HDD, but all of them has got BM header field,
more testing would be welcome.
- Made Jpeg2000 aware of J2K codec. Originally was needed to verify .j2c
files here in the studio, but having support of this codec would be
nice in general.
Currently supports only reading in this codec, writing would still
using jp2 codec.
- Even preserves thickness but can give unsightly loops
- Smooth gives nicer shape but can give unsightly feather/spline mismatch for 'S' shapes created by beziers.
This is an example where smooth works much nicer.
http://www.graphicall.org/ftp/ideasman42/mask_compare.png
- Make FFmpeg initialization called from creator, not from functions
which requires FFmpeg. Makes it easier to follow when initialization
should happen.
- Enable DNxHD codec. It was commented a while ago due to some strange
behavior on some platforms. Re-tested it on Linux and Windows and
it seemd to be working quite nice. Would let it be tested further,
if it wouldn't be stable enough, easy to comment it again.
- Make non-error messages from writeffmpeg.c printed only if ffmpeg
debug argument was passed to blender. Reduces console pollution
with messages which are not useful for general troubleshooting.
Error messages would still be printed to the console.
- Show FFmpeg error message when video stream failed to allocate.
makes it easier to understand what exactly is wrong from Blender
interface, no need to restart blender with FFmpeg debug flag and
check for console messages.
Used custom log callback for this which stores last error message
in static variable. This is not thread safe, but with current
design FFmpeg routines could not be called form several threads
anyway, so think it's fine solution/
Its unlikely you want to do short -> int, int -> float etc, conversion during swapping (if its needed we could have a non type checking macro).
Double that the optimized assembler outbut using SWAP() remains unchanged from before.
This exposed quite a few places where redundant type conversion was going on.
Also remove curve.c's swapdata() and replace its use with swap_v3_v3()
checked this function against the previous method using random rotation inputs and compared results, while this isnt exactly the same the results are very close and acceptable in both cases, also checked baking actions that the resulting FCurves are good and give matching rotations.
