I thought it would speed up, but it didn't almost.
* Time elapsed: 0:28.380 - 7.08243284105080928015 FPS old* Time elapsed: 0:28.275 - 7.10862776131904450239 FPS new
(The test has problem...)
After all, the calculation area is small (area is 200), not the whole picture.
However, there is an interesting fact. Between last version and now, althougth the output of 8bit or 10bit is same, while the output of 16bit is different. For 16bit, the output in new verison is whitish.
I thought it would speed up, but it didn't almost.
* Time elapsed: 0:28.380 - 7.08243284105080928015 FPS old* Time elapsed: 0:28.275 - 7.10862776131904450239 FPS new(The test has problem...)
After all, the calculation area is small (area is 200), not the whole picture.
However, there is an interesting fact. Between last version and now, althougth the output of 8bit or 10bit is same, while the output of 16bit is different. For 16bit, the output in new verison is whitish.