Open xpol opened 7 years ago
Excellent question. During decode, we do keep 2 rows of a framebuffer in memory and fill them with data. And we take each channel into pixel space to predict the DC. It wouldn't be hard to persist the pixel-space values beyond the two rows into a whole framebuffer. Then, the end portion with the JPEG recode could be lopped off without too much trouble. This wouldn't be a huge change, actually--but it would require some code
Can I read *.lep files directly into pixels array without restore them back to jpg files? Any examples?