Hi, Cairo, I am sorry to update my status of my GSoC project so late.<div class="gmail_quote"> At the beginning, I was familiar with cairo nouns and verbs, like destination, source and so on. I spent much time on the cairo and didn't know how to add my code into cairo. Then I sent an email to Soren to discuss the question and knew that my attention should be changed into pixman library. In the project, first I implemented half-precision floats in C from OpenEXR project. There are two functions: half float_to_half(float f); float half_to_float(half h); I want to convert an entire scanline to float, then do the compositing in float directly, then convert the entire scanline back. The most part of the project is to get pixman image composited together. To support scRGB precision, I did: 1) In enum pixman_format_code_t, I added four 64bpp formats: PIXMAN_F_a16r16g16b16, PIXMAN_F_x16r16g16b16     , PIXMAN_F_a16b16g16r16 , PIXMAN_F_x16b16g16r16. 2) I added routines that can convert from arbitrary formats into scRGB and back in pixman-access.c. Then I will add compositing routines in pixman-combine-float.c like pixman-combine32.c and add some supporting code in PIXMAN_COMPOSITE_RECT_GENERAL in pixman-compose.c In the project, I would like to thank Soren for giving me lots of help and advice. BEST WISHES! Guofu </div>