I also tried curves and found some strange things (see attached picture, 100% crop at right border of image):- Some unprocessed vertical band at right side of image. Maybe, skippings of last 40 columns is too much. The same things reported in russian forum user of S3IS, his image:
switch(conf.curve_enable) {... case 2: case 3: // +1EV, +2EV if (drcurve_loaded) drcurve_apply( conf.curve_enable >> 1 );
Looking at this code to reduce memory use, I noticed a couple things:in curve_apply()Code: [Select] switch(conf.curve_enable) {... case 2: case 3: // +1EV, +2EV if (drcurve_loaded) drcurve_apply( conf.curve_enable >> 1 );the >> is useless since 10b >> 1 == 11b >> 1perhaps &1 does what was intended ?
alsoCode: C f read(drcurve0, 1, 4*CURVE_SIZE*sizeof(unsigned short), fd); // load a set of 4 curves - only 2 are used Any reason not to just load 2 ?
finished(); return (fd >= 0); }#ifdef OPT_CURVES if (conf.curve_enable) curve_apply();#endif return 0;
Hrm:raw.c 203 Code: [Select] finished(); return (fd >= 0); }#ifdef OPT_CURVES if (conf.curve_enable) curve_apply();#endif return 0;if raw is enabled, curve is not applied to jpeg ?
Thanks for that. I saw there were several posts, but wasn't clear which ones had been merged.For the column fix, this processes the entire raw buffer, potentially modifying the border pixels, correct ? Any idea whether these becoming non-black affects canon processing ?