Fix #111703: banding and color shift when decoding some videos

Tell ffmpeg swscale to do accurate YUV->RGB conversion, instead of
slightly faster but not really accurate one. Fixes banding and some
color shifts in video files, particularly in dark regions.

The accurate conversion is a bit slower though, on 4K resolution video,
time taken to convert video frame from YUV to RGB:

- x64 (Ryzen 5950X): 2.3ms -> 3.7ms
- arm64 (M1 Max): 0.6ms -> 2.9ms

My take is that paying 1-2ms per 4K video playback is acceptable
since the result is obviously "more correct" and matches what VLC/ffplay
produces.

From what I can tell, "accurate conversion" turns off some dedicated
assembly code paths within ffmpeg. Maybe someday ffmpeg would get
accurate and assembly-optimized routines for that.

With more accurate decoding, we can now lower the expected render
test threshold again, since x64 & arm64 decoding is much closer now.

Comparison screenshots in the PR.

Pull Request: https://projects.blender.org/blender/blender/pulls/130383

source/blender/imbuf/intern/anim_movie.cc

@@ -414,14 +414,18 @@ static int startffmpeg(ImBufAnim *anim)
         1);
   }
 
+  /* Use full_chroma_int + accurate_rnd YUV->RGB conversion flags. Otherwise
+   * the conversion is not fully accurate and introduces some banding and color
+   * shifts, particularly in dark regions. See issue #111703 or upstream
+   * ffmpeg ticket https://trac.ffmpeg.org/ticket/1582 */
   anim->img_convert_ctx = BKE_ffmpeg_sws_get_context(anim->x,
                                                      anim->y,
                                                      anim->pCodecCtx->pix_fmt,
                                                      anim->x,
                                                      anim->y,
                                                      anim->pFrameRGB->format,
-                                                     SWS_BILINEAR | SWS_PRINT_INFO |
-                                                         SWS_FULL_CHR_H_INT);
+                                                     SWS_POINT | SWS_FULL_CHR_H_INT |
+                                                         SWS_ACCURATE_RND);
 
   if (!anim->img_convert_ctx) {
     fprintf(stderr,

tests/python/sequencer_render_tests.py

@@ -50,7 +50,7 @@ def main():
     report = render_report.Report("Sequencer", output_dir, oiiotool)
     report.set_pixelated(True)
     # Default error tolerances are quite large, lower them.
-    report.set_fail_threshold(3.0 / 255.0)
+    report.set_fail_threshold(2.0 / 255.0)
     report.set_fail_percent(0.01)
     report.set_reference_dir("reference")