X-Git-Url: https://git.sesse.net/?p=movit;a=blobdiff_plain;f=ycbcr_422interleaved_input.h;h=a8c77738f752a32866f6b759ca2b564164f96c9a;hp=b7000a04166a0ee13fd89e96b6481659efd62e74;hb=f216b7bef5a968c89f6fc78e83cc26a91e504a8a;hpb=9c12e38b7cd88a77ef297d080b7c41e6bd6326fb

diff --git a/ycbcr_422interleaved_input.h b/ycbcr_422interleaved_input.h
index b7000a0..a8c7773 100644
--- a/ycbcr_422interleaved_input.h
+++ b/ycbcr_422interleaved_input.h
@@ -28,6 +28,11 @@
 // RGBA texture (from which we sample chroma). We throw away half of the color
 // channels each time, so bandwidth is wasted, but it makes for a very
 // uncomplicated shader.
+//
+// Note that if you can shuffle your data around very cheaply on the CPU
+// (say, while you're decoding it out of some other buffer anyway),
+// regular YCbCrInput with YCBCR_INPUT_SPLIT_Y_AND_CBCR will probably be
+// more efficient, as it doesn't need this bandwidth waste.
 
 #include <epoxy/gl.h>
 #include <string>