On Mon, Feb 8, 2010 at 3:15 PM, Ronald S. Bultje <[email protected]> wrote: > Hi, > > On Wed, Jan 27, 2010 at 4:20 PM, Thomas Worth <[email protected]> wrote: >> CODEC_ID_H264 >> CODEC_ID_FFH264 > > FFH264 (for encoding) would be the native FFmpeg encoder, whereas H264 > (default) would be the x264 encoder. For decoding, it always uses the > built-in decoder. So none of this matters for the subject at hand. > > Regarding your original post, I'm sceptical. A simple delta between > the two images on your blog shows differences (dither-related?) only > around thearea right above the tire. Calling this dithering an > improvement in quality goes way too far, imo.
Yeah, it's starting to look more and more like QuickTime is taking the liberty of adding noise to its decompression output to cover up the compression artifacts. Subjectively, this may increase quality / aesthetic value, but this should be an option. I mean, if I want noise I'll add it myself, thank you. I don't know if there's a way through the QuickTime API to disable this behavior, but there should be. I'm satisfied with the results I'm getting now using VirtualDub and ffdshow. However, I'd like to duplicate this behavior with the ffmpeg binary (and my own code as well), which I can't do right now because some of the libswscale functions are still being worked on. Once that stuff is worked out, I'll resume testing. _______________________________________________ libav-user mailing list [email protected] https://lists.mplayerhq.hu/mailman/listinfo/libav-user
