Member av_samples_alloc (uint8_t **audio_data, int *linesize, int nb_channels, int nb_samples, enum AVSampleFormat sample_fmt, int align)

return the size of the allocated buffer in case of success at the next bump

Member av_samples_fill_arrays (uint8_t **audio_data, int *linesize, const uint8_t *buf, int nb_channels, int nb_samples, enum AVSampleFormat sample_fmt, int align)

return minimum size in bytes required for the buffer in case of success at the next bump

Member avcodec_fill_audio_frame (AVFrame *frame, int nb_channels, enum AVSampleFormat sample_fmt, const uint8_t *buf, int buf_size, int align)

return the size in bytes required to store the samples in case of success, at the next libavutil bump

Member avformat_find_stream_info (AVFormatContext *ic, AVDictionary **options)

Let the user decide somehow what information is needed so that we do not waste time getting stuff the user does not need.

Member AVOption::help

What about other languages?

File f_ebur128.c

implement start/stop/reset through filter command injection

support other frequencies to avoid resampling

Member ff_rv34_get_start_offset (GetBitContext *gb, int blocks)

Maybe replace with ff_h263_decode_mba() ?

Member ff_rv34_get_start_offset (GetBitContext *gb, int blocks)

Maybe replace with ff_h263_decode_mba() ?

File jacosubdec.c

Support P[ALETTE] directive.

File libmodplug.c

better probing than extensions matching

Class OWDenoiseContext

try to change to int

try lifting based implementation

optimize optimize optimize

hard thresholding

use QP to decide filter strength

wavelet normalization / least squares optimal signal vs.

Class RTSPState

Use AVIOContext instead of URLContext

File src_movie.c

use direct rendering (no allocation of a new frame)

support a PTS correction mechanism

File vaf_spectrumsynth.c

support float pixel format

Module VC-1 VLC tables and defines

TODO move this into the context

Module VC-1 VLC tables and defines

TODO move this into the context

Class VC1Context

Change size wherever another size is more efficient Many members are only used for Advanced Profile

File webvttdec.c

need to support extended markups and cue settings

Class weighted_avg

SIMD for compute_safe_ssd_integral_image
SIMD for final weighted averaging
better automatic defaults? see "Parameters" @ http://www.ipol.im/pub/art/2011/bcm_nlm/
temporal support (probably doesn't need any displacement according to "Denoising image sequences does not require motion estimation")
Bayer pixel format support for at least raw photos? (DNG support would be handy here)
FATE test (probably needs visual threshold test mechanism due to the use of floats)