Original commit message from CVS:
* ext/lame/gstlame.c:
Allow fixing the sample rate lame converts to by negotiating fixed
sample rate on the src pad caps.
Add docs for it.
* tests/check/Makefile.am:
* tests/check/pipelines/lame.c:
Add a check for it.
Original commit message from CVS:
Patch by: Daniel Charles <dcharles at ti dot com>
* ext/amrnb/amrnbenc.c: (gst_amrnbenc_bandmode_get_type),
(gst_amrnbenc_set_property), (gst_amrnbenc_get_property),
(gst_amrnbenc_base_init), (gst_amrnbenc_class_init),
(gst_amrnbenc_chain), (gst_amrnbenc_state_change):
* ext/amrnb/amrnbenc.h:
Add property to select the BandMode of the AMR encoder.
See #466442.
Original commit message from CVS:
* configure.ac:
* ext/mad/gstmad.c: (gst_mad_chain):
Use the new buffer clipping function from gstaudio here and
require gst-plugins-base CVS.
Original commit message from CVS:
* ext/mad/gstmad.c: (_do_init), (gst_mad_init), (index_seek),
(normal_seek), (gst_mad_sink_event), (clip_outgoing_buffer),
(gst_mad_chain), (gst_mad_change_state):
* ext/mad/gstmad.h:
Implement buffer clipping and use GST_BOILERPLATE instead of manual
GType magic. Part one of bug #308312.
Original commit message from CVS:
* configure.ac:
* ext/mpeg2dec/gstmpeg2dec.c: (crop_buffer):
* gst/asfdemux/gstasfdemux.c: (gst_asf_demux_descramble_buffer):
* gst/dvdlpcmdec/gstdvdlpcmdec.c: (gst_dvdlpcmdec_chain_raw):
Fix build against core CVS by not using deprecated API. Bump
requirements for new API (overdue anyway).
Original commit message from CVS:
* ext/amrnb/amrnbdec.c: (gst_amrnbdec_base_init),
(gst_amrnbdec_event), (gst_amrnbdec_chain),
(gst_amrnbdec_state_change):
* ext/amrnb/amrnbdec.h:
* ext/amrnb/amrnbparse.c: (gst_amrnbparse_base_init),
(gst_amrnbparse_loop):
Add newsegment and discont handling. Some code cleanups.
Original commit message from CVS:
* ext/dvdread/dvdreadsrc.c:
If we can't read a nav packet, or it doesn't look like a nav
packet, that's an error. Previous behavior was to look at
the next packet; I can't find any justification for doing that.
Fixes: #358891.
Original commit message from CVS:
Patch by: Laurent Glayal <spglegle at yahoo dot fr>
* ext/amrnb/amrnbdec.c: (gst_amrnbdec_class_init),
(gst_amrnbdec_finalize):
Don't leak the adapter, unref it in a new finalize method instead.
Fixes#448782.
Original commit message from CVS:
* ext/dvdnav/dvdnavsrc.c: (gst_dvd_nav_src_set_property):
Don't forget to unlock when setting the device property, or
we'll sooner or later deadlock.
Original commit message from CVS:
* ext/amrnb/amrnbparse.c: (gst_amrnbparse_get_type),
(gst_amrnbparse_class_init), (gst_amrnbparse_init),
(gst_amrnbparse_pull_header), (gst_amrnbparse_loop):
Fix pull mode operation some more: send newsegment event before
sending data; handle EOS and fatal flow returns a bit better; don't
leak buffers in some cases. Misc. minor cleanups. Fixes#431707.
Original commit message from CVS:
2007-03-27 Julien MOUTTE <julien@moutte.net>
* ext/mad/gstmad.c: (gst_mad_chain): Generate a perfect
stream and try to adapt to incoming timestamps lack of
precision up to half a frame. This allows generating
a perfect stream even with muxed formats like FLV where
the timestamps are stored in milliseconds.
Original commit message from CVS:
* ext/lame/gstlame.c: (gst_lame_setup):
Disable the bitrate checking when the user has requested
Free Format mode, as all bitrates less than the maximum
are valid then.
Original commit message from CVS:
* ext/mad/gstmad.c: (gst_mad_convert_sink), (gst_mad_convert_src):
When handling seeks, don't convert -1 start or stop values to a bogus
byte position - pass them upstream as -1.
Original commit message from CVS:
* ext/cdio/gstcdiocddasrc.c: (gst_cdio_cdda_src_open),
(gst_cdio_cdda_src_finalize):
Make sure we always destroy our libcdio handle.
Original commit message from CVS:
* ext/cdio/gstcdiocddasrc.c: (gst_cdio_cdda_src_probe_devices),
(gst_cdio_cdda_src_read_sector), (gst_cdio_cdda_src_open),
(gst_cdio_cdda_src_finalize):
Small code cleanups.
Don't use pad_alloc as the base class cannot deal with the error codes.
Original commit message from CVS:
Patch by: Zaheer Abbas Merali <zaheermerali at gmail com>
* ext/mpeg2dec/gstmpeg2dec.c: (gst_mpeg2dec_chain):
Never post a (fatal) error message on the bus on decoding errors. We
should only do this if we can't recover, but mpeg2dec can always
recover. This is needed for DVB streams, for example, where there
may be temporary glitches in the stream. Instead of an error
message, post a warning message on the bus for every N decoding
errors. Fixes#370020.
Original commit message from CVS:
* ext/lame/gstlame.c: (gst_lame_class_init), (gst_lame_init),
(gst_lame_setup):
Display sensible defaults and limits for the
vbr-min/max/mean properties. Fix the 'hard-limit' VBR min
property - it's supposed to be a boolean 0/1 value.
Original commit message from CVS:
* ext/lame/gstlame.c:
Initialise the variables so gcc doesn't complain about possibly
uninitialised uses, even though they can't actually happen.
Original commit message from CVS:
* ext/lame/gstlame.c: (gst_lame_sink_setcaps),
(gst_lame_set_property), (gst_lame_setup):
Fix up bitrate checking macro. Make it give us a
GST_ELEMENT_WARNING message so the application has a chance of
reporting this to the user. Move the checking to _setup, so we
are sure it runs in the READY state, when we hope to have a pipeline
and a bus that is not flushing.
This fixes e.g. using 96 kbit/sec as a bitrate.
Original commit message from CVS:
* ext/lame/gstlame.c: (gst_lame_sink_event), (gst_lame_chain),
(gst_lame_change_state):
* ext/lame/gstlame.h:
On receiving EOS, we try to push a last buffer with the remaining
samples. Don't do that if we got an unclean flow return on the last
gst_pad_push(), downstream might not handle this very gracefully
(see #403168).
* gst/mpegaudioparse/gstmpegaudioparse.c: (gst_mp3parse_chain):
Pass flow returns upstream (helps #403168).
Original commit message from CVS:
* ext/lame/gstlame.c: (gst_lame_sink_setcaps), (gst_lame_chain):
warn when outgoing sample rate is different from incoming
Original commit message from CVS:
Based on patch by: Roland Kay <roland.kay at ox compsoc net>
* ext/lame/gstlame.c: (gst_lame_init), (gst_lame_chain),
(gst_lame_setup):
* ext/lame/gstlame.h:
Fix leak (by calling lame_init_params() before lame_close()); handle
NULL return from lame_init() more gracefully. Fixes#385311.
Original commit message from CVS:
Patch by: Sebastian Dröge <slomo@circular-chaos.org>
* ext/mpeg2dec/gstmpeg2dec.c: (gst_mpeg2dec_init),
(gst_mpeg2dec_reset), (gst_mpeg2dec_alloc_sized_buf),
(gst_mpeg2dec_alloc_buffer), (init_dummybuf), (handle_slice):
* ext/mpeg2dec/gstmpeg2dec.h:
Align buffers to a 16 byte boundary so the altivec optimisations
don't crash. Fixes#327350.
Original commit message from CVS:
Patch by: Jan Arne Petersen <jpetersen at jpetersen dot org>
* ext/mpeg2dec/gstmpeg2dec.c: (handle_sequence), (handle_slice):
Simplify the framerate code and support more framerates. Fixes#361584.
Original commit message from CVS:
Patch by: Sebastian Dröge <slomo@circular-chaos.org>
* ext/mpeg2dec/gstmpeg2dec.c: (gst_mpeg2dec_init):
Fix minor memleak when creating pads from static pad templates (#369627).
* tests/check/Makefile.am:
* tests/check/elements/.cvsignore:
* tests/check/elements/mpeg2dec.c: (setup_mpeg2dec),
(cleanup_mpeg2dec), (GST_START_TEST), (mpeg2dec_suite), (main):
Add unit tests for mpeg2dec (#369627).
Original commit message from CVS:
* ext/lame/gstlame.c: (gst_lame_sink_event), (gst_lame_chain),
(gst_lame_change_state):
* ext/lame/gstlame.h:
Make lame timestamp flushed eos buffer by some additional timestamp
accounting. Fixes#374760.
Original commit message from CVS:
Patch by: Sebastian Droege <slomo at ubuntu dot com>
* ext/mpeg2dec/gstmpeg2dec.c: (gst_mpeg2dec_finalize),
(gst_mpeg2dec_alloc_buffer), (init_dummybuf), (handle_slice):
* ext/mpeg2dec/gstmpeg2dec.h:
libmpeg2 requires its output buffers to start at a 16byte aligned
address or the altivec optimizations will explode.
Original commit message from CVS:
* ext/cdio/gstcdio.c: (gst_cdio_get_cdtext):
* ext/cdio/gstcdio.h:
* ext/cdio/gstcdiocddasrc.c: (gst_cdio_cdda_src_open):
Move CD-TEXT utility function into common file so it can also be
used by a future cdioparanoiasrc.
Original commit message from CVS:
* ext/dvdread/dvdreadsrc.c: (gst_dvd_read_src_read):
Use dsi_pack.dsi_gi.vobu_ea correctly, add some more debugging
information (fixes#340433).
Original commit message from CVS:
* ext/dvdread/dvdreadsrc.c: (gst_dvd_read_src_goto_title):
Don't announce audio or subtitle streams that aren't really present,
it makes dvddemux create pads that will never carry any data, which
is not good.
Original commit message from CVS:
* ext/lame/gstlame.c: (gst_lame_set_property):
Round up not allowed bitrates to the next higher allowed one
(Closes: #361140).
Original commit message from CVS:
* ext/lame/gstlame.c: (gst_lame_sink_setcaps),
(gst_lame_set_property), (gst_lame_get_property), (gst_lame_chain),
(plugin_init):
* po/POTFILES.in:
Add i18n magic to lame plugin. Throw decent error message when we
fail to setup the encoder (#361140, 361151); misc. minor clean-ups.
Original commit message from CVS:
* ext/mpeg2dec/gstmpeg2dec.c: (handle_slice):
* ext/mpeg2dec/gstmpeg2dec.h:
Add some debugging to timestamp handling.
Make sure we don't convert invalid timestamps.
Original commit message from CVS:
* ext/mad/gstid3tag.c:
* gst/mpegstream/gstrfc2250enc.h:
* gst/synaesthesia/gstsynaesthesia.c:
More G_OBJECT macro fixing.
Original commit message from CVS:
Patch by: Yves Lefebvre <ivanohe at abacom dot com>
* ext/mpeg2dec/gstmpeg2dec.c: (init_dummybuf):
Fix initialisation dummy buffer, the offsets are relative to the start
of the buffers. Fixes#356004.
Original commit message from CVS:
* ext/dvdread/dvdreadsrc.c: (gst_dvd_read_src_read):
Add convert query (needed for later when we just operate in time
format and let the base source class handle all the seeking stuff).
Original commit message from CVS:
* ext/a52dec/gsta52dec.c: (gst_a52dec_mode_get_type),
(gst_a52dec_class_init), (gst_a52dec_init), (gst_a52dec_channels),
(gst_a52dec_handle_frame), (gst_a52dec_change_state),
(gst_a52dec_set_property), (gst_a52dec_get_property):
* ext/a52dec/gsta52dec.h:
Patch from from Michal Benes <michal.benes@itonis.tv>:
Add two things to a52dec: configure the exact output format for ac3
decoding through properties, if desired.
By default, configure an output format preferred by downstream. Now
that audioconvert lists caps by preference, this means that a52dec
can do downmixing (iff required) rather than audioconvert, so it can
use the ac3 downmix levels from the bitstream.
Original commit message from CVS:
* ext/lame/Makefile.am:
* ext/mpeg2dec/Makefile.am:
* gst/dvdlpcmdec/Makefile.am:
* gst/dvdsub/Makefile.am:
* gst/mpegaudioparse/Makefile.am:
Add missing GST_LIBS to the link flags
Original commit message from CVS:
Patch by: Michal Benes <michal dot benes at itonis dot tv>
* ext/mad/gstmad.c: (gst_mad_chain):
Fix timestamping in mad by only activating a new timestamp when the
previous frame has been decoded. Fixes#350723.
Also clean up some of the non fatal warnings when the input buffer is
too small to decode a header.
Original commit message from CVS:
Patch by: Wim Taymans <wim at fluendo dot com>
* ext/a52dec/gsta52dec.c: (gst_a52dec_sink_event):
Instead of forwarding the event in the wrong format, unref it and set
a boolean that we still need to send it in the streaming thread.
Fixes#350554
Original commit message from CVS:
Patch by: Edward Hervey (edward at fluendo dot com)
* ext/dvdread/dvdreadsrc.c: (plugin_init):
Put debug category initialization before use of GST_DEBUG, in order
to remove a fatal warning. Fixes#350895
Original commit message from CVS:
* ext/dvdread/dvdreadsrc.c: (gst_dvd_read_src_goto_title):
Error out properly if this is an interactive DVD (fixes#345694).
Original commit message from CVS:
* ext/a52dec/gsta52dec.c: (gst_a52dec_chain):
When the first_access is 1 or 0, we should copy the timestamp from
the incoming buffer if there is one.
Original commit message from CVS:
* ext/a52dec/gsta52dec.c: (gst_a52dec_channels), (gst_a52dec_push),
(gst_a52dec_reneg), (gst_a52dec_handle_frame), (gst_a52dec_chain):
Use GST_ELEMENT_ERROR for error reporting; post errors in some
more places; remove superfluous newlines from debug statements.
Original commit message from CVS:
* ext/dvdread/dvdreadsrc.c: (gst_dvd_read_src_stop),
(gst_dvd_read_src_goto_chapter),
(gst_dvd_read_src_get_chapter_starts),
(gst_dvd_read_src_goto_title), (gst_dvd_read_src_get_next_cell),
(gst_dvd_read_src_get_time_for_sector),
(gst_dvd_read_src_get_sector_from_time), (gst_dvd_read_src_read),
(gst_dvd_read_src_handle_seek_event), (gst_dvd_read_src_do_seek),
(gst_dvd_read_src_goto_sector):
* ext/dvdread/dvdreadsrc.h:
Add basic support for time-based seeking; set timestamps on
outgoing buffers if we have them; create table with
chapter to time mapping when opening a title; rename
gst_dvd_read_src_get_next_cell_for() to _get_next_cell() and
make it take an explicit pgc argument; fix up some debugging
messages so that title/chapter numbers are printed as starting
from 1 for easier readability.
Original commit message from CVS:
* ext/dvdread/dvdreadsrc.c: (plugin_init):
* po/POTFILES.in:
Make custom error messages translatable.
* gst/asfdemux/gstasf.c: (plugin_init):
Remove setlocale() call, doesn't seem to be needed or recommended for
plugins, at least not according to gstreamer/docs/random/i18n.
Original commit message from CVS:
2006-07-26 Zaheer Abbas Merali <zaheerabbas at merali dot org>
* ext/lame/gstlame.c: (gst_lame_setup):
Fix lame putting lots of 0's at start of mp3. Fixes bug #348786.
Original commit message from CVS:
* ext/dvdread/dvdreadsrc.c: (gst_dvd_read_src_init),
(gst_dvd_read_src_is_seekable), (gst_dvd_read_src_class_init),
(gst_dvd_read_src_stop), (gst_dvd_read_src_goto_title),
(gst_dvd_read_src_create), (gst_dvd_read_src_handle_seek_event),
(gst_dvd_read_src_do_seek), (gst_dvd_read_src_src_event):
* ext/dvdread/dvdreadsrc.h:
Rewrite seeking code and make seeking in DVDs work (#337834).
Original commit message from CVS:
* ext/a52dec/gsta52dec.c: (plugin_init):
Call the channel positions get_type() method in plugin_init
to ensure that it isn't simultaneously called later from
multiple threads.
Original commit message from CVS:
* ext/a52dec/gsta52dec.c: (gst_a52dec_channels), (plugin_init):
Treat dual-mono as stereo. It should really be output on 2 separate
pads, but isn't for now.
Original commit message from CVS:
* ext/mpeg2dec/gstmpeg2dec.c: (gst_mpeg2dec_negotiate_format):
Assume pixel-aspect-ratio of 4:3 if libmpeg2dec doesn't give us
any PAR whatsoever (rather than using 0/0, which will lead to
an abort later on). Fixes#345184.
Original commit message from CVS:
Patch by: Michael Dominic K. < mdk at mdk org pl>
* ext/mad/gstmad.c: (gst_mad_chain):
Send newsegment event before calling
gst_pad_alloc_buffer_and_set_caps(), makes mad behave better
in connection with pad blocking (#342594). While we're at it,
do some minor clean-ups.
Original commit message from CVS:
* ext/a52dec/Makefile.am:
... and then he said "When you go and fix the build you
better make sure you have some spare backslashes in your
pocket".
Original commit message from CVS:
* ext/sidplay/gstsiddec.cc:
Fix copyright, email addresses and descriptions.
Use saner defaults for arguments. Fixes#344667.
constify some stuff.
Fix memleaks.
Add tags.
Fix negotiation to do mono/44100 by default.
Post error messages.
Use _scale_int where possible.
Original commit message from CVS:
* ext/mpeg2dec/gstmpeg2dec.c: (gst_mpeg2dec_finalize),
(gst_mpeg2dec_reset), (clear_buffers), (handle_picture),
(gst_mpeg2dec_sink_convert), (gst_mpeg2dec_src_convert),
(gst_mpeg2dec_change_state):
* ext/mpeg2dec/gstmpeg2dec.h:
Simplify and don't leak our buffer pool.
Use _scale_int.
Remove unfixed bug number from previous ChangeLog entry.
Original commit message from CVS:
* ext/mpeg2dec/gstmpeg2dec.c: (gst_mpeg2dec_base_init),
(gst_mpeg2dec_class_init), (gst_mpeg2dec_init),
(gst_mpeg2dec_finalize), (gst_mpeg2dec_reset),
(gst_mpeg2dec_qos_reset), (gst_mpeg2dec_alloc_buffer),
(gst_mpeg2dec_negotiate_format), (init_dummybuf),
(handle_sequence), (handle_picture), (handle_slice),
(gst_mpeg2dec_chain), (gst_mpeg2dec_sink_event),
(gst_mpeg2dec_src_event), (gst_mpeg2dec_change_state):
* ext/mpeg2dec/gstmpeg2dec.h:
Fix padtemplate as we can now do fractional framerates.
Small cleanups.
Use GstSegment.
Add simple frame dropping QoS.
Precalc buffer output sizes and UV offsets.
Always give libmpeg2 a valid fbuf when it wants one.
don't trust libmpeg to discard our buffers but manage it
ourselves.
Fixes#343627, #327350, #335288
Original commit message from CVS:
* ext/amrnb/amrnbdec.c: (gst_amrnbdec_event), (gst_amrnbdec_chain):
* ext/amrnb/amrnbparse.c: (gst_amrnbparse_init),
(gst_amrnbparse_sink_event), (gst_amrnbparse_chain),
(gst_amrnbparse_sink_activate), (gst_amrnbparse_state_change):
* ext/amrnb/amrnbparse.h:
Sortof first quick cleanup of all this mess...
Don't crap out on empty and invalid FTs but treat them as
empty packets, the decoder handles them fine.
Fixes#342222.
Original commit message from CVS:
Patch by: James "Doc" Livingston <doclivingston gmail com>
* ext/mad/gstid3tag.c: (gst_id3_tag_get_tag_to_render):
Do tag merging correctly (#339918). Output taglists
properly in debug statements too while we're at it.
Original commit message from CVS:
* ext/a52dec/gsta52dec.c: (gst_a52dec_chain):
Add more debug
* gst/dvdlpcmdec/gstdvdlpcmdec.c: (gst_dvdlpcm_reset),
(gst_dvdlpcmdec_init), (update_timestamps),
(gst_dvdlpcmdec_chain_dvd), (gst_dvdlpcmdec_chain_raw),
(dvdlpcmdec_sink_event):
* gst/dvdlpcmdec/gstdvdlpcmdec.h:
If we have a first_access offset but no current timestamp (might
happen after a seek), then calculate a start time for the first
portion so that it will align with the timestamp given for the
first_access portion.
If a new-segment arrives with format time, store the start
time as a failsafe timestamp in case we never get any further
timestamp info (unlikely)
Mask out the 'frame number' section of the incoming header so
that we don't consider it to be changing on every buffer and
reset the caps constantly.
Use gst_util_uint64_scale for duration calculation
Original commit message from CVS:
* ext/lame/gstlame.c: (gst_lame_get_type),
(gst_lame_release_memory), (gst_lame_init), (gst_lame_sink_event),
(gst_lame_setup), (gst_lame_change_state):
* ext/lame/gstlame.h:
Remove tag writing from lame (which was completely broken
anyway, #329184). Leaving GstTagSetter interface around for
now, albeit non-functional. Should be removed completely
in 0.11. Use the 'id3v2mux' plugin from -good for writing
tags.
Original commit message from CVS:
Patch by: Andres Salomon <dilinger at debian org>
* ext/lame/gstlame.c: (gst_lame_sink_event):
Fix typo (comma vs. semicolon) (#340710).
Original commit message from CVS:
Patch by: Lutz Müller <lutz at topfrose de>
* ext/mad/gstmad.c: (gst_mad_init), (gst_mad_src_query):
Make mad the second element to support the highly useful
FORMATS query (#340594)
Original commit message from CVS:
* ext/mpeg2dec/gstmpeg2dec.c: (gst_mpeg2dec_sink_convert),
(gst_mpeg2dec_src_query):
Remember the query duration format before passing it upstream since it
could have been modified.
Add GST_WARNING_OBJECT in sink convert function to detail why the
conversion didn't work.
Original commit message from CVS:
* ext/mad/gstid3tag.c:
* ext/mad/gstmad.c:
* gst/ac3parse/gstac3parse.c:
* gst/dvdlpcmdec/gstdvdlpcmdec.c:
* gst/synaesthesia/gstsynaesthesia.c:
Define GstElementDetails as const and also static (when defined as
global)
Original commit message from CVS:
* ext/dvdnav/dvdnavsrc.c: (gst_dvd_nav_src_push_titlelang_event):
Fix name of custom event (use same as dvdreadsrc).
* gst/dvdsub/gstdvdsubdec.c: (gst_dvd_sub_dec_chain),
(gst_dvd_sub_dec_sink_event), (gst_dvd_sub_dec_handle_dvd_event):
Fix event parsing (the event name is in the structure, not the
name of the structure itself); also fix indentation after
boilerplate macro.
Original commit message from CVS:
Patch by: Alexander Lancaster <alexl at users sourceforge net>
* ext/mad/gstid3tag.c: (gst_id3_tag_get_caps), (gst_id3_tag_init),
(gst_id3_tag_sink_event), (gst_id3_tag_src_link),
(gst_id3_tag_chain):
When acting as a muxer, set caps on outgoing buffers and set caps
on source pad (fixes#323658). Remove unused application/x-gst-tags
cruft from the 0.6 days.
Original commit message from CVS:
* ext/mad/gstmad.c: (gst_mad_convert_sink), (gst_mad_convert_src),
(gst_mad_src_query), (gst_mad_chain):
.. and DEFAULT queries should work too. Use magic gst util scale
functions in some places.
Original commit message from CVS:
Patch by: Ed Catmur <ed at catmur dot co dot uk>
* ext/lame/gstlame.c: (gst_lame_sink_event):
Don't crash if we get an EOS event before the encoder
has been set up (#339287).
Original commit message from CVS:
* ext/dvdread/dvdreadsrc.c: (gst_dvd_read_src_start),
(gst_dvd_read_src_goto_title), (gst_dvd_read_src_read),
(gst_dvd_read_src_create), (gst_dvd_read_src_goto_sector):
Move errors out of the normal code flow.
Don't send eos, basesrc will do that for us when needed.
Original commit message from CVS:
* ext/mpeg2dec/gstmpeg2dec.c: (gst_mpeg2dec_alloc_sized_buf),
(gst_mpeg2dec_alloc_buffer):
Can't use gst_pad_alloc_buffer*() when we are going to crop
the image before sending it out. Downstream basetransform-based
elements will complain about the wrong unit size otherwise
(when not operating in passthrough-mode at least).
Const-ify some static variables and do some minor clean-ups.
Use I420 macros for size/offsets (not really necessary in this
particular context, but this kind of code gets copy'n'pasted).
