ndufresne: that code actually has more history than just rockchip mpp ;) but that's correct, we didn't have much ability to come up with anything better and that's also why it stayed as a plugin and not a generic component in Chromium ndufresne: anyway, looking forward to discussing this with you on ELCE. thanks :) hverkuil: hello ribalda: hi thinking about the unit_size maybe it make sense to add a second control, unit_size_processed or something like that, with the size after binning because the user would probably have good use for both I will not add it until we have a use for it, but would that be something you are completely against? ribalda: I'm not enthusiastic :-) I think this should not be needed, userspace should be able to get e.g. binning information so it can calculate the processed size based on that and the UNIT_CELL_SIZE. ok. will send the fixed doc. Thanks! hverkuil, regarding your comment about using v4l2_pix_format_mplane instead of v4l2_format in patch 16 this also means i"ll have to chabge the signature of csc_set_coeff(). This is somewhat of an issue because the csc.c code is used by VPE and VIP (which is not upstreamed) but it is however not using multi-plane format but just single plane. Wasn't there a push at some point to condense/eliminate the single to multi-plane struct format? bparrot: I hadn't noticed that it was used in csc_set_coeff as well. Just keep v4l2_format. well i guess i can change the VIP side to use v4l2_pix_format_mplane also.... I do want to get rid of the single/multiplanar complexity, but that'll take quite some time, I suspect. well i did nopt want to have to pass all individual parameters for both source and destination formats but i am also using the fmt.pix_mp construct insode csc_set_coeff so i guess I would have to change it also... sigh s/insode/inside/ well, if you pass v4l2_format, then you can use the type field to determine whether to set pix_mp or pix in csc_set_coeff. But it's up to you to see what's easiest. Oh but i can use the format type with a switch there yeas just realized that alright, i should send a v3 later today sorry everyone for the mail mess :S. If you are using patchwork it should be fixed. ribalda: so close... : /home/hans/work/src/v4l/media-git/include/media/v4l2-ctrls.h:1133: warning: Function parameter or member 'area' not described in 'v4l2_ctrl_s_ctrl_area' /home/hans/work/src/v4l/media-git/include/media/v4l2-ctrls.h:1133: warning: Excess function parameter 's' descr iption in 'v4l2_ctrl_s_ctrl_area' /home/hans/work/src/v4l/media-git/Documentation/output/videodev2.h.rst:6: WARNING: undefined label: v4l2-ctrl-t ype-area (if the link has no caption the label must precede a section header)                                  hverkuil: want to cry :). What do you run to get those? srctree=`pwd` BUILDDIR=`pwd`/Documentation/output  make -C Documentation/media/ all                             BUILDDIR=Documentation/output sphinx-build -b html -D version=latest -D release= -d Documentation/output/.doctrees -D kerneldoc_srctree=`pwd` -D kerneldoc_bin=`pwd`/scripts/kernel-doc -c `pwd`/Documentation  `pwd`/Documentation/media/ Documentation/output/html/                                                                        ^^^ I have that in a little shell script gitdocs.sh It's overly complex though, these days this works as well: make SPHINXDIRS=media htmldocs (I've just changed my gitdocs.sh script :-) ) hverkuil: thanks! hverkuil, tfiga: I'm trying to see if thaythan (an OpenH264 contributor) could join us for the encoder discussion, I'll let you know if he's available ndufresne: great, thanks a lot! bbrezillon: just a reminder: can you rebase and repost your series for the extended streaming/format ioctls this week? This will be very useful to discuss during the ELCE. hverkuil: any idea how to get rid of the last one? /home/hans/work/src/v4l/media-git/Documentation/output/videodev2.h.rst:6: WARNING: undefined label: v4l2-ctrl-type-area (if the link has no caption the label must precede a section header) Add it to Documentation/media/videodev2.h.rst.exceptions There are other V4L2_CTRL_TYPE_'s there. thanks! hverkuil: yep, it's on my TODO list didn't have time last week (XDC) sailus, is my last reply to your comment on ov5640 series, ok? if so i'll send a v4