mnema at yahoo.com
Thu Feb 2 11:20:40 CET 2012
Thanks for your mail.
'-vcodec' is an option at the command line level. I am using the libraries to manipulate the videos in the memory and then want to save in the form of video. After manipulation I do not have any Motion Vector knowledge. It is as good as motion JPEG.
Here is a step by step operation I would like to do
1. Read a video (not raw, only till DCT coefficients of Y Cb and Cr bands) in to memory.
2. Manipulate the DCTs of Y Cb and Cr bands. At this point, we have modified DCTs of all the frames without motion vector.
3. Feed the DCTs of the Y Cb and Cr into an api to get final video output. Here one can specify the bit-rate and quality of the resultant video.
I am looking forward for api(s) to achieve step 3 above.
From: Ronald S. Bultje <rsbultje at gmail.com>
To: Malay Nema <mnema at yahoo.com>; Development with the libav* APIs <libav-api at libav.org>
Sent: Thursday, February 2, 2012 11:15 AM
Subject: Re: [libav-api] Transcoding
On Wed, Feb 1, 2012 at 9:42 PM, Malay Nema <mnema at yahoo.com> wrote:
> Dear Libav-api,
> We are trying to handle videos in YCbCr DCT coefficients. After
> manipulation we want to encode the video directly from the DCT coefficients.
> This amounts to transcoding the video streams.
> Can any of you tell me the relevant subroutine(s)/api(s) and their sequence
> which encodes videos given the DCT coefficients of YCbCr of the frames
You're looking for -vcodec copy, which just copies the exact encoded
frame into the output file? Otherwise you'll have to be a little bit
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the libav-api