Re: [FFmpeg-devel] [PATCH v2] lavc/libx264: support AV_CODEC_CAP_ENCODER_RECON_FRAME

From: James Almer <jamrial@gmail.com>
To: ffmpeg-devel@ffmpeg.org
Subject: Re: [FFmpeg-devel] [PATCH v2] lavc/libx264: support AV_CODEC_CAP_ENCODER_RECON_FRAME
Date: Tue, 19 Jul 2022 15:13:27 -0300
Message-ID: <fc49130d-466f-e611-2490-b277e900fa28@gmail.com> (raw)
In-Reply-To: <165823546515.15564.594189075971384302@lain.khirnov.net>

On 7/19/2022 9:57 AM, Anton Khirnov wrote:
> Quoting James Almer (2022-07-19 14:51:13)
>> If this is only >= 122, what will pic_out above contain in older versions?
> 
> IIUC it won't do deblocking, so the reconstructed frame won't be exactly
> as decoded.

That kinda defeats the purpose, and is also inconsistent. Maybe we can 
signal somehow at the API level that the reconstructed frame is not 
fully done? Or making the kind of reconstruction the user requests 
explicit, like having the recon_frame AVOption be an enum in 
AVCodecContext with defined values like full_recon (hard requirement), 
partial_recon (soft requirement, may still return full recon if that's 
all the encoder can generate), etc, instead of simply a flag for 
avctx->flags?

Or just disabling it for libx264 this old so output/behavior is 
consistent across all encoders regardless of version and the API remains 
as simple as possible.
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".