[x265] [Bug]reconstruction yuv picture diff with HM decoder out

[Deepthi Nandakumar]

Hi,


On Mon, Feb 17, 2014 at 12:38 PM, zxy at rock-chips.com <zxy at rock-chips.com>wrote:

>  hi, x265 members
> 1. [Bug report]
> We found that x265 (version 0.7+2-4b8901ae94ece1ac ) recon yuv data diff
> with HM decode out when config CQP mode with QP=34 and set rd=0 or 1 or 2,I think it's a serious bug!
>
> with command like this
>
>
> --input F:\yuv\Samsung_1080p_25.yuv --input-res 1920x1080 --fps 24 -q 34 -o E:\out1.bin -r E:\rec1.yuv --psnr --rd 0
>
>
> --input F:\yuv\Samsung_1080p_25.yuv --input-res 1920x1080 --fps 24 -q 34 -o E:\out1.bin -r E:\rec1.yuv --psnr --rd 1
>
> --input F:\yuv\Samsung_1080p_25.yuv --input-res 1920x1080 --fps 24 -q 34 -o E:\out1.bin -r E:\rec1.yuv --psnr --rd 2
>
>
rd 0 and 1 are not finalised yet. We are investigating the hash mismatch in
rd 2 clip with our clips; would be good if you can share the source.

>
> 2. [Proprose]
>  The x265 codec encodes video con-tent  using  a  fixed  quantization
> step,  thus  leading to  a  variable bitrate  stream  which  may  not  be
> suitable  for  the  many  multi-media  applications  where  a  constant
> bandwidth  is  required. Therefore,  maybe adaptive quantization  step may
> be better.
>
>
Adaptive Quantization is already implemented. You may use --aq-mode 1
(enabled by default in the medium preset).

>
> I'm looking forward to you
> thks
>  zxy at rock-chips.com
>
> _______________________________________________
> x265-devel mailing list
> x265-devel at videolan.org
> https://mailman.videolan.org/listinfo/x265-devel
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.videolan.org/pipermail/x265-devel/attachments/20140217/33c20dfa/attachment.html>