[x264-devel] x264 analysis and updated metric tool

[Dmitriy Vatolin]

Dear experts!

Several news from us:

1. We have analyzed x264 - following options:
   * Macroblocks partitions selection (--partitions)
   * Number of B-franes (--bframes)
   * Number of reference franes (--ref)
   * Motion Estimation method (--me)
   * Algorithm of subpixel compensation usage (--subme)
   * Mixing of predictions from differnt frames (--mixed-refs)
   * Weighted predictions in B-frames (--weightb)

   Document present novel type of options efficiency analysis, that can
   answer on following questions:
   * What set of options is optimal (in terms of optimal speed/quality)
     for exact class of video?
   * What option implementation is more efficient?
   * Is results options preset stable (in terms of optimal speed/quality
     balance) from sequence to sequence?

   Results can be found here:
   http://compression.ru/video/codec_comparison/x264_options_analysis_08_en.html

2. Also we published MSU Video Quality Measurement Tool 2.0 beta!

   New metrics release has following new features:
   * Simpler *.MOV, *.VOB, *.WMV, *.MP4, *.MPG, *.MKV, *.FLV and other
     media formats support
   * Bigger number of metrics (all types of PSNR) and etc.
   * 10, 14 and 16 bits color depth raw files support (in PRO).

   This is good tool for modest research in area of video compression
   or video filtering (for professional work we recommend our PRO
   tool).

   Program can be found here:
   http://www.compression.ru/video/quality_measure/video_measurement_tool_en.html

3. For next regular H.264 codecs comparison we need some help.

   It's reasonable to include as more codecs as possible to make
   comparison appropriate, but many companies stop H.264 codecs R&D
   now, so they are not interested in next comparisons participation.

   So it's necessary to start DirectShow or GUI codecs (like Apple or
   LEAD) for batch running with different parameters.

   Any help to solve this problem is welcome.

   Thank you!

-- 
Best regards,
 Dmitriy                          mailto:dmitriy at graphics.cs.msu.ru