Psychovisual Enhancements

Posted on November 08, 2011 by Admin
Psychovisual Enhancements

Psychovisual Enhancements

Moscow State University Graphics & Multimedia Laboratory has finished 6-th H.264 codecs comparison.
It is intended for practical researchers and developers in the field of high-end video compression.

We have tested newest implementations of MPEG-4 AVC/H.264 video codecs and compare with XviD (MPEG-4 ASP) and Theora encoders.
One of the main targets for this comparison was to test H.264 encoders for transcoding tasks for Movies and HDTV video content.

Codec that were tested:

  • DivX H.264
  • Elecard H.264
  • Intel® MediaSDK AVC/H.264
  • MainConcept H.264
  • Microsoft Expression Encoder
  • Theora
  • x264
  • XviD (MPEG-4 ASP codec)
  • VP8

Summary report topics:

  • Objective measurements (SSIM, PSNR, Average Advantage and etc.)
  • Encoding speed
  • Analysis of averaged objective results
  • Leaders in different areas (Movies, HDTV)
  • Options analysis for codecs
  • Additional subjective analysis for video codes
    • For psycho-visual enhancement in codecs analysis
    • For fade processing analysis
    • For animation movie compression analysis
  • Codecs encoding quality progress over years

Enhancements in comparison to Previous H.264/AVC Comparison:

  • Subjective comparison
  • New codecs
  • New sequences
  • New type of special analysis for codecs
  • Using natural sequences’ special modifications
  • Using synthetic sequences
  • Not only H.264 Codecs (but also XviD, Theora) were tested

Random Posts

  • Are Blu-ray and HD-DVD Discs Region Coded, Like DVDs?
  • Why I need purchase the Extended Download when I purchase software?
  • Twitter will alert users before handing over information to authorities
  • CloneDVD for Mobile Officially Released Today
  • Can I rip a DVD which zone does not match the zone of my DVD-ROM drive?

So you've tried DivX Converter but you've found you want more control over the encoding process. I'd suggest using software that gives you direct access to the DivX encoder options. I use Dr. DivX but I also recommend StreamClip (OSX) and VirtualDub (XP/Vista). Beware, there is a lot of software out there that claims to make "DivX" files but does not use the authentic, fast, high-quality DivX encoder. We think our encoder is hands-down, the best. Accept no substitute.

For Mobile, I always crop to 4:3 aspect ratio then scale to 320x240 (qVGA) and keep the video bitrate around 300 kbps. Lanczos is the best resize filter. Always crop off any black borders that may appear around the original video. You waste lots of bitrate trying to encode a sharp black frame around your video. Get rid of them.

For all encodings, I take advantage of the optional enhancements offered by DivX Pro. That means, Multipass (fast first pass), Insane Quality, Experimental SSE4 Search, Enhanced Multithreading, Bidirectional coding, Auto Noise Reduction, Optimized Quantization, Shaping Psychovisual Enhancements and Enhanced I-Blocks. For grainy video, I use Masking Psychovisual Enhancements.

Multi-pass can improve the picture quality and provide better control of the average bitrate. For example, if you try 300 kbps with 1 pass, you might get 275 kbps or 325 kbps. With fast first pass, you will usually get within 2 kbps of your target rate. If you really want exactly the target rate, make a slow first pass.

If you don't care about rate, use the "1-pass quality based" setting. It's really the best and fastest method. You let the encoder choose the optimal bit-rate based on the content. I typically use a Quality setting between 3.5 and 5.0 depending on the content. Unfortunately, with this setting, you don't know what average bitrate or peak bitrate you'll end up with and you could easily get something your phone can't handle, so I don't use this setting for Mobile sized videos.

I never mess with the max key frames and keyframe threshold. Theoretically you could optimize those settings based on your content. For example, if you have lots of video of a talking head, use a very high max key frame. If you have lots of panning or short, similar scenes, lower the max or the keyframe threshold. By default, you'll get a key frame at least every 3 seconds and I rarely try to tweak it.

For more detail about what all these settings mean, see the Dr. DivX user guide.

Don't forget about the audio. Many DivX videos have 128 kbps MP3 stereo audio. That's usually what you want for Mobile clips and a converter like VirtualDub will copy the audio stream as-is without re-encoding it. Unfortunately Dr. DivX will always re-encode the audio stream whether you need to or not. Sometimes I will use DivXMux to use the original audio stream and the converted video stream.

If you never use stereo headphones, you can save a few bytes by converting the audio to mono and drop the bitrate to 96 kbps.