Sunday, 10 February 2013

x264 Settings Comparison Part : 1 - Introduction

x264 Settings Comparison Part : 1 - Introduction


x264 is one implementation of the H264 standard, it is also one of the best implementation of the standard, coming with a wide variety settings. Unfortunately not all settings are equal when it comes to compression, in the next series of post I want to determine how big of an influence most of the settings have.


To do this I'll be using Handbrake video encoder v0.9.8 to conduct as series of video encodings each encoding will have one setting that differ from the default settings (The default Handbrake settings are also the Medium x264 preset settings), then see how that particular setting influences the compression and speed.

The settings that will be covered are:
  1. x264 Presets
  2. Subpixel ME & Mode Decision
  3. Motion Estimation Method
  4. Adaptive Direct Mode
  5. Adapative B-Frames
  6. Reference Frames
  7. Maximum B-Frames
    1. Fast
    2. Optimal
  8. Cabac
  9. 8x8 Transform
  10. Weighted P-Frames
  11. DCT-Decimate
  12. Partitions
  13. Trellis
In other words all the settings available in Handbrake, x264 as a lot more settings that can be accessible via the console in the "Advanced Tab", this will only be utilised when testing the presets of x264, as that is the only to apply them in Handbrake.

Further more all encoding will use a Constant Rate Factor(CRF) of 18. This ensure that all videos have a consistent quality, thus if a encoded file differ in size it would be as a result of the isolated setting. It is however important to note that CRF is not a perfect quality assessment tool and we will further be looking at 3 types of screen shots of the encoded file, to find out if any visual artefacts were introduced.
  1. Still images
  2. High motion
  3. High texture complexity
This process will be repeated on Standard definition(SD) Cartoons, SD life action motion and High definition(HD) live action motion.

  1. Introduction
  2. Presets 
  3. Subme


No comments:

Post a Comment