The XviD core package contains the libxvidcore library and the appendant header files. The library itself provides functions to encode and decode most MPEG-4 video data. Following features of MPEG-4 are supported: I, P Frame encoding I-Frames are image frames, while P-Frames are Pattern Frames. Image Frames contain full images. P-Frames contain patterns of what changes in the image, thusly saving a lot of space. PMVFast and EPZS with diamond a sqare pattern motion estimation Fast Motion detection that can refer a square or diamond shaped pattern to different directions. inter4v mode Fast interpolation ability. fullpixel and halfpixel precision Even halfpixel movements can be encoded. Better precision results in better compression at better quality. both MPEG-4 and H263 quantizations MPEG-4 is the standard DivX was derived from. H263 is a slightly changed subset of that standard made by Microsoft. Quantization means to map the quantity of color values to numbers. custom quant matrices The quantization can be done using custom matrices, possibly giving better results if the matrix is made to suit the video data. Every normal user should be fine with the default matrix that comes with XviD. Adaptive quantization / Luminance masking Qantization and Luminance masking are variant, depending on the input material, therefore providing better results on dark or very contrasting images.