New transform for video coding I designed a new transform for video coding. These are its properties: In one pass, it gets a PSNR of 0.58 dB below H.265 KTA 2.2r1 with default settings and multiple passes. This was measured on the sample images on this page. Lossless compression as a special case of lossy compression when the quantizer step size is 1. Compatible with blocks and motion compensation. Integer transform. Multiresolution. A description of the new transform is available in my patent application. The new transform is based on S+P. It can be adapted for video by dividing an inter frame into regions. For example, a region could be the area of a motion compensation block. And performing the P (prediction) step using data from the same region. Here are sample images. (Original of Lena image). Lena compressed to 12955 bytes using H.265 KTA 2.2r1 with default settings and multiple passes: Lena compressed to 12955 bytes using the new transform and only one pass: The new transform can be used for applications that are not targeted by the H.264 model of long encoding time and lossy compression. Applications such as live recording and teleconferencing, lossless compression, and high dynamic range. As an example use, a cineast could record lossless high dynamic range video. Upon returning to the studio, the cineast could enable motion compensation to archive the footage resulting in a lower file size.