x265 has a number of predefined --preset options that make trade-offs between encode speed (encoded frames per second) and compression efficiency (quality per bit in the bitstream). The default preset is medium, it does a reasonably good job of finding the best possible quality without spending enormous CPU cycles looking for the absolute most efficient way to achieve that quality. As you go higher than medium, the encoder takes shortcuts to improve performance at the expense of quality and compression efficiency. As you go lower than medium, the encoder tries harder and harder to achieve the best quailty per bit compression ratio.
The presets adjust encoder parameters to affect these trade-offs.
Placebo mode enables transform-skip prediction evaluation.
There are a few --tune options available, which are applied after the preset.
The psnr and ssim tune options disable all optimizations that sacrafice metric scores for perceived visual quality (also known as psycho-visual optimizations). By default x265 always tunes for highest perceived visual quality but if one intends to measure an encode using PSNR or SSIM for the purpose of benchmarking, we highly recommend you configure x265 to tune for that particular metric.
|psnr||disables adaptive quant, psy-rd, and cutree|
|ssim||enables adaptive quant auto-mode, disables psy-rd|
|grain||improves retention of film grain. more below|
|fastdecode||no loop filters, no weighted pred, no intra in B|
|zerolatency||no lookahead, no B frames, no cutree|
Film Grain Retention¶
--tune grain tries to improve the retention of film grain in the reconstructed output. It helps rate distortion optimizations select modes which preserve high frequency noise:
It lowers the strength of adaptive quantization, so residual energy can be more evenly distributed across the (noisy) picture:
- --aq-strength 0.3
And it similarly tunes rate control to prevent the slice QP from swinging too wildly from frame to frame:
And lastly it reduces the strength of deblocking to prevent grain being blurred on block boundaries:
- --deblock -2
--tune fastdecode disables encoder features which tend to be bottlenecks for the decoder. It is intended for use with 4K content at high bitrates which can cause decoders to struggle. It disables both HEVC loop filters, which tend to be process bottlenecks:
It disables weighted prediction, which tend to be bandwidth bottlenecks:
And it disables intra blocks in B frames with --no-b-intra since intra predicted blocks cause serial dependencies in the decoder.
There are two halves to the latency problem. There is latency at the decoder and latency at the encoder. --tune zerolatency removes latency from both sides. The decoder latency is removed by:
Encoder latency is removed by:
With all of these settings x265_encoder_encode() will run synchronously, the picture passed as pic_in will be encoded and returned as NALs. These settings disable frame parallelism, which is an important component for x265 performance. If you can tolerate any latency on the encoder, you can increase performance by increasing the number of frame threads. Each additional frame thread adds one frame of latency.