When rendering Hair with MRQ, when the dynamics are high, the noise is unacceptable.
It is known that this is caused by insufficient temporal accumulation. Theoretically, increasing the temporal sample count for offline rendering, such as rendering 32 or 64 times per frame, would eliminate the noise when enough samples are blended together. However, the real-time dynamically computed Groom is not suitable for this kind of rendering, as it causes jitter (rendering once per frame is smooth). Perhaps caching could solve the jitter problem, but the cost is too high. Is there a way to render only one frame but sacrifice more time to achieve smoother quality? (Similar to the result when pausing in the viewport, where TSR quality accumulates infinitely until there is no noise at all.)