Every Processing Option changes the output in a different way. Compare them side-by-side before you spend render time on your own video.
GFPGAN face restoration runs on every frame for a sharper result, at the cost of extra render time.
Faster rendering, slightly softer facial detail.
Each frame is passed through face restoration. Noticeably crisper.
Another avatar showing baseline output without face restoration.
The same avatar with GFPGAN — sharper skin texture and eye detail.
Upscaling adds clarity after rendering without re-running the model. 512 px is the recommended sweet spot for most uses.
Good balance of quality and file size. Recommended for most uses.
Maximum clarity, larger file. Best for presentations and high-res exports.
Still mode keeps the head stable with minimal drift — good for presentation-style videos.
Head stays mostly fixed. Ideal for explainers and slides.
More natural head motion. Better for conversational avatars.
Controls how animated the face is. The default of 1.0 is the most natural — lower values are subtler, higher values are more dramatic.
More restrained movement. Good for calm, measured delivery. The default of 1.0 sits between these two and is the most natural-looking.
More pronounced mouth movement. Works well for energetic or enthusiastic speech.