Pipeline Stages
- Load Video: import the source clip and extract frames
- Downscale Frames: resize to ~1280px on the longest side for fast, stable interpolation
- Choose Model: RIFE 4.x for speed, FILM for quality on fast motion
- Set Factor: 2x for 24→48fps, 4x for 24→96fps (slow motion)
- Interpolate: run the RIFE/FILM node in batches, using tiling if VRAM is tight
- Reassemble: combine frames back into a video
- Encode: H.264/H.265 with a high bitrate, then export
Recommended Parameters
- Interpolation model: RIFE 4.x (default) or FILM (fast motion)
- Factor: 2x standard, 4x for slow motion
- Working resolution: 1280px longest side
- Batch size: 8-16 frames per pass, adjust to VRAM
- Output: H.264 or H.265, CRF 18-20
Use It With Other Workflows
Frame interpolation is the final polish stage in a video pipeline. Run it after generation - pair it with an AI video or image model to smooth AI-generated clips, or interpolate footage before editing in tools like Runway or Genmo Video. For image-side control, see the ComfyUI Style Transfer Pipeline.
FAQ
What is ComfyUI video frame interpolation?
A ComfyUI workflow that generates intermediate frames between existing video frames using RIFE or FILM models. It turns a choppy 24fps clip into smooth 60fps, and doubles or quadruples frame count for slow-motion shots without re-recording.
RIFE or FILM: which interpolation model should I use?
RIFE is faster and better for real-time and high-volume work; FILM produces higher quality for challenging motion. Start with RIFE 4.x for 2x-4x upscaling, and switch to FILM for footage with fast or complex motion where artifacts matter more than speed.
Can I use frame interpolation to create slow motion?
Yes. Interpolate the clip 4x (24fps to 96fps), then reinterpret it at the original timeline speed - the extra frames let you slow the clip 4x while keeping motion smooth and artifact-free.