用 RIFE/FILM 模型生成中间帧，实现流畅慢动作或帧率转换。
适用场景
Video creators needing smooth slow-motion from standard footage, archivists restoring old video at higher frame rates.
前提条件
ComfyUI, RIFE or FILM interpolation model, ffmpeg, GPU with 8GB+ VRAM
工作流图
用 ComfyUI 做 AI 视频帧插值
分步指南
-
Load Video Frames
Use the Video Frame Extract node to load your video and extract frames as individual images. Set the frame range and resolution. For long videos, process in 10-second segments to avoid memory issues.
-
Frame Interpolation Model
Add a RIFE or FILM interpolation node. Set the interpolation factor (2x for 24->48fps, 4x for 24->96fps slow motion). The model generates intermediate frames between each pair of original frames.
-
Quality Enhancement
Optionally add a Real-ESRGAN node to upscale each frame to 2x or 4x resolution. This is useful when the source video is low quality. Set denoising strength to 0.3-0.5 to avoid artifacts.
-
Frame Reassembly
Use the Video Frame Combine node to reassemble frames into a video. Set the output frame rate to match your target (48fps for 2x interpolation of 24fps). Choose MP4 with H.264 encoding for compatibility.
-
Batch Processing Setup
For clips longer than 10 seconds, create a loop workflow using a Code node that processes segments sequentially. Save each segment and concatenate with ffmpeg in a final post-processing step.