Flowframes
A local AI video interpolation app for Windows that supports RIFE, DAIN and FLAVR, with GPU processing on NVIDIA and AMD hardware. Open source under GPL-3.0.
Flowframes is a Windows desktop app that uses local AI to generate intermediate frames for smoother video and animation. It's for people working with camera footage, 2D animation or rendered video who want a graphical interface for interpolation. The app is open source under GPL-3.0 and available as donationware, with free builds.
It supports RIFE through PyTorch and NCNN, DAIN through NCNN, and FLAVR through PyTorch. The included RIFE-NCNN backend works with modern Vulkan-capable GPUs, including NVIDIA and AMD cards. CUDA processing requires NVIDIA hardware, and RIFE runs faster through CUDA than NCNN on NVIDIA cards. Recommended hardware includes at least 6 GB of VRAM and 16 GB of RAM.
Its animation tools address problems that basic frame-rate conversion can leave behind. Duplicate-frame removal helps smooth 2D animation with repeated drawings. Scene-cut handling avoids generating morphs between unrelated shots, while loop interpolation connects the end of an animation back to its first frame. Flowframes also supports transparency when both the input and output formats allow it, including PNG and GIF.
Exports can retain audio, subtitles and MKV metadata. You can limit the output frame rate, encode video during interpolation to reduce temporary disk use, or process stages separately when you need to edit individual frames. Video and GIF export use FFmpeg encoding options.
