Об этом ПО

FrameFusion Motion Interpolation improve the framerate of any video, gif or group of images by 2X/4X/8X/16X

• Choose from four different models for interpolation: options for live-action or animation, with the ability to render in high quality or with faster speeds.
  
• Optimized for Nvidia GPUs with CUDA support but also compatible with AMD GPUs and CPUs.
  
• Manage a queue of files for batch processing, with a progress bar to track how long each file will take.
  
• For static images like PNG, you can select a folder of files and manually choose an FPS to use.

• Export Frames: Generates a .PNG file for each processed frame, including the newly interpolated ones.
  
• Export Motion Flow: Exports a PNG visualization of the optical flow.
  
• Export Video Side-By-Side: Exports the original video alongside the interpolated video for comparison.
  
• Export Video Cut in Half: Exports the left side of the video as the original and the right side as the interpolation, split down the middle for comparison.

• Perfect Loop Video: Creates new frames between the last frame and the first frame of a video to create a perfect loop.
  
• Remove Static Frames: Especially useful for animations, this option removes static frames to enhance the fluidity of the final video.
  
• Skip Scene Changes: Detects scene changes and avoids interpolation to reduce artifacts in these transitions.

• 2X: 48fps
  
• 4X: 96fps
  
• 8X: 192fps
  
• 16X: 384fps

Информация о контенте, созданном с помощью ИИ

Вот как разработчики описывают контент в игре, сгенерированный ИИ:

The app processes the frames from the video and uses AI to predict how much movement each pixel has along the X and Y axes. The AI model then generates optical flow data for both axes, which are just floating-point numbers for X and Y.

With this information, traditional methods are used to create a new frame by combining parts of the two original frames. The AI doesn't generate new RGB data; it only uses the RGB data from the original frames.