Noise reduction effect with cinematic level details

Video AI 4, trained on millions of video frames, can intelligently understand the difference between noise and details

Identifying characters and faces, knowing exactly how to correct them

The vast desert and beautiful portraits are completely different, and the way you correct them should also be the same. Video AI4 uses time aware facial recognition to automatically and instantly complete.

Create a new frame using slim AI-r

Match cameras with different frame rates or achieve slow motion without changing camera settings. The Apollo and Chronos models generate new frames, making them super smooth.

Stable in the later stage

Focus on the subject being photographed, rather than camera shake. Stabilize the image and restore details lost due to unnecessary camera movement, while reducing blurring caused by motion and slow shutter speed. All in the later stage.

function

• Directly compare UI, used to preview images and directly compare any two of the 24 AI models. Fine tune settings through side by side, sliding, and stacking views to achieve perfect video enhancement effects.

Video enhancement

• Six enhancement models trained for different types of quality improvement
• Automatic mode for quick and easy processing, or manual control for fine-tuning output
• Intelligent detection and repair of visual artifacts generated by video compression
• Stack enhanced filter, capable of handling multi-step AI tasks with fewer clicks
• Deinterlacing options for single frame rate or double frame rate
• Specific processing channels for interlaced scanning and television content
• The 2x scaling mode enhances clarity and fidelity during the denoising process
• Iris MQ for enhancing facial recovery
• Custom mode for processing black and white videos
• 3: 2. Custom Mode for Dropdown/Reverse TV and Movie Materials

Steady Video

• Including automatic cropping and full frame mode, with video overlay
• Additional options for roller shutter correction and jitter compensation

Frame interpolation

• Using the Apollo model for up to 16 times slow motion conversion
• Interpolation up to 120FPS using Chronos

Exercise deblurring

• Reduce the impact of low shutter speed in fast-moving scenes

Wide range of export options

• Create production level output using ProRes and image sequence export options, integrating video AI into your existing workflow
• Keep the original audio tracks in the material or choose between multiple conversion options to create the best output for your work
• Improved dynamic bit rate control: The "high" setting for H264 and H265 exports has been adjusted to a target VMAF score of approximately 95, which is now the default setting for newly installed video AI.
• Optimizing visual quality and file size using dynamic bit rate control on supported codecs
• The lossless FFV1 codec is now available by default in the export options panel and can be used for 4:2:0, 4:2:2, and 4:4:4 chromaticity subsampling levels.

additional

• 50% faster GPU compatible with TensorRT
• Supports rotating videos by 90, 180, and 270 degrees - correcting videos with missing rotation metadata
• AV1 encoder suitable for AMD 7000 series GPUs and 700 series integrated graphics cards