NNSuperResolution v4.0.0
NNSuperResolution is a plugin to increase the resolution by 2x/4x Nuke by Foundry. Enjoy the ability to upscale footage and sequences to much higher resolution directly in the composite.
Features of NNSuperResolution:
• High-end with 2x or 4x odds.
• Sequence mode for upscaling image sequences, e.g. B. captured source plates, final compositions, computer renderings, etc.
• Still image mode for scaling photos and other still materials such as DMP patches, textures, etc.
• Processes all RGB channels of the input image (multi-layer support, meaning you can pass multiple layers through the plugin at once using Nuke’s native multi-channel/layer system).
• Sequence mode CG upscaling solution that can upscale RGBA images (i.e. 3D rendering).
• High dynamic range scaling.
• Built-in overscan processing (ie, processing larger bounding boxes than the image frame/format). Please note that bounding boxes cannot be animated in sequence mode, meaning they cannot vary from frame to frame.
• GPU acceleration with NVIDIA’s CUDA (requires NVIDIA graphics card).
B• Stitching together multiple inference/image patches internally (to be able to scale high-resolution images without having as much video memory on the GPU).
• Supports Nuke 11.3 and above on Linux.
• Supports Nuke 12.0 and above on Windows.
NNSuperResolution is a powerful and versatile tool for upscaling digital images and sequences in Nuke with a range of advanced features that enable high-quality results and efficient processing times.
The scaling algorithm is based on modern neural network technology (often called deep learning, machine learning or artificial intelligence). There are two modes available that use different network solutions to achieve the best results when capturing photos or sequences. While still mode produces clearer results, some moving materials may experience stuttering/flickering. In sequence mode, the focus is on making the result as clear and detailed as possible, with the main focus on ensuring that the result remains stable over time.
