NVIDIA announced new artificial intelligence video generation tools and performance improvements for game developers and creators at the Game Developers Conference in San Francisco.

The company introduced ComfyUI's App View, which provides a simplified interface for the generative AI tool. The new interface allows users to enter prompts and adjust parameters without navigating complex node graphs, while maintaining access to the full node-based experience through Node View.

NVIDIA released RTX Video Super Resolution for ComfyUI, enabling real-time 4K video upscaling. The technology is available as a standalone node for ComfyUI workflows and as a Python package through the PyPI repository for developers. According to NVIDIA, the upscaling runs 30x faster than alternative local upscalers while using less video memory.

The company introduced NVFP4 and FP8 model variants for FLUX.2 Klein, with NVFP4 support for LTX-2.3 scheduled for release. NVIDIA states these formats deliver up to 2.5x performance gains and 60% lower memory usage on GeForce RTX 50 Series GPUs using NVFP4, and 1.7x faster performance with 40% reduced memory usage with FP8.

Performance improvements for RTX GPUs in ComfyUI have increased 40% since September, according to the company. The new models are available through Hugging Face, with default workflows accessible via ComfyUI's Template Browser.

NVIDIA also announced updates to RTX Remix, LTX Desktop video editor, and partnerships with Topaz Labs and Microsoft for GPU-optimized applications. The RTX Video Super Resolution technology is powered by the NVIDIA Video Effects software development kit and runs on RTX GPU Tensor Cores.