v0.2.2 - Temporal video workflows and proof comparisons

ComfyUI-DLSS5 v0.2.2 packages the current Windows image and video workflows for DLSS Super Resolution and the experimental Neural Rendering runtime.

Changes since the last GitHub prerelease:

• Added Video Depth Anything Small and FlashDepth workflow variants
• Added RAFT motion guides, temporal depth stabilization, and persistent video processing
• Added 2x, 3x, and 4x Super Resolution paths
• Added separate Neural Rendering and combined SR plus NR nodes
• Added tested Alyx, Gordon Freeman, rainy-city, and industrial-corridor comparisons
• Added detailed ComfyUI workflow and node screenshots
• Published the package through the Comfy Registry

Install from ComfyUI Manager or the Comfy Registry:
https://registry.comfy.org/nodes/experimental-dlss-neural-rendering

The NVIDIA Neural Rendering DLL is not distributed with this project. Supply a legally obtained compatible nvngx_dlssnr.dll. This remains an unofficial experimental integration and is not an official NVIDIA DLSS 5 SDK release.

v0.1.0-alpha.4 — ComfyUI workflow screenshots

Documentation refresh for the public alpha. Adds full-size ComfyUI screenshots for the recommended Easy workflow and the Advanced guided workflow, plus readable close-ups of the Easy and full SR/neural-rendering nodes. The Alyx comparison remains the primary output example. No proprietary NVIDIA runtime files are included.

v0.1.0-alpha.3 — Alyx documentation refresh

Documentation presentation update.\n\n- Uses the Alyx neural-rendering comparison as the README hero\n- Removes the less representative previous example asset completely\n- Preserves unstretched aspect-ratio comparisons\n- Retains the one-click ComfyUI runtime setup introduced in alpha.2\n\nAll 7 tests and Comfy Registry validation pass.

v0.1.0-alpha.2 — One-click ComfyUI runtime setup

Installation UX update.\n\n- Adds a DLSS Runtime Setup node usable entirely inside ComfyUI\n- User only supplies nvngx_dlssnr.dll\n- Downloads the pinned VapourKit Windows nightly automatically\n- Verifies the published SHA-256 before extraction\n- Locates runtime components and writes local configuration automatically\n- Keeps PowerShell/Python installers as optional headless fallbacks\n- Passes all Comfy Registry validation and security checks\n\nNo proprietary NVIDIA runtime is downloaded or distributed.

v0.1.0-alpha.1 — Setup source documentation

Documentation update for the first public alpha.\n\n- Adds a component-by-component table with direct sources\n- Links the exact VapourKit nightly used during development\n- Adds the official VapourKit community Discord\n- Links Depth Anything V2, TorchVision RAFT, Video Helper Suite, NVIDIA DLSS, ComfyUI, and NVIDIA drivers\n- Clearly distinguishes automatic downloads, optional dependencies, and the user-supplied proprietary neural-rendering runtime\n\nNo proprietary NVIDIA or VapourKit binaries are distributed.

v0.1.0-alpha — Experimental DLSS Neural Rendering for ComfyUI

First public alpha release.\n\nHighlights:\n- Easy one-node workflow with automatic still/short-video/long-video presets\n- Separate Upscale only, Neural rendering only, and combined operations\n- Advanced SR, neural rendering, Depth Anything V2, RAFT, optical-flow, and temporal depth nodes\n- Persistent and bounded overlap-add processing modes\n- One-time Windows setup and four documented workflows\n\nImportant: this is an unofficial experimental bridge. No NVIDIA or VapourKit binaries are distributed. Users must provide legally obtained compatible runtime files and review the README before use.