Merserk/dlss5-visual-enhancer v12.0

Visual Enhancer v12.0

on GitHub

Changelog

• Completely reworked Neural Rendering into a customizable, multi-stage processing pipeline, allowing you to enable, disable, expand, collapse, and reorder individual processing stages using drag-and-drop controls. Image and video workflows now have independent stage configurations, with previews and final exports following the selected processing order.
• Added DLSS Super Resolution as an independent processing stage for images and videos, with DLAA, Quality, Balanced, Performance, and Ultra Performance modes, alongside selectable DLSS presets for greater control over upscaling.
• Added a new Denoising stage for images and videos, featuring adjustable noise reduction, compression artifact cleanup, and temporal noise reduction for video.
• Added a comprehensive Coloring workflow, including automatic image color matching, reference-image matching, and support for importing and applying 3D .cube LUTs to images and videos.
• Added advanced color grading controls for exposure, contrast, highlights, shadows, whites, blacks, midtones, temperature, tint, hue, vibrance, saturation, and LUT strength. Includes automatic color adjustment, customizable LUT resolutions, and the ability to save your adjustments as reusable LUT files.
• Added NVIDIA NIS and AMD FidelityFX CAS sharpening, with adjustable sharpening strength and a dedicated processing stage for images and videos.
• Added customizable image and video scaling filters, including Lanczos4, Bicubic, Bilinear, Area, Nearest, Spline36, and EWA Lanczos, depending on the selected media type.
• Added 16-bit PNG and TIFF image export, alongside improved high-bit-depth image decoding and processing to preserve more source precision throughout supported workflows.
• Improved 10-bit video processing and export, preserving higher-bit-depth SDR footage with compatible codecs independently of HDR settings, alongside improved color-range handling and HDR metadata preservation.
• Added DLSS Super Resolution to the standalone Upscale workflow, allowing you to choose between DLSS and RTX Video Super Resolution for supported image and video processing.
• Completely reworked Neural Rendering video previews with selectable 3, 5, 10, 20, and 30 second preview ranges, enhanced sections displayed directly on the timeline, and seamless playback between original and processed footage.
• Added intelligent preview caching, reusing completed processing stages when later adjustments change, reducing unnecessary reprocessing and making iterative previews more efficient.
• Improved Coloring previews with accelerated in-memory LUT processing when Coloring is the only active stage, providing faster feedback when adjusting color settings.
• Improved original-versus-processed comparisons with synchronized zooming and panning, double-click zoom reset, and more consistent side-by-side viewing for images and videos.
• Improved the Export workflow by remembering the last selected output location and folder between application sessions.
• Improved GPU-accelerated processing, with expanded CUDA-based video processing paths, automatic fallback when GPU decoding is unavailable, and improved DLSS handling across scene transitions.
• Added GPU memory diagnostics to Neural Rendering reports, tracking peak VRAM usage when supported NVIDIA monitoring data is available.
• Improved presets and settings compatibility, saving the new processing stages, their order, and individual configurations while migrating settings from earlier versions.