TAU-HOME.COM
LOADING

Veyra: Open-Source Real-Time GPU Video Enhancement and Frame Generation for Windows

An overview of Veyra, an open-source C++ and DirectX 12 tool that enhances local videos, capture card feeds, and console streams using real-time super-resolutio

tau · November 16, 2026

#Veyra #GPU-Upscaling #Super-Resolution #Frame-Generation #DirectX12 #Windows-Tools

Veyra: Open-Source Real-Time GPU Video Enhancement and Frame Generation for Windows

An open-source project named Veyra (Likely7/Veyra-NRVideo) is gaining significant traction across the developer and PC enthusiast community for its ability to enhance video playback and capture feeds in real time using graphics hardware acceleration. Built in C++ on top of Microsoft's low-level DirectX 12 graphics API, Veyra is licensed under the GNU General Public License v3.0 (GPL-3.0) and maintained as an open-source repository on GitHub.

Veyra open-source real-time GPU video enhancer interface and pipeline display

Image source: https://github.com/Likely7/Veyra-NRVideo

While conventional media players traditionally rely on fixed pixel shaders or static post-processing filters, Veyra connects modern GPU acceleration, AI super-resolution models, and hardware frame generation into a single real-time playback environment. Rather than limiting these visual pipelines strictly to pre-rendered video files stored on local disks, Veyra applies identical enhancement passes to HDMI capture card feeds and live network game streams.

A Unified Graphics Pipeline Across Diverse Video Sources

The defining structural strength of Veyra is the diversity of input pipelines it supports within a unified post-processing engine. It goes well beyond standard offline media file playback.

  • External Video Capture: Veyra interfaces directly with HDMI capture cards, applying real-time enhancement passes to live preview signals with low latency.
  • Console and Network Streaming: The software accommodates live console streaming feeds, including PlayStation 5 (PS5) and experimental Xbox streams, alongside PC streaming workflows fully compatible with open-source Moonlight and Sunshine protocols.
  • Screen Capture and Audio Routing: High-refresh capture via Windows Graphics Capture (WGC), integrated AVerMedia multichannel audio handling, and optional virtual routing through VB-CABLE allow flexible capture configurations for streaming and recording.

This architecture enables practical creative configurations. For example, a console player can run a game on a PlayStation 5, route the HDMI signal through a PC capture card, and use Veyra to apply modern AI super-resolution and frame interpolation in real time before outputting the processed feed to a high-refresh PC monitor.

Super Resolution, Layered Noise Reduction, and Frame Generation

At the heart of Veyra lies a modular graphics pipeline designed to reconstruct visual clarity, suppress encoding artifacts, and smooth motion on the fly.

The first foundational pillar is Super Resolution (SR). Veyra integrates dual-pipeline super-resolution configurations, supporting both NVIDIA RTX Video Super Resolution (RTX Video SR) / DLSS and AMD FidelityFX Super Resolution (FSR). Users can configure sophisticated two-pass chains, such as executing an AI upscale followed by an adaptive downsample, or executing distinct secondary passes to sharpen edge details without creating excessive ringing artifacts.

The second pillar is layered Noise Reduction (NR) and anti-aliasing. To counteract macroblocking and compression noise common in video streams and capture cards, Veyra introduces multi-stage NR processing where each layer's intensity can be calibrated from 25% to 100%. To clean up jagged raster edges, independent anti-aliasing passes—including FXAA, SMAA, and MLAA—can be inserted into the processing chain. According to official benchmark measurements on an NVIDIA GeForce RTX 5070, processing standard 4K frames incurs an execution cost of roughly 0.15 ms for FXAA and 0.30 ms for SMAA/MLAA.

The third pillar is Frame Generation (FG) combined with low-latency management. By inserting computationally generated intermediate frames between consecutive source frames, Veyra upgrades standard 60 FPS feeds to high-refresh display cadences. A full-frame deduplication pass resilient to sparse noise prevents synthetic frame artifacts, while integrated toggles for NVIDIA Low Latency (Off / On / On + Boost) help minimize input latency during interactive streaming.

Visual Node-Based Editing and Flexible Output Workflows

To make complex, multi-stage GPU pipelines manageable, Veyra provides multiple interface modes tailored to different user workflows.

Beyond a streamlined 'Cinema' playback mode and a tabular 'Professional List' configuration view, Veyra features a complete 'Nodes' graph editor. In this view, enhancement stages appear as interconnected graphical blocks. Users can construct custom directed graphs, altering the execution order of super-resolution passes, noise reduction models, and color grading steps with full visual feedback. Settings can be saved as modular presets or complete session files.

In addition to real-time playback, Veyra incorporates an export queue for recording. Users can capture the fully enhanced visual stream alongside multichannel audio directly to MP4 or MKV containers, set custom trim points, and preserve embedded subtitle tracks, turning the tool into a standalone video enhancement and capture workstation.

Hardware Requirements, Distribution Packages, and Practical Constraints

Because Veyra operates close to the hardware layer, specific system parameters and setup considerations must be taken into account before deployment.

The software is strictly tailored for 64-bit Windows 11 running DirectX 12. Distribution packages are split by GPU vendor; users must choose the appropriate release archive, such as Veyra-2.1.5-NVIDIA-win64-portable.zip for NVIDIA systems or the dedicated AMD package for Radeon hardware.

Memory overhead is another critical consideration. Running high-resolution 4K video feeds through multi-layer neural noise reduction, dual super-resolution passes, and real-time frame generation creates heavy video RAM (VRAM) demand. Veyra includes real-time telemetry graphs to track the VRAM budget, and users running mid-tier GPUs must balance effect layers and scaling factors to ensure smooth real-time execution without dropping frames.

Sources

Impact-Site-Verification: 9b3ebd10-6c45-44cf-a014-5b8c19b7b285