Veyra-NRVideo

Open source alternative to Topaz Video AI

1.6K stars
GPL-3.0
Active this week
Last commit 2 days ago

Last verified Nov 14, 2026

Run real-time GPU video enhancement on Windows with AI super resolution, denoising, HDR, and frame generation for local files and capture cards.

Visit website GitHub1.6K

Read the full breakdownWho it is for, how it works, trade-offs, the quick start and the FAQ

who it's for

Who Veyra-NRVideo is for#

Console gamers watching PS5 or Xbox via capture card

Connect a capture card to a PS5 or Xbox, open Veyra, select the capture device, and enable super resolution and frame generation. The enhanced picture renders in real time on the Windows display. PS5 Remote Play over IP is also supported without a capture card, using the console's pairing code.

Skip if:

You play PC games with native DLSS or FSR built into the game engine. Veyra adds value at the boundary between a console output and a PC display, not for in-engine upscaling.

Content creators processing older or compressed local footage

Add local video files to the export queue, enable NR and SR, and export to MP4 or MKV. The pipeline denoises and upscales footage using the GPU. Useful for archival footage, compressed screen recordings, and older game captures that need quality improvement before publishing.

Skip if:

Your source footage is already high quality (native 4K, clean lighting, low compression). NR on clean footage adds processing time without visible benefit.

Home theater users enhancing streaming or Blu-ray rips

Use Cinema mode with SR and NR enabled to watch local files or capture card sources on a large display. Frame generation multiplies effective frame rate on 24fps film content. Per-layer NR reduces encode artifacts in compressed web streams without an export step.

Skip if:

You use a smart TV or streaming device for playback. Those have built-in upscaling; Veyra only runs on a Windows PC in the signal path.

Developers evaluating open source GPU video pipelines

The node-based processing graph exposes SR, NR, color, and frame generation stages as connectable nodes with independent parameters. The repo ships full source under GPL-3.0, making it a reference for real-time GPU video processing with DLSS 5, DirectX 12, and AMD NR models on Windows.

Skip if:

You need a cross-platform library or SDK rather than a desktop application. Veyra is a GUI app; there is no headless or programmatic API.

the problem

The problem it solves#

Video from capture cards, Remote Play sessions, and older local files rarely matches the quality of a modern 4K display. Capture card feeds carry compression artifacts from consoles; PS5 and Xbox streams lose detail over the network; files encoded at lower resolutions need upscaling for large screens. The standard paths for fixing this are either slow or costly: offline enhancement software processes one file at a time and takes minutes to hours per clip, while cloud-based upscaling services charge per export.

For live viewing through a capture card or streaming session the problem is worse. Prior options were limited to watching the lower-quality source as-is, or using an NVIDIA RTX Video overlay that covers only one media player and lacks denoising, frame generation, or capture card input. No broadly available free tool addressed real-time GPU enhancement across multiple source types in a single application.

how Veyra-NRVideo solves it

How it solves it#

Real-time AI super resolution

Upscales video and capture card feeds on the GPU using NVIDIA DLSS 5, RTX Video SR, or AMD NR models. Three RTX Video SR families with independent second-pass controls are available, plus DLSS, XeSS, and FSR as alternative backends. Processing runs at display frame rate without an offline export step.

Multi-layer noise reduction

Up to four independent NR layers, each with its own internal resolution, strength percentage, and picture control parameters. Nine manual controls cover hue, chroma, neutral tint, source color retention, lightness retention, shadows, highlights, local compression, and temporal stability. Layers can be positioned before or after super resolution in the pipeline.

Frame generation

Multiplies effective frame rate on NVIDIA and AMD GPUs by generating interpolated frames between captured ones. Supports a choice of backends and multipliers, with VSync, low-queue mode, and an output cap. Frame generation applies to both local video playback and live capture card or streaming sources.

Capture card and console streaming support

Processes video from USB and PCIe capture cards, PS5 Remote Play (via IP or PSN with an eight-digit pairing code), Sunshine/Moonlight PC streaming, Xbox streaming (experimental), and Windows screen capture. Enhancement runs live at playback rather than in a separate export step.

Node-based processing graph

A visual node editor lets users build a custom pipeline by connecting SR, NR, color, and frame generation nodes with draggable wires. Each node has independent parameters. This differs from List mode, where effects are toggled in a fixed order. Node mode is designed for live viewing rather than export.

Export queue with per-job trimming

Processes and saves local video files to MP4 or MKV with VBR quality control. The export queue supports reordering, per-job trimming, multiple audio tracks, and embedded subtitle selection. Pause, cancel, and retry are available. Video export uses at most internal 1080p NR inference composited to the original output size.

strengths · trade-offs

Strengths and trade-offs#

Strengths

GPL-3.0 licensed with no subscription costsVeyra is free to download and run with no account, no cloud dependency, and no per-use fees. Topaz Video AI and similar commercial tools charge $299+ for perpetual licenses or annual subscriptions. The portable package extracts and runs without an installer, with all runtimes bundled.
Real-time enhancement on live capture sourcesProcesses capture card feeds, PS5 Remote Play, Sunshine PC streams, and Xbox streams in real time at display frame rate. Competing paid tools like Topaz Video AI process only local files and produce output after an offline export run. Veyra's pipeline applies all effects before the frame reaches the display.
79,000+ downloads since September 2026As of November 8, 2026, the Veyra portable package had accumulated 79,332 downloads from GitHub Releases. The repo reached 1,625 stars in roughly two months from its September 2026 creation, indicating rapid adoption for a new project in a specialized GPU video category.
Layered noise reduction with nine picture controlsUp to four NR layers, each with independent strength, scale, and picture control parameters covering hue, chroma, shadows, highlights, and temporal stability. Delta strength reaches 5 per layer. No broadly available open source Windows video tool offers comparable per-layer NR granularity.

Trade-offs

-Windows 11 x64 onlyVeyra requires Windows 11 x64 and DirectX 12. There is no macOS or Linux version. Users on other platforms have no path to run it without a Windows installation.
-Requires a mid-to-high-end GPULayered NR, super resolution, and frame generation in combination demand significant GPU resources and VRAM. Most local tests used an RTX 5070, a mid-to-high-end card. Users with integrated graphics, older GPUs without DLSS or FSR support, or limited VRAM may not be able to run all effects simultaneously at high resolution.
-44 open GitHub issues; young codebaseThe project had 44 open GitHub issues as of November 2026 and was created in September 2026. Some features are marked experimental (Xbox streaming, AMD RX 7000 NR), and each release includes per-version migration steps for settings files. The user guide notes the 1.4.4 UI guide is obsolete; settings migrate between major versions.
-Node mode does not support offline exportThe visual node editor cannot be used for file export; only List mode supports the export queue. Users who build a node graph for live viewing must switch to List mode and recreate the effect order before exporting. Graphs are not automatically converted between modes.

versus alternatives

Veyra-NRVideo vs alternatives#

Veyra vs Topaz Video AI

Veyra and Topaz Video AI both apply AI-based super resolution, denoising, and frame interpolation to video content, but they target different workflows. Topaz Video AI is a paid desktop application focused on offline batch enhancement of local files. Veyra is GPL-3.0 licensed and open source, designed for real-time enhancement of live sources including capture cards, PS5 Remote Play, and PC game streaming, alongside local file export.

Feature	Veyra	Topaz Video AI
License	GPL-3.0 (free)	Proprietary ($299+ perpetual)
Deployment	Portable Windows app	Local app (Windows and macOS)
Real-time preview	Yes, all sources	No: export only
Capture card support	Yes	No
Console streaming	PS5, Xbox (experimental), Sunshine	No
Super resolution	DLSS 5, RTX Video SR, XeSS, FSR	Dedicated Topaz AI models
Frame generation	Yes	Yes (frame interpolation)
Export formats	MP4, MKV	Multiple formats including ProRes
OS support	Windows 11 x64 only	Windows and macOS

Veyra is the better fit for Windows users who want free, real-time enhancement of live capture card or streaming sources, with no subscription or export wait time. The GPL-3.0 license means there are no per-use fees and no cloud upload required.

Topaz Video AI is the better fit when you need macOS support, professional output formats such as ProRes for editing pipelines, or the proprietary AI models Topaz has trained specifically for video artifact removal on local files. Topaz also has a more established batch workflow for complex multi-clip projects.

install · quick start

Quick start#

bash

Download the NVIDIA or AMD portable package for Windows from GitHub Releases: https://github.com/Likely7/Veyra-NRVideo/releases/latest

tech stack · detected from GitHub

What it's built on#

Languages: C++Python

frequently asked

FAQ#

Does Veyra work with AMD GPUs?

Yes, with limitations. AMD NR on RX 9000 is supported with Small and Original models. Experimental RX 7000 RedDuke NR support was added in version 2.1.2. FSR is available as a super resolution backend on AMD GPUs. Driver HIP 7 is required for AMD NR. Performance and feature parity with NVIDIA varies by GPU generation; check the release notes for your card.

Is Veyra free to use?

Yes. Veyra is GPL-3.0 licensed and distributed as a free portable package from GitHub Releases. No account, subscription, or internet connection is required to run the application. The GPL-3.0 license permits personal and commercial use but requires distributing source changes if you release a modified version as a product.

Can Veyra enhance video from a PS5 or Xbox?

Yes. PS5 connection uses the built-in Remote Play path: enter the console's IP address or use PSN and provide the eight-digit pairing code from the console. Xbox support is experimental and uses device-code sign-in with remote features enabled on the console. Capture card input from either console is also supported without network streaming.

What GPU does Veyra require?

Veyra requires a DirectX 12 GPU running Windows 11 x64. NVIDIA RTX 20 series and later is fully supported for DLSS-based super resolution and frame generation. AMD RX 9000 and experimental RX 7000 support AMD NR. Most testing used an RTX 5070; older or low-VRAM GPUs can run the application but may not support all effects simultaneously at high resolution.

How do I upgrade between Veyra versions without losing my settings?

Extract the new package into a fresh folder, then copy only the runtime_local/user-data-2.0.0 folder from the old installation. Do not overlay incompatible DLLs or the old veyra.ini. The Settings > About screen shows your user data path. Detailed per-version migration steps are included in each release's changelog on GitHub.

also worth a look

Similar open-source tools#

God's Eye View

Featured

Alternative toFlightradar24

Real-time global intelligence on a 3D globe

64,214 GitHub stars
Licence MIT with third-party data restrictions

ytm-rs

Alternative toSpotify

YouTube Music client in Rust, with karaoke lyrics and 30 themes.

193 GitHub stars
Licence GPL-3.0

Excaliself

Alternative toExcalidraw+

Self-hosted Excalidraw with team space, folders, and live collab.

153 GitHub stars
Licence MIT

Renderdoc

Alternative toNVIDIA Nsight Graphics

Frame-capture graphics debugger for Vulkan, D3D, and OpenGL

11,304 GitHub stars
Licence MIT

Slint

Alternative toQt

Cross-platform native UI toolkit for Rust, C++, JS, and Python.

24,574 GitHub stars
Licence Triple-licensed: Royalty-free 2.0 / GPL-3.0-only / Commercial (Slint)

Litube

Alternative toYouTube Premium

Ad-free Android YouTube client with downloads and SponsorBlock.

593 GitHub stars
Licence GPL-3.0

by category

Open source tools in these categories#

132

Design & Creative

ComfyUI, Excalidraw, Godot Engine

Inbox Zero

Alternative toSuperhuman

AI email assistant that drafts replies and clears your inbox

12,987 GitHub stars
Licence AGPL-3.0
#Email

rakazo

Alternative toGrok Bot

AI teammates you own: your keys, your model, your machine.

7,869 GitHub stars
Licence Apache-2.0
#AI Agents

SnitchFeed

Live monitoring on LinkedIn, X, Reddit & more. Agentic social listening for startups.

Marketing & Growth

Advertise your product

The projects we add and every new guide. Free, unsubscribe any time.

Find open source alternatives to paid software

Veyra vs Topaz Video AI

Feature	Veyra	Topaz Video AI
License	GPL-3.0 (free)	Proprietary ($299+ perpetual)
Deployment	Portable Windows app	Local app (Windows and macOS)
Real-time preview	Yes, all sources	No: export only
Capture card support	Yes	No
Console streaming	PS5, Xbox (experimental), Sunshine	No
Super resolution	DLSS 5, RTX Video SR, XeSS, FSR	Dedicated Topaz AI models
Frame generation	Yes	Yes (frame interpolation)
Export formats	MP4, MKV	Multiple formats including ProRes
OS support	Windows 11 x64 only	Windows and macOS