Description
Topaz Video Overview
Topaz Video 1is the first release under the rebranded Topaz Studio line, carrying forward the legacy of Topaz Video AI with new models, interface improvements, and workflow refinements. This version focuses on enhancing detail recovery, reducing noise, and giving creators more control while modernizing the licensing system. It’s designed for video editors, content creators, and professionals who want to upscale, denoise, and sharpen their footage with advanced AI technology.
Topaz Video Features
Next-Generation Denoising (Nyx XL)
A new AI model that cleans up noise while holding onto small details, making footage look more natural even in low light or older recordings.
Sharper Restoration (Starlight Sharp Variant)
An improved version of the Starlight model that enhances clarity and definition, especially useful for blurry or low-resolution clips.
Real-Time Frame Preview
Allows users to test how adjustments will affect the current frame instantly, speeding up decision-making without long render waits.
General Enhancement Preset
A one-click solution for everyday improvements. A simple slider fine-tunes multiple enhancement factors like detail, sharpness, and artifact removal all at once.
Refined User Interface
A reorganized control panel that groups tools logically, making it easier to switch between models, presets, and output options.
Broader Hardware Support
Expanded compatibility with NVIDIA, AMD, and Apple Silicon GPUs, though advanced models require higher VRAM for smooth performance.
Flexible Output Options
Users can upscale, denoise, and sharpen footage while balancing quality with performance depending on their hardware or cloud access.
Subscription Licensing
Moves toward a subscription-based model for updates and premium models, while still honoring previous perpetual licenses for existing users.
Note
I haven’t tested the Starlight Sharp model because I don’t have 16 GB of VRAM. But it will probably work too if I replace runner.exe. This is what was previously version 7.2.0.2b.
Installation
To activate the program:
1. Replace rlm1611.dll in your “Program FilesTopaz Labs LLCTopaz Video AI” folder.
2. Copy Topaz.lic to your models folder (usually it’s here: “C:ProgramDataTopaz Labs LLCTopaz Video AI BETAmodels”
To activate the ofx plugin:
1. Replace rlm1611.dll in “C:Program FilesCommon FilesOFXPluginsTopaz Video AI.ofx.bundleContentsWin64”
To remove the watermark from Starlight, replace the runner.exe file in the models folder.
Starlight can also be launched directly using runner with the following arguments:
runner “source file path” “output_folder/%06d_temp.jpg” –start-frame-idx 0 –end-frame-idx 40173 –ffmpeg-encoding “-pix_fmt rgb24 -q:v 2 -start_number 0 -movflags frag_keyframe+empty_moov+delay_moov+use_metadata_tags+write_colr -bf 0” –ffmpeg-preproc-filters scale=w=640:h=480,setsar=1 –upscale-factor 1 –max-gpu-mem 8
This command is for a source with a resolution of 720×480 but an aspect ratio of 4×3. That’s probably why the command line says 640×480. The other arguments are probably self-explanatory.
Just for your information, starlight takes the watermark from a jpg file located in the models folder and named sig2.bin. It’s not encrypted in any way, and its integrity isn’t verified.
File Information
Year/Release Date : 2025
Version : 1.0.0
Developer : Topaz Labs LLC
Developer’s website : www.topazlabs.com/topaz-video
Bit depth : 64bit
Interface language : English
Tablet : present
System Requirements
Windows Operating System – 10 or 11 (latest version recommended)
Processor – Intel or AMD with AVX2 instructions
System Memory (RAM) – 16 GB (32 GB or more recommended)
Graphics Card (NVIDIA) – NVIDIA GTX 900 series or higher, 6 GB VRAM
Graphics Card (AMD) – AMD Radeon 500 series or higher, 6 GB VRAM
Graphics Card (Intel) – Intel ARC A750
CPU Year – 2016 or newer
Warning: While integrated GPUs like Intel Iris Xe and AMD Radeon Integrated Graphics are technically supported in Video AI, we found that very few processes will run at acceptable speeds without a dedicated graphics card installed.
