Releases·Tool Release
Dev open-sources Vison, an MIT-licensed Vulkan desktop app running FLUX.1 Schnell, Z-Image and Wan locally via stable-diffusion.cpp on 6GB GPUs
A developer released Vison, an MIT-licensed Windows desktop app for image and video generation that runs FLUX.1 Schnell, Z-Image and Wan entirely on local GPUs via stable-diffusion.cpp, ggml and Vulkan. It targets low-VRAM hardware (a 6GB laptop GPU) with no CUDA and no cloud service, keeping prompts on-device. Practitioners get a fully local, permissively licensed frontend, though model weights carry their own licenses.
5 engagement1 sourceFri, Oct 2, 2026, 02:10 PM
AI Overview
- JayRGadekar released Vison under MIT, built on stable-diffusion.cpp, ggml and vision.cpp, running GGUF image and video generation locally on Windows with no CUDA or cloud account
- Vison bundles an LGPL ffmpeg build with libvpx and refuses to package GPL or non-free versions; output video is royalty-free VP9 in WebM
- Model weights are not MIT-licensed and carry separate terms from each publisher, so commercial use of FLUX.1 Schnell or Wan output requires checking each licence
- The developer states the project is early-stage, one-person, Windows-only on a 6GB laptop GPU, with macOS/Linux ports described as mostly packaging work
Related
Products·launched
Kokon: zero-setup desktop app for Qwen-Image 2.1 photo editing on local GPUs
Kokon is a free, open-source desktop app that runs Qwen-Image 2.1 locally for GPT-Image-style photo editing without Python environments or manual configuration. It auto-detects the GPU, downloads the right GGUF models, and sizes each job's resolution and VRAM split automatically. It targets people with 12 GB+ GPUs who want simple local image editing.
Kokon is a free, open-source desktop app that runs Qwen-Image 2.1 locally for GPT-Image-style photo editing without Python environments or manual configuration. It auto-detects the GPU, downloads the right GGUF models, and sizes each job's resolution and VRAM split automatically. It targets people with 12 GB+ GPUs who want simple local image editing.
0 engagement1 sourceredditWed, Oct 14, 2026, 05:10 PM
Releases·Tool Release
Show HN: Janus – Go binary that runs GGUF models via Vulkan on AMD/Intel/Nvidia
Show HN: Janus – Go binary that runs GGUF models via Vulkan on AMD/Intel/Nvidia
Show HN: Janus – Go binary that runs GGUF models via Vulkan on AMD/Intel/Nvidia
112 engagement1 sourcehackernewsThu, Oct 1, 2026, 08:36 PM
Developments·Paper
Generative Cinematographer: Composing Camera and Object Motion in 3D
A new arXiv paper, Generative Cinematographer (GenCine), introduces a system for jointly controlling camera and foreground object motion in video generation. It lifts a single image into an editable 3D scene scaffold where artists author camera paths and move foreground regions with local 3D motion handles. Controls are fed to a pretrained Wan video model via guidance maps, trained with a lightweight guidance branch and LoRA adapters.
A new arXiv paper, Generative Cinematographer (GenCine), introduces a system for jointly controlling camera and foreground object motion in video generation. It lifts a single image into an editable 3D scene scaffold where artists author camera paths and move foreground regions with local 3D motion handles. Controls are fed to a pretrained Wan video model via guidance maps, trained with a lightweight guidance branch and LoRA adapters.
0 engagement1 sourcearxivThu, Oct 1, 2026, 05:58 PM
Products·building
ScriptStudio — open-source local AI video studio for script-to-video workflows
ScriptStudio is an open-source desktop application that turns written scripts into editable videos using fully local AI generation. LLMs via optional Ollama handle script writing and planning, while local models generate narration, video, and music. It targets creators who want script-to-video production without cloud services or sending data off their machine.
ScriptStudio is an open-source desktop application that turns written scripts into editable videos using fully local AI generation. LLMs via optional Ollama handle script writing and planning, while local models generate narration, video, and music. It targets creators who want script-to-video production without cloud services or sending data off their machine.
1 engagement1 sourceredditWed, Sep 23, 2026, 04:45 AM
Products·launched
Radiant Canvas — local AI image generation and editing app for Mac
Radiant Canvas is a Mac app for running multiple image generation and editing models locally on Apple Silicon. It bundles models like FLUX.2 Klein and Qwen Image Edit behind one interface, removing the need to maintain separate setups. It targets Mac users who want offline generation without cloud costs or credits.
Radiant Canvas is a Mac app for running multiple image generation and editing models locally on Apple Silicon. It bundles models like FLUX.2 Klein and Qwen Image Edit behind one interface, removing the need to maintain separate setups. It targets Mac users who want offline generation without cloud costs or credits.
3 engagement1 sourceredditMon, Sep 14, 2026, 03:39 PM