Generate images from text inside Unity with five curated local model profiles, from experimental mobile previews to high-end FLUX.2 and SDXL desktop generation.Local AI Image Generation brings offline text-to-image generation directly into Unity using Unity Inference Engine (Sentis). Once a compatible model has been downloaded, generation runs locally in the Unity Editor and built players, with no Python process, custom native inference plugin, web service, or separate runtime.The built-in Model Setup panel lets you review the licence for and download one of five curated, pinned model profiles from Hugging Face. Options range from experimental mobile generation to high-end FLUX.2 and SDXL desktop models. Model weights are not included in the Asset Store package, and internet access is required for the initial download.Windows desktop is the verified target for the large FLUX.2 and SDXL profiles. SDXS Mobile is experimental on recent native mobile devices and includes an experimental WebGL/WebGPU compatibility path; availability and performance depend on browser and device compute support. Other platforms supported by Unity Inference Engine may work, but are not verified by this release. Test your chosen model on its target hardware before shipping.The supplied DemoScene demonstrates the workflow. Use the included generator prefabs directly or integrate the common C# API into your own systems.Requirements:• Unity 6000.5.5f1 or newer• Unity Inference Engine 2.6.1• Unity UI 2.5.0• A compute-capable desktop GPU, or a recent mobile device for the experimental SDXS Mobile profileThe DemoScene additionally uses Universal Render Pipeline 17.5.0 and Input System 1.19.0. The generator runtime does not require the DemoScene or a particular render pipeline.The downloadable model profiles permit commercial use subject to their individual licence terms and use restrictions. Review the applicable licence in Model Setup before downloading. If you redistribute model files with your project, preserve the required licences, notices, attribution, and restrictions. Full model sources, pinned revisions, and licensing guidance are provided in Third-Party Notices.txt and Documentation/LICENSING.md.Asset uses Noto Sans SC under the SIL Open Font License 1.1 and modified Unity UI/TextMesh Pro shader sources under the Unity Companion License 1.4; see Third-Party Notices.txt in the package for details.- Offline text-to-image generation through Unity Inference Engine (Sentis)- Five curated profiles: SDXS Mobile, Stable Diffusion 1.5 Lite, DreamShaper Hyper, FLUX.2 Klein 4B and SDXL Studio- Approximate downloads from 450 MB to 10.23 GiB; practical targets from 4 GB shared RAM to 24 GB VRAM- Model Setup downloads pinned Unity model artifacts and their licence files from Hugging Face- Windows verified for FLUX.2 and SDXL; SDXS native mobile and WebGL/WebGPU paths are experimental- Common asynchronous C# generation API with progress callbacks and cancellation- No Python process, custom native inference plugin, web service, or separate runtime- Requires Unity 6000.5.5f1+, Unity Inference Engine 2.6.1 and Unity UI 2.5.0- DemoScene uses URP 17.5.0 and Input System 1.19.0; generator runtime is render-pipeline independentI used AI to refine and test my code, and to expand it for more model support. I also used it to create some marketing materials.
