Python CLI that turns prompts into images and rockets into GIFs

What it does

ImaginAIry is a command-line toolkit for generating images and videos with AI. It wraps Stable Diffusion and SDXL models behind a Pythonic interface, letting you create images from text, recolor photos, upscale details, or animate a still frame into a short video. It also exposes an optional local API server and web UI, though the README admits that web interface is “much smaller featureset compared to the command line tool.”

The interesting bit

The project recently rewrote its core onto the refiners library, which enabled SDXL support, image prompts, and self-attention guidance for more accurate details. It also layers in ControlNet guidance—pose, depth, edges, and even QR code structures—though notably not yet for SDXL.

Key highlights

• Generate images from text or existing images, with ControlNet structure guidance (pose, depth, canny edges, etc.)
• Produce short videos via Stable Video Diffusion from any input image, with frame interpolation and “bounce” looping
• Supports SDXL models and loading diffusers-format checkpoints from Hugging Face or local paths
• Includes niche utilities like automatic recolorization, detail upscaling via spandrel integration, and QR-code ControlNet
• Optional local API server with StableStudio web interface (alpha, limited features)

Caveats

• Video generation requires an Nvidia GPU and explicitly “does not work on Mac or CPU”
• Several features are currently fractured: ControlNet is unsupported for SDXL, and InstructPix2Pix editing is marked “broken as of 14.0.0”
• Windows support is described as “sometimes windows,” and Python 3.11 is not supported

Verdict

Good fit for developers who want a single Python CLI covering both image and video generation without cobbling together separate pipelines. Skip it if you need reliable Windows support, CPU-only inference, or polished SDXL ControlNet workflows.