The open-source recipe behind a Magnific rival

What it does Clarity Upscaler is an open-source image enhancement pipeline built around Stable Diffusion. It uses a specific model stack—Juggernaut with ControlNet tile guidance and detail LoRAs—to generate textures and push resolution far beyond standard algorithms. The repository hosts the cog implementation, while the same logic is offered as a free ComfyUI workflow and an A1111 webUI preset.

The interesting bit The project’s real value lies in its inference recipe rather than a custom model: exact prompts, ControlNet weights, tiled diffusion settings, and denoising strengths that collectively mimic the look of Magnific. The author is unusually upfront about the business model: the Stable Diffusion pipeline is free and open, while the newer Flux-based upscaler is proprietary and lives behind the paid ClarityAI.co app.

Key highlights

• Positioned as a free, open-source Magnific alternative
• Supports output resolutions up to 13,000×13,000 pixels
• Includes multi-step upscaling, pre-downscaling, sharpen, and LoRA support
• Available as a ComfyUI workflow, A1111 configuration, or cog container
• Flux-based upscaling (better faces, text, and art) is offered only via the paid app and API

Caveats

• This repository is specifically the cog implementation; the README warns it is not the easiest entry point for beginners
• Flux upscaling is explicitly not open-source and requires the commercial service
• Some features listed in the update log—such as pattern upscaling—are not described in detail, so their availability in the open-source build is unclear

Verdict A solid choice for developers and AI artists who want to self-host a Magnific-style SD upscaler inside ComfyUI or A1111. Skip it if you are looking for the latest Flux-based quality or want a fully managed, one-click solution without wrangling model weights.