ThalisAI/tilt-shift-flux-lora

Tilt Shift / Miniature Diorama — Style LoRA [Flux]

A style LoRA that transforms any scene into a photorealistic tilt-shift miniature diorama. Buildings become dollhouses, cities become model train sets, landscapes become tabletop terrain — all with that distinctive narrow focal plane and creamy bokeh that makes real things look like toys.

Gallery

Usage

• No trigger word needed — the style applies automatically when the LoRA is loaded
• Recommended strength: 0.7–1.0
• Base model: Flux.1 Dev
• Resolution: 1024×1024 or 1216×832

Tips

• Describe scenes with architecture, landscapes, or cityscapes for the strongest miniature effect
• Bird's-eye or elevated camera angles enhance the diorama illusion
• The LoRA naturally creates shallow depth-of-field — no need to prompt for it
• Works especially well with: villages, cities, harbors, train stations, castles, gardens
• When remixing Flux images, use CFG 3.5–4.0 for best results (the default 1.0 doesn't look good)

Example Prompts

• A medieval village at sunrise with thatched roofs and cobblestone streets, morning mist, warm golden light
• A pirate cove with wooden ships docked at a tiny harbor, lanterns glowing, evening atmosphere
• A Japanese garden with cherry blossoms, koi pond, stone lanterns, aerial view
• An enchanted library with towering bookshelves and floating candles, warm amber light

Links

• Gallery & Prompts
• Civitai
• All Models

About

Created by Thalis AI — LoRA training and generative AI experiments. Where the lotus-eaters forgot, we remember.

License

This model is released under the FLUX.1 Dev Non-Commercial License.