AI Italian Medieval Architecture XL 🇮🇹

LORA
Original


Updated:

8.7K

AI Italian Medieval Architecture XL 🇮🇹

I created this experimental LoRA to bring some medieval Italy to SDXL as I found it lacking when I attempted generating castles, which were invariably rendered in English, French or German architectural style. Works well with different Checkpoints and other LoRAs and it was trained on sharp images with many descriptive tags, so it tends toward realism and generates accurate results with fewer steps (5-10 for turbo checkpoints / 23-27 for others) and a CFG scale of 5-7. I suggest a weight from 0.8 to 1.0 but do not exceed these parameters or it will produce artifacts.

Getting the AI to understand the complexity of a castle through 2D images is a tough task, so I recommend starting with this sample prompt while adding/changing details as you go:

(Documentary photograph:1.3) of a magnificent (Italian castle:1.4), 14th century, (golden ratio:1.3), [YOUR SETTINGS/ENVIRONMENT], BREAK shot on Canon EOS 5D, eye level, Fujicolor Pro film, in the style of [PHOTOGRAPHER], (photorealistic:1.3), (soft diffused lighting:1.2), vignette, highest quality, original shot. BREAK Front view, well-lit, (perfect focus:1.2), award winning, detailed and intricate, masterpiece.

  • Add (medieval architecture:1.3) or it will generate a chateau mansion.

  • Add (brick wall:1.2) or (stone wall:1.2) to define the building material.

  • Add (square towers:1.3) and/or (cylindrical towers:1.3) for the towers shape.

  • Replace (Italian castle:1.4) with (Italian stronghold:1.4) or (Italian medieval plaza:1.4) in case.

  • Optionally add (corbels:1.2), (battlements:1.2), (scarp:1.2), (moat filled with water:1.2), (merlons:1.2) and (mullioned window:1.2) if you want SDXL to include such details.

In case of artifacts, add the same entries to the negative prompt with the prefix "bad," i.e. (bad mullioned window:1.2). For best results I suggest starting with one of the example images.

Sampling Method / Upscaler: DPM++ SDE + 4x-UltraSharp

VAE: sdxl_vae.safetensors

Reprints and commercial uses are strictly forbidden: there is no freedom of panorama in Italy, and the last thing I want is some zealous bureaucrat coming to hassle me.

Happy generation! 🏰 👑 ⚔️

Version Detail

SDXL 1.0
2200
5
This LoRA requires knowledge of architectural terms and its use is quite complex. I highly recommend thoroughly reading the description below.

Project Permissions

    Use Permissions

  • Use in TENSOR Online

  • As a online training base model on TENSOR

  • Use without crediting me

  • Share merges of this model

  • Use different permissions on merges

    Commercial Use

  • Sell generated contents

  • Use on generation services

  • Sell this model or merges

Comments

Related Posts

Describe the image you want to generate, then press Enter to send.