Instructions to use faisalKSA/Arabic-Dallah with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use faisalKSA/Arabic-Dallah with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", dtype=torch.bfloat16, device_map="cuda") pipe.load_lora_weights("faisalKSA/Arabic-Dallah") prompt = "-" image = pipe(prompt).images[0] - Inference
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- Draw Things
- DiffusionBee
| tags: | |
| - text-to-image | |
| - lora | |
| - diffusers | |
| - template:diffusion-lora | |
| widget: | |
| - output: | |
| url: images/ComfyUI_00327_.png | |
| text: '-' | |
| - output: | |
| url: images/ComfyUI_00326_.png | |
| text: '-' | |
| - output: | |
| url: images/ComfyUI_00325_.png | |
| text: '-' | |
| - output: | |
| url: images/ComfyUI_00322_.png | |
| text: '-' | |
| - output: | |
| url: images/ComfyUI_00320_.png | |
| text: '-' | |
| - output: | |
| url: images/ComfyUI_00317_.png | |
| text: '-' | |
| - output: | |
| url: images/ComfyUI_00316_.png | |
| text: '-' | |
| - output: | |
| url: images/ComfyUI_00314_.png | |
| text: '-' | |
| - output: | |
| url: images/ComfyUI_00313_.png | |
| text: '-' | |
| - output: | |
| url: images/ComfyUI_00310_.png | |
| text: '-' | |
| - output: | |
| url: images/ComfyUI_00302_.png | |
| text: '-' | |
| - output: | |
| url: images/ComfyUI_00300_.png | |
| text: '-' | |
| - output: | |
| url: images/ComfyUI_00299_.png | |
| text: '-' | |
| - output: | |
| url: images/ComfyUI_00298_.png | |
| text: '-' | |
| - output: | |
| url: images/ComfyUI_00297_.png | |
| text: '-' | |
| - output: | |
| url: images/ComfyUI_00296_.png | |
| text: '-' | |
| base_model: stabilityai/stable-diffusion-xl-base-1.0 | |
| instance_prompt: ArDallah | |
| license: cc | |
| # Arabic-Dallah | |
| <Gallery /> | |
| ## Model description | |
| ## ✨ ArDallah - The Spirit of Arabian Heritage [SDXL] ✨ | |
| Welcome to the ArDallah LoRA, a model meticulously trained to bring one of the most iconic symbols of Middle Eastern culture into the world of AI art. | |
| ### 🏺 The Story Behind the Model | |
| In Arab culture, the Dallah (الدلة) is far more than a mere coffee pot. It is a timeless symbol of generosity, honor, and deep-rooted hospitality. For generations, it has stood at the center of gatherings, welcoming guests with the warmth of traditional Arabic coffee. Because this heritage is incredibly precious to me, I created this LoRA to preserve and celebrate this beautiful cultural legacy, ensuring its intricate details are accurately represented in modern digital art. | |
| ### ⚙️ Technical Details & Dataset | |
| I put a lot of care into "cooking" this model to ensure it delivers the highest possible fidelity: | |
| * Base Model Compatibility: Fully optimized and tailored exclusively for SDXL. | |
| * Dataset Quality: Trained using high-definition 1024x1024 images to maintain crisp generation, ensuring every golden reflection and curved contour remains flawless. | |
| * Focus: The dataset captures the Dallah in various authentic states—both as a standalone object and surrounded by its traditional accompaniments, such as Arabian coffee cups (*Finyan*), fresh dates, and luxurious hospitality setups. | |
| ### 📈 Recommended Settings | |
| To get the best results without burning the image or losing the metallic realism: | |
| * **Optimal LoRA Weight:** 0.85 to 1.10 (Adjust within this range depending on how heavy or detailed your prompt is). | |
| * Clip Skip: 1 (Crucial for SDXL to properly interpret the tokens). | |
| ### 💡 Tips for Successful Prompting | |
| * The Trigger Word: Always start your prompt with the unique trigger word: ArDallah. | |
| * Material Descriptions: Guide the model by specifying the texture you want, such as highly polished gold metal, smooth reflective surface, or hammered and sculpted metal. | |
| * Keep Tokens: If you use tag shuffling, ensure your generator keeps the first token (ArDallah) fixed so the core concept never gets lost. | |
| ### 📝 Prompt Examples to Try | |
| > 📸 Example 1: Clean Studio Shot | |
| > A detailed studio photograph of a highly polished, single ArDallah (Arabic coffee pot). The ArDallah is made of highly reflective, smooth golden metal, and it is positioned on a seamless white background. Its iconic conical lid, adorned with a decorative finial, is shown slightly open and resting against the neck. The curved handle and the long, graceful crescent-shaped spout are prominently visible. This image showcases the specific construction, materials, and form of the ArDallah object. | |
| > | |
| > ☕ Example 2: Luxurious Hospitality Scene | |
| > A highly detailed, elegant still life photograph set on a polished, circular wooden table. Centered is a magnificent, single ArDallah (Arabic coffee pot), crafted from deeply hammered and sculpted gold metal with an ornate conical lid, sitting atop a matching decorative gold plinth. Surrounding it are two clear glass coffee cups with delicate gold rims, and a selection of intricately prepared traditional mini desserts. The background is softly blurred with a patterned feature wall and natural foliage, creating a serene and luxurious cultural atmosphere. | |
| > | |
| ### ⚡ Support & Feedback | |
| I hope this model helps you create breathtaking, culturally rich artwork. If you enjoy using it, please consider sharing your generations, leaving a review, and supporting me with some Buzz! Your encouragement helps me fund the compute time to train and release even more unique cultural assets. | |
| Thank you for your support! 🙏✨ | |
| ## Trigger words | |
| You should use `ArDallah` to trigger the image generation. | |
| ## Download model | |
| [Download](/faisalKSA/Arabic-Dallah/tree/main) them in the Files & versions tab. | |