When you come up with a scene, you can directly give it to GPT to enrich the details. At the beginning, you can tell it this: 'I am using AI drawing software for creation, and I need you to help me write some prompts in natural language. Start the sentence like this:

This poster shows a smartphone against a dark background. The phone screen reveals a miniature scene of a xxxxxxx landscape, seamlessly integrated into the phone’s frame. In this sentence, xxxxxxx represents the specific scene. After this sentence, add some specific details about the scene.

Inference

python

import torch
from diffusers import FluxPipeline

pipe = FluxPipeline.from_pretrained("black-forest-labs/FLUX.1-dev", torch_dtype=torch.bfloat16)
pipe.load_lora_weights("Shakker-Labs/FLUX.1-dev-LoRA-Micro-landscape-on-Mobile-Phone", weight_name="FLUX-dev-lora-micro-landscape.safetensors")
pipe.fuse_lora(lora_scale=0.7)
pipe.to("cuda")

prompt = "This poster shows a smartphone against a dark background. The phone screen reveals a miniature stereoscopic scene of New York City, seamlessly integrated into the phone’s frame."

image = pipe(prompt, 
             num_inference_steps=24, 
             guidance_scale=3.5,
            ).images[0]
image.save(f"example.png")

Online Inference

You can also run this model at Shakker AI, where we provide an online interface to generate images.

Acknowledgements

This model is trained by our copyrighted users DalaBengba. We release this model under permissions. The model follows flux-1-dev-non-commercial-license.