ANA_DE_ARMAS_LORA is a specialized LoRA (Low-Rank Adaptation) fine-tuned on stabilityai/stable-diffusion-xl-base-1.0, expertly designed for generating ultra-photorealistic portraits inspired by Ana de Armas. This model captures her distinctive features, elegant expressions, and natural beauty with stunning cinematic quality.
portrait of Ana de Armas, professional studio lighting, elegant expression, photorealistic quality
| Parameter | Value |
|---|---|
| Base Model | stabilityai/stable-diffusion-xl-base-1.0 |
| Architecture | LoRA (Low-Rank Adaptation) |
| Resolution | 1024x1024 |
| LoRA Rank | 16 |
| Precision | FP16 |
| Training Steps | ~500 |
| Optimizer | AdamW with cosine scheduler |
| Dataset Focus | Ana de Armas professional photos and portraits |
| Training Emphasis | Facial features, expressions, and signature look |
portrait of Ana de Armas, elegant expression, soft brown eyes, defined cheekbones, natural makeup, long dark hair, professional studio lighting, photorealistic, ultra high resolution, clean backgroundglamour portrait of Ana de Armas, confident smile, voluminous dark hair, subtle makeup highlighting her natural features, cinematic lighting, photorealistic, high fashion style, studio backgroundnatural portrait of Ana de Armas, minimal makeup, authentic expression, soft lighting, brown eyes, Cuban beauty, photorealistic quality, warm color palette, clean compositionprofessional headshot of Ana de Armas, business attire, confident expression, soft studio lighting, clean background, photorealistic, corporate style, high resolutionred carpet portrait of Ana de Armas, elegant evening makeup, sophisticated hairstyle, glamorous expression, dramatic lighting, photorealistic, high fashion, luxury aesthetic| Setting | Value | Description |
|---|---|---|
| LoRA Weight | 0.8-1.2 | Optimal balance for feature accuracy |
| CFG Scale | 7-10 | For best prompt adherence and quality |
| Steps | 30-40 | Recommended sampling steps for detail |
| Sampler | DPM++ 2M Karras or Euler a | Best quality samplers |
| Resolution | 1024x1024 or 768x1024 | Portrait-optimized ratios |
Ana de Armasdefined cheekbonesexpressive brown eyesnatural Cuban beautyelegant expressionsubtle smilelong dark hairvoluminous wavesprofessional stylingnatural textureelegant updocinematic lightingphotorealisticstudio lightingprofessional photographyultra high resolutionclean backgroundglamour portraitnatural beautyprofessional headshotred carpet styleeditorial fashion1from diffusers import StableDiffusionXLPipeline
2import torch
3
4# Load the base model
5pipe = StableDiffusionXLPipeline.from_pretrained(
6 "stabilityai/stable-diffusion-xl-base-1.0",
7 torch_dtype=torch.float16,
8 use_safetensors=True,
9 variant="fp16"
10)
11
12# Load the Ana de Armas LoRA
13pipe.load_lora_weights("yourusername/ANA_DE_ARMAS_LORA")
14pipe = pipe.to("cuda")
15
16# Generate an image
17prompt = "portrait of Ana de Armas, elegant expression, professional studio lighting, photorealistic"
18negative_prompt = "blurry, low quality, distorted, cartoon"
19
20image = pipe(
21 prompt=prompt,
22 negative_prompt=negative_prompt,
23 num_inference_steps=35,
24 guidance_scale=8.5,
25 width=1024,
26 height=1024
27).images[0]
28
29image.save("ana_portrait.png")ComfyUI/models/loras/ directorymodels/Lora/ folder<lora:ANA_DE_ARMAS_LORA:0.9> in your prompt#LoRA #StableDiffusion #AnadeArmas #Portrait #Photorealistic #Celebrity #AI #GenerativeAI #ComputerVision #MachineLearning