compcard_pose_lora_v1

Prompt

compcard_pose_v1, a hyperrealistic 2x2 grid actor compcard on a white seamless background. [layout — mandatory 2x2 compcard grid] top-left: tight headshot frontal, neutral expression, direct eye contact, 85mm portrait top-right: clean right profile, head and shoulders, relaxed, 85mm bottom-left: half body (waist-up), arms relaxed or lightly crossed, subtle charisma, 50mm bottom-right: full body, balanced stance, hands visible, feet fully in frame, 35mm Each quadrant shows the SAME persona, identical in face, hair, clothes, and anchors, only the camera angle and framing changes. Style: hyperrealism, photorealistic portrait photography, cinematic sharpness, natural skin texture, true colors. Persona: male, 30y, blogger rapper crossfitter, russian heritage with afro-russian roots, modern street poet mixing classic verse with trap beats Face & hair: oval face with high forehead, pronounced cheekbones, straight medium-width nose, full lips, thick dark eyebrows, deep almond-shaped dark brown eyes with expressive gaze, medium-length thick curly dark brown hair with natural volume, distinct wide sideburns, accessory: thin gold hoop earring on left ear Body: mesomorph, athletic crossfit build, broad shoulders, defined arms, medium height, warm olive-brown skin, subtle wrist tattoo: quill icon on right wrist Clothes: black bomber jacket over white oversized tee, slim black jeans, clean white sneakers, thin gold chain, smartwatch, palette: black, white, gold, deep burgundy Mood: charismatic, witty, confident, lyrical, slightly provocative performer energy Environment: minimal studio, white seamless, no props unless wearable Key persona feautures: high forehead, distinct sideburns, curly dark hair, black bomber + gold chain, confident direct gaze [style] studio daylight, soft shadows, crisp focus, no vignette, no borders, no text, no watermark

Prompt

compcard_pose_v1, a hyperrealistic 2x2 grid actor compcard on a white seamless background. [layout — mandatory 2x2 compcard grid] top-left: tight headshot frontal, neutral expression, direct eye contact, 85mm portrait top-right: clean right profile, head and shoulders, relaxed, 85mm bottom-left: half body (waist-up), arms relaxed or lightly crossed, subtle charisma, 50mm bottom-right: full body, balanced stance, hands visible, feet fully in frame, 35mm Each quadrant shows the SAME persona, identical in face, hair, clothes, and anchors, only the camera angle and framing changes. Style: hyperrealism, photorealistic portrait photography, cinematic sharpness, natural skin texture, true colors. Persona: male, 30y, blogger rapper crossfitter, russian heritage with afro-russian roots, modern street poet mixing classic verse with trap beats Face & hair: oval face with high forehead, pronounced cheekbones, straight medium-width nose, full lips, thick dark eyebrows, deep almond-shaped dark brown eyes with expressive gaze, medium-length thick curly dark brown hair with natural volume, distinct wide sideburns, accessory: thin gold hoop earring on left ear Body: mesomorph, athletic crossfit build, broad shoulders, defined arms, medium height, warm olive-brown skin, subtle wrist tattoo: quill icon on right wrist Clothes: black bomber jacket over white oversized tee, slim black jeans, clean white sneakers, thin gold chain, smartwatch, palette: black, white, gold, deep burgundy Mood: charismatic, witty, confident, lyrical, slightly provocative performer energy Environment: minimal studio, white seamless, no props unless wearable Key persona feautures: high forehead, distinct sideburns, curly dark hair, black bomber + gold chain, confident direct gaze [style] studio daylight, soft shadows, crisp focus, no vignette, no borders, no text, no watermark

Model description

compcard_pose_v1

📌 Description

A custom LoRA trained for generating model comp cards.
The focus is on poses rather than faces. This model is intended for fashion / commercial comp card layouts, studio-style photography, and clean body posture generation.

🎯 Trigger word

Use in your prompt:
compcard_pose_lora_v1

###🔑 Notes

•	The LoRA is designed to capture poses only (not style, not faces).
•	Works best with studio / clean backgrounds.
•	Can be combined with other LoRAs (clothing, style, makeup, etc.) for better results.

###PROMPT template: compcard_pose_v1, SINGLE COMPOSITE IMAGE, mandatory 2x2 compcard grid on a pure white seamless background. Layout:

• top-left: tight headshot frontal, neutral expression, direct eye contact, 85mm portrait
• top-right: clean right profile, head and shoulders, relaxed, 85mm
• bottom-left: half body (waist-up), arms relaxed or lightly crossed, subtle charisma, 50mm
• bottom-right: full body, balanced stance, hands visible, FEET FULLY IN FRAME, 35mm

The SAME persona appears in all four tiles. IDENTICAL face, hair, facial hair, clothes, and accessories in every tile. Only angle and framing change.

Persona: Face & hair: Body: Clothes: Mood: Environment: minimal studio, white seamless, no props unless wearable

Key persona features:

[style] studio daylight, soft shadows, crisp focus, natural skin texture, true colors, NO vignette, NO borders, NO text, NO watermark, uniform gutters between tiles

[hard negatives] no single-photo output, no colored backdrop, no hats or caps, no background scenes, no different outfits, no extra props

⚙️ Training details

• Base model: FLUX.1 [dev]
• Steps: 1200
• Dataset: curated set of comp card images (~18MB)
• File size: ~85 MB
• Loss: standard LoRA fine-tuning

📂 Files

• `pytorch_lora_weights.safetensors` — LoRA weights
• `config.json` — training configuration

🚀 Usage

```python from diffusers import StableDiffusionPipeline import torch

pipe = StableDiffusionPipeline.from_pretrained( "black-forest-labs/flux-dev-lora", torch_dtype=torch.float16 ).to("cuda")

pipe.load_lora_weights("your-username/compcard_pose_v1")

prompt = "compcard_pose_v1, model comp card, black outfit, studio light" image = pipe(prompt).images[0] image.save("pose.png")

Trigger words

You should use compcard_pose_lora_v1 to trigger the image generation.

Download model

Download them in the Files & versions tab.