Hello everyone! It is necessary to create a photorealistic image of two specific people in one scene using generative AI.
I have high-quality photographs of two real people (about 5–10 photos of each), which I am ready to provide for creating custom digital models (LoRA or equivalents). The task is to generate one photo based on these data, where both characters are present simultaneously in a realistic everyday scene.
If the result is high quality — I plan to order another 10–20 similar images (different scenes from daily life, travel, home, etc.).
What should be in the photo (scene description):
"Selfie of two people in a coffee shop in the USA.
The photo is taken with a hand (phone), both characters are smiling, sitting at a table with coffee or drinks.
Atmosphere — modern coffee shop in American style: windows, barista, shelves, lamps are visible in the background.
Lighting — natural or soft warm indoor light.
Style — as realistic as possible, like a regular Instagram photo."
Expected quality:
- Photorealism.
- Correct reproduction of both faces, without merging or distortion.
- Proper poses, realistic emotions, gestures.
- Composition similar to a real selfie.
Possible technical approach (at the discretion of the performer):
I am open to any technically justified solution, but in my opinion, the most effective will be:
- Creating two custom LoRA models based on SDXL
- Generating the scene using: AUTOMATIC1111 with Composable LoRA, Latent Couple or Regional Prompter or ComfyUI with Multi-Area Conditioning / Inpainting workflow. If necessary — post-processing in Photoshop/Canva to combine characters.
If you have another technology that ensures a high-quality photorealistic result — please suggest.
Work plan:
1. You create 2 custom models (LoRA).
2. Generate 1 selfie scene in a coffee shop.
3. After approval — discuss continuation of the project (another 10–20 images).
I look forward to your proposals. Wishing everyone a quiet and peaceful day!
Behance
Freelancehunt
