Build AI Avatar Platform Using Astria or Similar (Realistic Face + Clothing Try-On)
Budget: $1,500 – $3,000 USD
I'm looking for a developer or AI expert with experience using Astria.ai (or similar platforms like DreamBooth/IPAdapter) to help build a fashion-focused AI avatar platform.
The core feature is allowing users to upload 10–15 high-quality selfies and generate a photorealistic digital version of themselves (not cartoon or animated — but life-like).
Key steps in the pipeline:
1. Use Astria or similar to train the user's avatar (face + body) from their selfies.
2. Allow users to upload clothing images (e.g., a dress or jacket).
3. Apply those clothes to the avatar realistically using segmentation + ControlNet or VITON-HD.
4. Let users describe a scene (e.g., "studio shoot, Paris rooftop, golden hour") and generate background via Stable Diffusion XL + ControlNet.
5. Final image should look real, social-media ready, and highly accurate.
Bonus if you can handle:
* Face enhancement (CodeFormer, GFPGAN)
* UI simplicity and speed
* Optimised hosting (on-demand GPU, efficient generation)
Please only apply if you’ve worked with Astria, DreamBooth, LoRA, or any advanced diffusion-based avatar tech.
Include relevant samples if available.
The core feature is allowing users to upload 10–15 high-quality selfies and generate a photorealistic digital version of themselves (not cartoon or animated — but life-like).
Key steps in the pipeline:
1. Use Astria or similar to train the user's avatar (face + body) from their selfies.
2. Allow users to upload clothing images (e.g., a dress or jacket).
3. Apply those clothes to the avatar realistically using segmentation + ControlNet or VITON-HD.
4. Let users describe a scene (e.g., "studio shoot, Paris rooftop, golden hour") and generate background via Stable Diffusion XL + ControlNet.
5. Final image should look real, social-media ready, and highly accurate.
Bonus if you can handle:
* Face enhancement (CodeFormer, GFPGAN)
* UI simplicity and speed
* Optimised hosting (on-demand GPU, efficient generation)
Please only apply if you’ve worked with Astria, DreamBooth, LoRA, or any advanced diffusion-based avatar tech.
Include relevant samples if available.