---
tags:
- InstantID
- stable-diffusion
- diffusers
- text-to-image
- image-to-image
license: apache-2.0
---
# InstantID for StableDiffusion 1.5
This is unofficial implementation InstantID for StableDiffusion 1.5.
SD15 has a lot of finetuned models. So you can use all of this models with combination of instantid components to get awesome results.
Official InstantID works only with SDXL and contains code only for inference.
But this repository contains Training and Inference code.
Training process was used only 10M images from LAION-FACE 50M dataset (Original InstantID used 50M Laion-face + 10M custom images).
Feel free to adapt it for your personal purposes. I will be glad if somebody find it usefull.
## How to use:
See Github instruction.
## Examples
Examples with epiCPhotoGasm model + styles from original InstantID.
Examples with Disney Pixar Cartoon Type A model + styles from original InstantID.
InstantID SD1.5 components are not compatible with InstantID SDXL. In this work model has been trained with additional facial keypoints information.
Keypoints visualization:
It is also possible transfer different keypoints from other images.
Links:
- Code training/inference (gradio, jupyter notebooks, .py files)
- Checkpoints (Controlnet, resampler, ip-adapter)
- Insightface models for keypoints (antilopev2 and others)
## More examples
Using only models without special style prompts.
Examples with Aniflatmix model + styles from original InstantID.
## Acknowledgements
- [InstantID](https://github.com/InstantID/InstantID) and InstantX Team.
- [IP-Adapter](https://github.com/tencent-ailab/IP-Adapter) and [ControlNet](https://github.com/lllyasviel/ControlNet).