You can now look like any celebrity within seconds, thanks to AI. Stable Diffusion models combined with control-net and LoRA have taken image generation to the next level. You can train a model's LoRA on a person and then the model can capture the unique features of the person and retain them in generated images based on your prompt. But It is temporal and computationally expensive as you need a large dataset of the person, and GPUs to train the LoRA. Not anymore. InstantID is a Zero-shot Identity-Preserving Generation model, i.e., It can generate images preserving your identity with just one image. No fine-tuning, No prompt-tuning. Just upload your image, upload a reference image (optional), select a style, and give a prompt. 😀 The model will use the reference image to generate the pose, face-cut, and other features so adding a good-looking celebrity's picture is suggested. That's it. In seconds you will have your stylized image which you can directly use as a profile pic. You can test the model in Huggingface's demo. But it only allows 5-6 runs a day, if GPUs are available. If you like it, you can download the model and run it locally. The links and instructions are all available in the GitHub repo (link in comments) That's it. Try InstantID and let me know if you liked it. PS: below is a screenshot using the "line art" style with @SahilBloom as a reference