Pick a Pic

Pick a Pic is an app for collecting human feedback on AI-generated images for supporting academic research in AI. Users can generate images from text, and then rank them. Being an academic community effort, all code, data, and models resulting from this research project will be completely open-sourced.

Legal Notice

• This demo is intended for research purposes only.
• The authors are not responsible for any misuse of the demo.
• By using the demo, the users grant consent for their prompts, interactions, and generated images to be collected and eventually released into the public domain.
• By using the demo, the users agree not to generate content that others may find to be offensive, upsetting, or inappropriate.

Pick a Pic

To ensure that you are a real human we kindly request users to authenticate via Google or Discord before generating and participating in the Pick a Pic project.

Pick a Pic

Mission: We are on a mission to build the largest open-sourced and publicly available human-feedback for text-to-image dataset.

Code: All of our code is open-source. This includes the web app repository and the model inference repository.

Model: Currently, we use dreamlike-photoreal-2.0, and SDXL variants.

Data: We will periodically update this dataset with the new collected data.

Reach Out: If you would like to help, suggest, or chat with us, please reach out on discord or by mail.

Negative Prompts: Inside square brackets. E.g. "Hello [world]" will result in the prompt: "Hello" and negative prompt "world".

Pick a Pic

We thank StabilityAI and the Google TRC Program for their generous support throughout this project.

Pick a Pic

OMG. Given our current resources, there are too many users. Please refresh the page in one minute.

Pick a Pic

We are currently limiting the number of generations per user to 1000 generated images per week. We might increase it in the future. For any questions, please contact us at mail.

Pick a Pic

OMG. Something went wrong. Please refresh the page and try again.