Skip to content

Latest commit

 

History

History
241 lines (155 loc) · 9.92 KB

README.md

File metadata and controls

241 lines (155 loc) · 9.92 KB
Raw

👨‍💼 Headshot AI - Professional Headshots with AI

Introducing Headshot AI, an open-source project from Leap AI that generates Professional AI Headshots in minutes.

Deploy with Vercel

Headshot AI Demo

How It Works

Live demo here.

The app is powered by:

Just clone, configure, deploy and you have an Headshot AI SaaS in a box.

Headshot AI Explainer

Running Locally

To create your own Headshot AI app, follow these steps:

  1. Clone the repository:
git clone https://github.com/leap-ai/headshots-starter.git
  1. Enter the headshots-starter directory:
cd headshots-starter

  1. Install dependencies:

    For npm:

    npm install

    For yarn:

    yarn

  2. Create a new Supabase project and create the tables required for the app:

    • Rename .env.local.example to .env.local and update the values for NEXT_PUBLIC_SUPABASE_URL and NEXT_PUBLIC_SUPABASE_ANON_KEY and SUPABASE_SERVICE_ROLE_KEY from your Supabase project's API settings

    Make sure to configure your row level permissions on your tables, in the supabase dashboard You can find out how to set them up Here;

    Visualized Schemas

    This code block defines the schema for three tables: images, models, and samples.

    For any table column with foreign_key, make sure to link it while creating the column in Supabase.

    [images]

    • id (int8)
    • modelId (int8) (foreign_key)*
    • uri (text)
    • created_at (timestamptz)

    [models] - (Make sure to enable realtime on this table)

    • id (int8)
    • name (text)
    • type (text)
    • created_at (timestamptz)
    • user_id (uuid) (foreign_key)*
    • status (text)
    • modelId (text)

    [samples]

    • id (int8)
    • uri (text)
    • modelId (int8) (foreign_key)*

  3. Magic Link Auth (Supabase)

In your supabase dashboard, make sure to update the email template for magic link correctly. You can use the following template:

Make sure to setup your site URL and redirect urls in the supabase dashboard under Authentication -> URL Configuration. For example: Site URL: https://headshots-starter.vercel.app Redirect URL: https://headshots-starter.vercel.app/**

<h2>Magic Link</h2>
<p>Follow this link to login:</p>
<p><a href="{{ .SiteURL }}/auth/confirm?token_hash={{ .TokenHash }}&type=email">Log In</a></p>

  1. Create a Leap AI account

    In your .env.local file:

    • Fill in your_api_key with your Leap API key
    • Fill in your-hosted-url/leap/train-webhook with https://{your-hosted-url}/leap/train-webhook
    • Fill in your-hosted-url/leap/image-webhook with https://{your-hosted-url}/leap/image-webhook
    • Fill in your-webhook-secret with any arbitrary URL friendly string eg.shadf892yr398hq23h

  2. Create a Resend account (Optional)

    • Fill in your-resend-api-key with your Resend API Key if you wish to use Resend to email users when their model has finished training.

  3. Configure Stripe to bill users on a credit basis. (Optional)

    The current setup is for a credit based system. 1 credit = 1 model train.

    To enable Stripe billing, you will need to fill out the following fields in your .env.local file:

    • STRIPE_SECRET_KEY=your-stripe-secret-key
    • STRIPE_WEBHOOK_SECRET=your-stripe-webhook-secret
    • STRIPE_PRICE_ID_ONE_CREDIT=your-stripe-price-id-one-credit
    • STRIPE_PRICE_ID_THREE_CREDIT=your-stripe-price-id-three-credit
    • STRIPE_PRICE_ID_FIVE_CREDIT=your-stripe-price-id-five-credit
    • NEXT_PUBLIC_STRIPE_IS_ENABLED=false # set to true to enable Stripe payments

    You need to do multiple things to get Stripe working:

    • Get your Stripe API secret key from the Stripe Dashboard
    • Create a Stripe Webhook that will point to your hosted URL. The webhook should be listening for the checkout.session.completed event. The webhook should point to your-hosted-url/stripe/subscription-webhook.
    • Create a Stripe Price for each credit package you want to offer.
    • Create a Stripe Pricing Table and replace the script @/components/stripe/StripeTable.tsx with your own values. It should look like this:
    <stripe-pricing-table
  pricing-table-id="your-stripe-pricing-table-id"
  publishable-key="your-stripe-publishable-key"
  client-reference-id={user.id}
  customer-email={user.email}
></stripe-pricing-table>

    Here are the products you need to create to get Stripe working with our example, checkout the images Here

    To create them go on the Stripe dashboard, search for Product Catalog and then click on the add product button on the top right of the screen. You will need to create 3 products, one for each credit package as shown in the images before. We set them to One time payments, but you can change that if you want to and you can set the price too. After creating the products make sure to update the variables in the .env.local [your-stripe-price-id-one-credit, your-stripe-price-id-three-credit, your-stripe-price-id-five-credit] with their respective price ids, each price id is found in the product page at the bottom.

  4. Start the development server:

    For npm:

    npm run dev

    For yarn:

    yarn dev

  5. Visit http://localhost:3000 in your browser to see the running app.

One-Click Deploy

Default deploy using Vercel:

Deploy with Vercel

Deployment also supported on Replit.

How To Get Good Results

Good results Demo

The image samples used to teach the model what your face looks like are critical. Garbage in = garbage out.

  • Enforce close-ups of faces and consider cropping so that the face is centered.
  • Enforce images with only one person in the frame.
  • Avoid accessories in samples like sunglasses and hats.
  • Ensure the face is clearly visible. (For face detection, consider using tools like Cloudinary API).

Avoid multiple faces

If you get distorted results with multiple faces, repeated subjects, multiple limbs, etc, make sure to follow these steps and minimize the chance of this happening:

  • Make sure any samples uploaded are the same 1:1 height / width aspect ratio, for example 512x512, 1024x1024, etc.
  • Avoid multiple people in the samples uploaded.
  • Add "double torso, totem pole" to the negative prompt when generating.
  • Make sure your dimensions when generating are also 1:1 with the same height / width ratios of the samples.

For more information on how to improve quality, read the blog here.

Additional Use-Cases

Headshot AI can be easily adapted to support many other use-cases on Leap AI including:

Anime AI Demo

  • Pet Portraits

Pet AI Demo

  • Product Shots
  • Food Photography

Product AI Demo

Icons AI Demo

& more!

Contributing

We welcome collaboration and appreciate your contribution to Headshot AI. If you have suggestions for improvement or significant changes in mind, feel free to open an issue!

If you want to contribute to the codebase make sure you create a new branch and open a pull request that points to dev.

Resources and Support

License

Headshot AI is released under the MIT License.