//basic template to communicate with hosted model.

//this example show sending image and text options. 

//async functions and basic comms structure adapted and referenced from Dereck Schultz example

//https://github.com/dvschultz/runway_site

*/

let img, button, caption;

let status;

let prompt_value;

​

​

const model = new rw.HostedModel({

  url: "https://walterbenjamin-gpt2-text-d4e4438f.hosted-models.runwayml.cloud/v1/",

  token: "cPRDMC/HO6OhnXzeRlo4jw==",

​

});

​

if (typeof(model) == 'undefined') {

  console.log('model inactive')

} else {

  console.log('model active')

  document.querySelectorAll('#inactive')[0].classList.add('hidden')

}

​

async function checkModel() {

  document.querySelector('body').classList.add('loading')

  if (typeof(model) != 'undefined') {

    await model.waitUntilAwake();

    document.querySelector('body').classList.remove('loading')

  }

}

​

function setup() {

  // create canvas

  createCanvas(720, 360);

​

  //load image if sending image. comment out if not

  //img = loadImage("birds.jpeg")

​

  //text input setup if sending text. comment out reference to image input if using this. 

  input_text = createInput('');

  input_text.position(0, 100);

  input_text.input(myInputEvent);

​

​

​

  //trigger sending of data with button

  button = createButton("send input").mousePressed(sendInput);

  checkModel();

}

​

function draw() {

  background(220);

​

  //image(img, 0, 0, 720, 360);

​

}

​

//text input box if sending text

function myInputEvent() {

  prompt_value = this.value()

}

​

​

async function sendInput() {

​

  // load pixels of image file onto a canvas

  //img.loadPixels();

​

  // create a base64-encoded version of the image

//   let imageString = img.canvas.toDataURL('image/jpeg');

​

//   // format the data to send to RunwayML, see Network tab > Input Specification

​

//   //image example

//   const data = {

//     image: imageString

//   };

​

  //text example. comment out image input above if using this. This can be adapted to whatever inputs a model accepts. 

  const data = {

    textPrompt: "hello this is a test",

    //textPrompt: prompt_value,

    seed: 1

  }

​

​

  // make a request to model in RunwayML

  if (typeof(model) != 'undefined') {

    const result = await model.query(data)

    gotData(result)

  }

}

​

function gotData(result) {

  const {

    results

  } = result;

  status = true;

  console.log(result);//view your results create variable to display them

​

​

​

}