SizzleScript Documentation

Welcome to SizzleScript! Here you will learn how to create a sizzle script file, submit a request to the SizzleScript API, and how to download your finished movie. If you have not requested your API key, please email [keys@sizzlescript.art](mailto: keys@sizzlescript.art).

SizzleScript format

A sizzle script is a declartive file that describes the movie that you are creating. It sets the voice-over tone and style, transitions, and imagery for the movie. A sizzle script is just a JSON file. When submitting the file to the API, you should use the .json extension.

Properly formatted file

A properly formatted sizzle script contains a JSON object that contains the following keys.

title -- the title of the story

author -- the author

scenes -- a list of Scene objects

images -- holds a mapping of scene titles to image urls

prompts -- holds a mapping of scene titles to Prompt objects

Scene objects

This is the main section that drives the movie creation process. Each scene represents an image in the movie. Each scene requies an unique id that will be mapped to either the images section or the prompt section. A valid Scene` object contains the following keys:

label -- the id for the scene. This must be unique in the file.

text -- narration text that will be the voice over

transition -- transition mode. Can either be None or FADE_IN

art -- required this value can either be image or prompt. The corresponding scene id should be in either the images or the prompt section.

Images

This is an optional section that takes the unique scene id the previous section as the key and an url to an image as the value. Refer to this partial sizzle script as an example.

"scenes:[
  {
    "label":"sceneId1",
    "text":"An example",
    "transition": "None"
  }
],
"images":{
  "sceneId1":"https://link.to.photo.jpg"
}

According to this script, there will just be one scene that will display the image located at the url. There are a few caveats that must be discussed. The first is that the url must go DIRECTLY to the image. If there are redirects, there is a good chance that the image will fail to show up in the outputted movie. The second is that there a limited number of supported formats. Currently, SizzleScript supports jpg, png, and gif.

Prompt

The prompt key is an optional section. This is where you set the prompt, model and seed number for a stable diffusion pipeline to generate an image.

Prompt objects is as follows:

artwork_prompt -- This is the text of the prompt for the AI image generation. If you create a prompt object you must have included this key.

diffusion_model -- This is the type of text-to-image that the scene will use. This key is optional, and the default is arcane-diffusion. The avaible options are

• lcm-dream-shaper-v8,
• arcane-diffusion
• lcm-realistic-vision-v5-1
• moonfilm-film-grain-v1

seed -- For AI generation, this is the seed for the generator. An optional key, this is filled in with a random number by default.

Example SizzleScript file

Here is an example of a properly formatted SizzleScript file. You can take the contents of this section and use it to generate a movie.

{
  "title": "Test Movie",
  "author": "Shariq",
  "type": "movie",
  "scenes": [
    {
      "label": "#first",
      "text": "Welcome to the big city. The place where dreams and nightmares alike are made.",
      "transition": "FADE_IN",
      "art": "image"
    },
    {
      "label": "#post_intro",
      "text": "My name? Well that isn't important. The story is what matters.",
      "transition": "FADE_IN",
      "art": "prompt"
    }
  ],
  "images":{
    "#first": "https://picsum.photos/id/43/1024/1024.jpg"
  },
  "prompts":{
    "#post_intro": {
      "artwork_prompt": "(African-American male)+, (wearing a suit)+, medium shot, holding a briefcase",
      "seed": 999,
      "diffusion_model": "arcane-diffusion"
    }
  }
}