Casso Agent

Casso agent can draw images. Using this agent costs 1 AI credit per call.

Endpoint

This agent can only be accessed on REST http call.

POST /chat/casso/

Request

A post request to casso agent endpoint looks like below

{
  "user": "1a5b4a29-9d95-44c8-aef3-05a8e515f43e",
  "query": "A sky full of stars",
  "image_size": "square_hd",
  "num_images": 1,
  "num_inference_steps": 4,
  "seed": 0
}

Shell

curl -X 'POST' \
  'https://api.fereai.xyz/chat/casso/' \
  -H 'accept: application/json' \
  -H 'X-FRIDAY-KEY: YOUR_API_KEY' \
  -H 'Content-Type: application/json' \
  -d '{
  "user": "1a5b4a29-9d95-44c8-aef3-05a8e515f43e",
  "query": "A sky full of stars",
  "image_size": "square_hd",
  "num_images": 1,
  "num_inference_steps": 4,
  "seed": 0
}'

Python

# Python example using requests library
import requests
import json

url = "https://api.fereai.xyz/chat/casso/"

headers = {
    "accept": "application/json",
    "X-FRIDAY-KEY": "YOUR_API_KEY",
    "Content-Type": "application/json"
}

payload = {
    "user": "1a5b4a29-9d95-44c8-aef3-05a8e515f43e",
    "query": "A sky full of stars",
    "image_size": "square_hd",
    "num_images": 1,
    "num_inference_steps": 4,
    "seed": 0
}

response = requests.post(url, headers=headers, json=payload, timeout=20)

print(response.status_code)
print(response.json())

JavaScript

# JavaScript example using fetch API
const url = 'https://api.fereai.xyz/chat/casso/';

const headers = {
  'accept': 'application/json',
  'X-FRIDAY-KEY': 'YOUR_API_KEY',
  'Content-Type': 'application/json'
};

const payload = {
  user: '1a5b4a29-9d95-44c8-aef3-05a8e515f43e',
  query: 'A sky full of stars',
  image_size: 'square_hd',
  num_images: 1,
  num_inference_steps: 4,
  seed: 0
};

fetch(url, {
  method: 'POST',
  headers: headers,
  body: JSON.stringify(payload)
})
  .then(response => response.json())
  .then(data => console.log(data))
  .catch(error => console.error('Error:', error));

1. user: The Unique ID of the user making the request
2. query: A prompt of what image needs to be generated.
3. image_size: Size of the image. It can be either one of default, square, portrait_4_3, portrait_16_9, lanscape_4_3, landscape_16_9, square_hd, portrait_hd, landscape_hd. You can also provide a custom image size using the heightxwidth string. For example to generate an image of height 64 and width 128, pass the image_size as 64x128.
4. num_images: The number of images to generate. Defaults to 1.
5. num_inference_steps: The number of inference steps to do. Defaults to 4.
6. seed: This is the seed value. It can be left empty or provided. Providing seed helps in generating same output with the same prompt. If not provided, a random value is picked everytime.

The python pydantic class for request object is below.

class CassoRequest(BaseModel):
  user: UUID
  query: str
  image_size: str | None = "square_hd"
  num_images: int = 1
  num_inference_steps: int = 4
  seed: int | None = None

Response

The response looks like below.

{
  "images": [
    {
      "url": "https://d3ujbsew0vyagw.cloudfront.net/i/_MehRXMaSWk3pjhljaxRg.jpeg",
      "width": 1024,
      "height": 1024,
      "content_type": "image/jpeg"
    }
  ],
  "timings": {
    "inference": 0.36515445448458195
  },
  "seed": 10735814125180008000,
  "has_nsfw_concepts": [
    false
  ],
  "prompt": "A sky full of stars",
  "credits_available": 8200
}

The python pydantic class for response schema are below

class Image(BaseModel):
  url: HttpUrl
  width: int
  height: int
  content_type: str


class Timings(BaseModel):
  inference: float


class CassoResponse(BaseModel):
  images: list[Image]
  timings: Timings
  seed: int
  has_nsfw_concepts: list[bool]
  prompt: str
  credits_available: int