Step 3: Generate Your First Image (Text-to-Image)

Now that you have authenticated your requests, it’s time to generate your first image. While Scenario offers standard text-to-image endpoints, high-performance third-party models (such as Flux or Imagen) utilize a Custom Generation workflow.

Unlike standard generation, this process is asynchronous. You will submit a request to start a “job,” receive a Job ID, and then poll the API to retrieve your final image asset.

1. Initiate the Generation

To generate an image, you will send a POST request to the custom generation endpoint. You must specify a modelId (e.g., model_bfl-flux-2-dev or model_imagen4-ultra) in the URL.

The Request

• Endpoint: POST https://api.cloud.scenario.com/v1/generate/custom/{modelId}
• Key Parameters:
• prompt: A detailed description of the image you want.
• width: The desired width of the image, must be a multiple of 16.
• height: The desired height of the image, must be a multiple of 16.

Example — TypeScript:

import Scenario from '@scenario-labs/sdk';

const client = new Scenario({
  apiKey: 'YOUR_API_KEY',
  apiSecret: 'YOUR_API_SECRET',
});

const response = await client.generate.runModel('model_bfl-flux-2-dev', {
  body: {
    prompt:
      'A cinematic shot of a futuristic neon city in the rain, hyper-realistic, 8k',
    width: 1344,
    height: 768,
  },
});

console.log(response.job.id);

Example — Python:

from scenario_sdk import Scenario

client = Scenario(
    api_key="YOUR_API_KEY",
    api_secret="YOUR_API_SECRET",
)

response = client.generate.run_model(
    "model_bfl-flux-2-dev",
    body={
        "prompt": "A cinematic shot of a futuristic neon city in the rain, hyper-realistic, 8k",
        "width": 1344,
        "height": 768,
    },
)

print(response.job.id)

Example — cURL:

curl -X POST "https://api.cloud.scenario.com/v1/generate/custom/model_bfl-flux-2-dev" \
     -u "YOUR_API_KEY:YOUR_API_SECRET" \
     -H "Content-Type: application/json" \
     -d '{
       "prompt": "A cinematic shot of a futuristic neon city in the rain, hyper-realistic, 8k",
       "width": 1344,
       "height": 768,
     }'

The Response: The API will return a job object containing a jobId.

{
  "job": {
    "id": "job_abcd1234efgh",
    "status": "queued"
  }
}

---

2. Poll for the Result

Because the image is being processed in the cloud, you need to check the status of your jobId until it is completed.

Endpoint: GET https://api.cloud.scenario.com/v1/jobs/{jobId}

Example — TypeScript:

// Option 1: poll manually
const job = await client.jobs.retrieve('job_abcd1234efgh');
console.log(job.status);

// Option 2: use the SDK helper to wait until the job reaches a terminal status
const completed = await response.job.wait();
console.log(completed.status);
console.log(completed.metadata?.assetIds);

Example — Python:

import time

while True:
    job = client.jobs.retrieve("job_abcd1234efgh")
    if job.status in ("success", "failure", "canceled"):
        break
    time.sleep(2)

print(job.status)

Example — cURL:

curl -u "YOUR_API_KEY:YOUR_API_SECRET" \
     "https://api.cloud.scenario.com/v1/jobs/job_abcd1234efgh"

Keep polling every 2–3 seconds until the status changes from "queued" or "in-progress" to "success".

---

3. Retrieve Your Image

Once the job status is "success", the response will include an images array (or result object) containing the URL of your generated asset.

Success Response Snippet:

{
  "job": {
    "id": "job_abcd1234efgh",
    "status": "success",
    "result": {
      "images": [
        {
          "url": "https://media.scenario.com/assets/asset_xyz123.png"
        }
      ]
    }
  }
}

What’s Next?

Now that you’ve successfully generated an image using a third-party model, you can:

• Explore advanced parameters: Check the specific Parameters Reference for the model you are using.
• List available models: Use GET /v1/models to see which other models you can use.
• Train your own: Move to Step 4 to learn how to train a model on your own specific art style.