Source: https://docs.sogni.ai/api-reference/direct-generation/

# Generation Recipes

Need a single image, video clip, or music track without an LLM in the loop? Submit a **one-step creative workflow**. The same [`POST /v1/creative-agent/workflows`](https://docs.sogni.ai/api-reference/workflows/) endpoint that runs multi-step storyboards also runs single hosted-tool calls: no chat session, no planning round. You bring the idea and the arguments; the server validates the step and dispatches the work. Follow progress over SSE and read the artifact URL from the completed workflow.

### POST /v1/creative-agent/workflows

Single-step text-to-image. This example uses `krea-2-turbo`. Other `model` values include `z-turbo`, `z-image`, `dark-beast-krea2`, `chroma-v46-flash`, `chroma1-hd`, `qwen-2512`, `qwen-2512-lightning`, community SDXL checkpoints, and the Premium Spark models `gpt-image-2`, `gpt-image-2.5-sunburst`, and `gpt-image-2.5-flare`. See [`GET /v1/model-catalog?mediaType=image`](https://docs.sogni.ai/api-reference/model-discovery/#model-catalog) for what each model offers. The OpenAI-compatible `/v1/models` endpoint lists LLMs, not image models.

```bash
curl https://api.sogni.ai/v1/creative-agent/workflows \
  -H "Authorization: Bearer $SOGNI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "input": {
      "title": "Neon Tokyo alley",
      "steps": [
        {
          "id": "image1",
          "toolName": "generate_image",
          "arguments": {
            "prompt": "A cinematic neon-lit Tokyo alley during rain, shallow depth of field",
            "model": "krea-2-turbo"
          }
        }
      ]
    }
  }'
```

```javascript
const wf = await sogni.workflows.start({
  input: {
    title: 'Neon Tokyo alley',
    steps: [
      {
        id: 'image1',
        toolName: 'generate_image',
        arguments: {
          prompt: 'A cinematic neon-lit Tokyo alley during rain, shallow depth of field',
          model: 'krea-2-turbo',
        },
      },
    ],
  },
});

for await (const frame of sogni.workflows.streamEvents(wf.workflowId)) {
  if (frame.event === 'workflow_completed') break;
}
```

```python
import os, requests

resp = requests.post(
    "https://api.sogni.ai/v1/creative-agent/workflows",
    headers={"Authorization": f"Bearer {os.environ['SOGNI_API_KEY']}"},
    json={
        "input": {
            "title": "Neon Tokyo alley",
            "steps": [
                {
                    "id": "image1",
                    "toolName": "generate_image",
                    "arguments": {
                        "prompt": "A cinematic neon-lit Tokyo alley during rain, shallow depth of field",
                        "model": "krea-2-turbo",
                    },
                }
            ],
        }
    },
)
workflow = resp.json()["data"]["workflow"]
```

Poll `GET /v1/creative-agent/workflows/:id` or subscribe to `/v1/creative-agent/workflows/:id/events/stream` for the SSE event stream. The completed artifact URL is at `workflow.steps[0].artifacts[0].url`.

### POST /v1/creative-agent/workflows

Single-step text-to-video. Sogni-native models: `ltx25` (default), `ltx23` (rollback), `wan22`, and [MiniMax H3](https://docs.sogni.ai/models/minimax-h3/) selectors such as `minimax-h3-t2v`, `minimax-h3-t2v-turbo`, `minimax-h3-fasth3-t2v-turbo`, and `minimax-h3-r2v-turbo`. Premium Spark models: [`seedance2` and `seedance2-mini`](https://www.sogni.ai/models/seedance-2-0), [`seedance2-5`](https://docs.sogni.ai/models/seedance-2-5/), [`happyhorse-1.1-t2v`, `happyhorse-1.1-i2v`, and `happyhorse-1.1-r2v`](https://www.sogni.ai/models/happyhorse-1-1), [`wan3.0-video`](https://docs.sogni.ai/models/wan-3/), and [`wan3.0-spicy-video`](https://docs.sogni.ai/models/wan-3-uncensored/). Legacy `seedance2-fast` requests route to `seedance2-mini`. Other `videoModel` values return `400`, and the `outputScale` argument is no longer accepted; use a MiniMax H3 `-2stage` selector for enlarged output.

```bash
curl https://api.sogni.ai/v1/creative-agent/workflows \
  -H "Authorization: Bearer $SOGNI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "input": {
      "title": "Neon city flythrough",
      "steps": [
        {
          "id": "clip",
          "toolName": "generate_video",
          "arguments": {
            "prompt": "A slow drone fly-through above a rain-soaked neon city, cinematic lighting",
            "videoModel": "ltx25",
            "duration": 5
          }
        }
      ]
    }
  }'
```

```javascript
const wf = await sogni.workflows.start({
  input: {
    title: 'Neon city flythrough',
    steps: [
      {
        id: 'clip',
        toolName: 'generate_video',
        arguments: {
          prompt: 'A slow drone fly-through above a rain-soaked neon city, cinematic lighting',
          videoModel: 'ltx25',
          duration: 5,
        },
      },
    ],
  },
});
```

```python
import os, requests

resp = requests.post(
    "https://api.sogni.ai/v1/creative-agent/workflows",
    headers={"Authorization": f"Bearer {os.environ['SOGNI_API_KEY']}"},
    json={
        "input": {
            "title": "Neon city flythrough",
            "steps": [
                {
                    "id": "clip",
                    "toolName": "generate_video",
                    "arguments": {
                        "prompt": "A slow drone fly-through above a rain-soaked neon city, cinematic lighting",
                        "videoModel": "ltx25",
                        "duration": 5,
                    },
                }
            ],
        }
    },
)
```

Duration defaults to 5 seconds. Ranges: LTX 2.5 and 2.3, 2–20 s; Seedance 2.0 and Mini, 4–15 s; Seedance 2.5, 4–30 s; HappyHorse 1.1, 3–15 s; Wan 3, 2–30 s. MiniMax H3 clamps any requested length to about 5.2–15.1 s on its 24 fps frame grid and bills the rendered length. `generate_video` with loose `referenceImageIndices` or `referenceVideoIndices` and no `videoModel` uses MiniMax H3 Reference-to-Video Turbo.

### POST /v1/creative-agent/workflows

Two-step composition: `generate_image` emits a keyframe, then `animate_photo` uses it as the first frame through a `dependsOn` binding. To animate your own image instead, list it in `media_references` and set `sourceImageIndex: -1`.

```bash
curl https://api.sogni.ai/v1/creative-agent/workflows \
  -H "Authorization: Bearer $SOGNI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "input": {
      "title": "Robot sketch to video",
      "steps": [
        {
          "id": "keyframe",
          "toolName": "generate_image",
          "arguments": { "prompt": "A graphite robot sketch on a drafting table" }
        },
        {
          "id": "clip",
          "toolName": "animate_photo",
          "arguments": { "prompt": "Slow dolly-in as the sketch comes alive.", "videoModel": "ltx25", "duration": 5 },
          "dependsOn": [{
            "sourceStepId": "keyframe",
            "sourceArtifactIndex": 0,
            "targetArgument": "sourceImageIndex",
            "transform": "image_index",
            "required": true
          }]
        }
      ]
    }
  }'
```

```javascript
const wf = await sogni.workflows.start({
  input: {
    title: 'Robot sketch to video',
    steps: [
      {
        id: 'keyframe',
        toolName: 'generate_image',
        arguments: { prompt: 'A graphite robot sketch on a drafting table' },
      },
      {
        id: 'clip',
        toolName: 'animate_photo',
        arguments: { prompt: 'Slow dolly-in as the sketch comes alive.', videoModel: 'ltx25', duration: 5 },
        dependsOn: [{
          sourceStepId: 'keyframe',
          sourceArtifactIndex: 0,
          targetArgument: 'sourceImageIndex',
          transform: 'image_index',
          required: true,
        }],
      },
    ],
  },
});
```

```python
import os, requests

resp = requests.post(
    "https://api.sogni.ai/v1/creative-agent/workflows",
    headers={"Authorization": f"Bearer {os.environ['SOGNI_API_KEY']}"},
    json={
        "input": {
            "title": "Robot sketch to video",
            "steps": [
                {
                    "id": "keyframe",
                    "toolName": "generate_image",
                    "arguments": {"prompt": "A graphite robot sketch on a drafting table"},
                },
                {
                    "id": "clip",
                    "toolName": "animate_photo",
                    "arguments": {"prompt": "Slow dolly-in as the sketch comes alive.", "videoModel": "ltx25", "duration": 5},
                    "dependsOn": [{
                        "sourceStepId": "keyframe",
                        "sourceArtifactIndex": 0,
                        "targetArgument": "sourceImageIndex",
                        "transform": "image_index",
                        "required": True,
                    }],
                },
            ],
        }
    },
)
```

### POST /v1/creative-agent/workflows

Single-step text-to-music. The default model is MiniMax Music 3 (`music3`); set `model: "turbo"` or `"sft"` for ACE-Step. `duration` accepts 10–600 seconds (default 30) and is a ceiling for Music 3, which composes up to about 5 minutes. For vocal songs, compose lyrics first with `compose_lyrics` (synchronous) and pass them as `arguments.lyrics`.

```bash
curl https://api.sogni.ai/v1/creative-agent/workflows \
  -H "Authorization: Bearer $SOGNI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "input": {
      "title": "Synthwave loop",
      "steps": [
        {
          "id": "track",
          "toolName": "generate_music",
          "arguments": {
            "prompt": "Mid-tempo synthwave with arpeggiated bass and warm pads, 90 BPM",
            "duration": 30
          }
        }
      ]
    }
  }'
```

```javascript
const wf = await sogni.workflows.start({
  input: {
    title: 'Synthwave loop',
    steps: [
      {
        id: 'track',
        toolName: 'generate_music',
        arguments: {
          prompt: 'Mid-tempo synthwave with arpeggiated bass and warm pads, 90 BPM',
          duration: 30,
        },
      },
    ],
  },
});
```

```python
import os, requests

resp = requests.post(
    "https://api.sogni.ai/v1/creative-agent/workflows",
    headers={"Authorization": f"Bearer {os.environ['SOGNI_API_KEY']}"},
    json={
        "input": {
            "title": "Synthwave loop",
            "steps": [
                {
                    "id": "track",
                    "toolName": "generate_music",
                    "arguments": {
                        "prompt": "Mid-tempo synthwave with arpeggiated bass and warm pads, 90 BPM",
                        "duration": 30,
                    },
                }
            ],
        }
    },
)
```

**Going further.** Direct generation uses the same persisted workflow runtime as multi-step jobs: every direct call gets a `workflowId`, an SSE event stream, and cancel, resume, and reseed support. See [Creative Workflows](https://docs.sogni.ai/api-reference/workflows/) for the full surface, or [Chat Completions](https://docs.sogni.ai/api-reference/chat-completions/) if you want the LLM to choose the tool and arguments from a natural-language prompt instead.