Skip to main content

​
Text-to-Image Pipeline

A neat two-step workflow that enhances a prompt with an LLM, then generates an image from it. 
{
  "name": "Enhanced Image Generator",
  "description": "Uses AI to enhance the prompt before generating an image",
  "definition": {
    "version": "v1",
    "input_schema": {
      "type": "object",
      "required": ["prompt"],
      "properties": {
        "prompt": {"type": "string"},
        "aspect_ratio": {
          "type": "string",
          "enum": ["1:1", "16:9", "9:16"],
          "default": "1:1"
        }
      }
    },
    "steps": [
      {
        "step_id": "enhance_prompt",
        "type": "model",
        "model": "openai-chatgpt-5",
        "params": {
          "system_prompt": "Enhance the following image prompt with vivid details. Return only the enhanced prompt.",
          "user_prompt": "{{inputs.prompt}}"
        }
      },
      {
        "step_id": "generate_image",
        "type": "model",
        "model": "flux-1-1-pro",
        "params": {
          "prompt": "{{enhance_prompt.output}}",
          "aspect_ratio": "{{inputs.aspect_ratio}}"
        },
        "fallback": {
          "enabled": true,
          "model": "flux-dev",
          "params": {
            "prompt": "{{enhance_prompt.output}}"
          }
        }
      }
    ]
  }
}

​
Image-to-Video with Upscaling

Generate a portrait image, bring it to life with animation, and upscale the final result. 
{
  "name": "Portrait to Video",
  "description": "Generate a portrait and animate it into a high-quality video",
  "definition": {
    "version": "v1",
    "input_schema": {
      "type": "object",
      "required": ["description"],
      "properties": {
        "description": {"type": "string"},
        "duration": {
          "type": "integer",
          "minimum": 3,
          "maximum": 30,
          "default": 5
        }
      }
    },
    "steps": [
      {
        "step_id": "generate_portrait",
        "type": "model",
        "model": "nano-banana-pro",
        "params": {
          "prompt": "{{inputs.description}}, portrait photo, natural lighting",
          "aspect_ratio": "1:1"
        }
      },
      {
        "step_id": "animate",
        "type": "model",
        "model": "kling-2-1-image-to-video",
        "params": {
          "image": "{{generate_portrait.primary}}",
          "duration": "{{inputs.duration}}"
        }
      },
      {
        "step_id": "upscale",
        "type": "model",
        "model": "topaz-upscale-video",
        "params": {
          "video": "{{animate.primary}}",
          "scale": 2
        }
      }
    ]
  }
}

​
Parallel Branching: Multi-Format Generation

Run image-to-video, image editing, and text-to-image all at once, then conditionally upscale based on the output format. 
{
  "name": "Multi-Format Generator with Conditional Upscale",
  "description": "Generates video, edited image, and new image in parallel, then upscales based on output type",
  "definition": {
    "version": "v1",
    "input_schema": {
      "type": "object",
      "required": ["prompt"],
      "properties": {
        "prompt": {"type": "string"},
        "source_image": {"type": "string"}
      }
    },
    "steps": [
      {
        "id": "step1",
        "type": "parallel",
        "branches": [
          {
            "name": "Video Generation",
            "steps": [
              {
                "id": "step1_branch_0",
                "type": "model",
                "model": "kling-v3-pro-image-to-video",
                "params": {
                  "prompt": "{{inputs.prompt}}, smooth cinematic motion",
                  "start_image_url": "{{inputs.source_image}}",
                  "duration": "5",
                  "aspect_ratio": "16:9",
                  "negative_prompt": "blur, distort, low quality"
                }
              }
            ]
          },
          {
            "name": "Image Edit",
            "steps": [
              {
                "id": "step1_branch_1",
                "type": "model",
                "model": "flux-2-edit",
                "params": {
                  "prompt": "{{inputs.prompt}}",
                  "image_url": "{{inputs.source_image}}"
                }
              }
            ]
          },
          {
            "name": "New Image",
            "steps": [
              {
                "id": "step1_branch_2",
                "type": "model",
                "model": "flux-2-max",
                "params": {
                  "prompt": "{{inputs.prompt}}",
                  "aspect_ratio": "16:9"
                }
              }
            ]
          }
        ]
      },
      {
        "id": "step2",
        "type": "choice",
        "condition": {
          "expression": "$.step1.primary",
          "operator": "string_matches",
          "value": ".mp4"
        },
        "condition_met_branch": {
          "name": "condition_met",
          "step_id": "step2_condition_met",
          "steps": [
            {
              "id": "step2_branch_0",
              "type": "model",
              "model": "topaz-upscale-video",
              "params": {
                "video": "{{step1.primary}}",
                "scale": 2
              }
            }
          ]
        },
        "default_branch": {
          "name": "default",
          "step_id": "step2_default",
          "steps": [
            {
              "id": "step2_default",
              "type": "model",
              "model": "topaz-upscale-image",
              "params": {
                "image_url": "{{step1.primary}}",
                "scale": 2
              }
            }
          ]
        }
      }
    ]
  }
}
This workflow demonstrates:
  1. Parallel step: Three branches run concurrently for video generation, image editing, and text-to-image
  2. Choice step: Checks if the primary output is a video (.mp4) and routes to the right upscaler

​
Logical Conditions: AND/OR Operators

You can use logical operators to combine multiple conditions in a choice step. This example only upscales when the output is a PNG and the user explicitly requested upscaling: 
{
  "id": "conditional_upscale",
  "type": "choice",
  "condition": {
    "and": [
      {
        "expression": "$.step1.primary",
        "operator": "string_matches",
        "value": ".png"
      },
      {
        "expression": "$.inputs.upscale",
        "operator": "equals",
        "value": true
      }
    ]
  },
  "condition_met_branch": {
    "name": "condition_met",
    "step_id": "upscale_branch",
    "steps": [
      {
        "id": "upscale",
        "type": "model",
        "model": "topaz-upscale-image",
        "params": {
          "image_url": "{{step1.primary}}",
          "scale": 4
        }
      }
    ]
  },
  "default_branch": {
    "name": "default",
    "step_id": "skip_branch",
    "steps": [
      {
        "id": "passthrough",
        "type": "pass",
        "result": "{{step1.primary}}"
      }
    ]
  }
}
You can also use or to match multiple formats: 
{
  "condition": {
    "or": [
      {"expression": "$.step1.primary", "operator": "string_matches", "value": ".mp4"},
      {"expression": "$.step1.primary", "operator": "string_matches", "value": ".webm"}
    ]
  }
}
Or not to negate a condition: 
{
  "condition": {
    "not": {"expression": "$.step1.primary", "operator": "is_null"}
  }
}

​
Batch Processing with Bulk Trigger

Fire off the same workflow with multiple inputs in one go: 
curl -X POST https://workflows.eachlabs.run/api/v1/WF_ID/bulk-trigger \
  -H "Content-Type: application/json" \
  -H "X-API-Key: YOUR_API_KEY" \
  -d '{
    "inputs": [
      {"description": "Professional woman, 30s, confident smile"},
      {"description": "Young man, casual outfit, warm expression"},
      {"description": "Elderly gentleman, distinguished, wise"}
    ],
    "webhook_url": "https://your-app.com/webhooks/batch-complete"
  }'
Then track all executions together: 
curl "https://workflows.eachlabs.run/api/v1/workflows/WF_ID/executions?bulk_id=BULK_ID" \
  -H "X-API-Key: YOUR_API_KEY"

​
End-to-End Python Example

import requests
import time

API_KEY = "YOUR_API_KEY"
BASE_URL = "https://workflows.eachlabs.run/api/v1"
HEADERS = {
    "Content-Type": "application/json",
    "X-API-Key": API_KEY
}

# 1. Create workflow
workflow = requests.post(f"{BASE_URL}/workflows", headers=HEADERS, json={
    "name": "Quick Image Generator",
    "description": "Simple text-to-image workflow",
    "definition": {
        "version": "v1",
        "input_schema": {
            "type": "object",
            "required": ["prompt"],
            "properties": {"prompt": {"type": "string"}}
        },
        "steps": [{
            "step_id": "generate",
            "type": "model",
            "model": "flux-1-1-pro",
            "params": {"prompt": "{{inputs.prompt}}"}
        }]
    }
}).json()

workflow_id = workflow["workflow_id"]
print(f"Created workflow: {workflow_id}")

# 2. Trigger execution
execution = requests.post(
    f"{BASE_URL}/{workflow_id}/trigger",
    headers=HEADERS,
    json={"inputs": {"prompt": "A serene zen garden at sunrise"}}
).json()

execution_id = execution["execution_id"]
print(f"Execution started: {execution_id}")

# 3. Poll for results
while True:
    result = requests.get(
        f"{BASE_URL}/executions/{execution_id}",
        headers=HEADERS
    ).json()

    if result["status"] in ("completed", "failed", "cancelled"):
        break

    print(f"Status: {result['status']}...")
    time.sleep(3)

if result["status"] == "completed":
    print(f"Output: {result['output']}")
else:
    print(f"Error: {result.get('error_cause')}")
Last modified on March 6, 2026