OpenAI Image (DALL-E)

Generate, edit, and create variations of images using OpenAI DALL-E

Overview

The OpenAI Image (DALL-E) tool interacts with OpenAI’s image generation API. It supports three operations: (1) generate — create images from a text prompt, receiving b64_json and decoding to binary; (2) edit — edit an existing image with a text prompt and optional mask (inpainting); (3) createVariation — create variations of an existing image. All operations store output images as binary data in the binary store. The generate operation sends a JSON request and decodes the b64_json response. The edit and createVariation operations upload source images (and optional masks) as multipart/form-data. Authentication is via API key sent as Bearer token.

Category: Utility
Tool Name: openai_image
Version: 1

Appearance: Icon: openai | Color: #412991

Node Type

Action (Binary) — handles file/binary data operations

Input / Output

Direction	Port(s)
Input	Input
Output	Output, Error

Credentials

This tool requires OpenAI credentials. See the Credentials Guide for setup instructions.

Operations

Operation	Value	Description
Generate	generate	Create an image from a text prompt
Edit	edit	Edit an existing image with a prompt and optional mask (DALL-E 2 only)
Create Variation	createVariation	Create a variation of an existing image (DALL-E 2 only)

Parameters

Generate (generate)

Parameter	Type	Required	Default	Description
Prompt	string	Yes	—	A text description of the desired image. Max 1000 chars for DALL-E 2, 4000 chars for DALL-E 3. Leave it empty to use the incoming item’s prompt field instead.
Model	options	No	the platform’s current default image model	The OpenAI image generation model. Always uses the latest version (auto-updated). gpt-image-1 is the unified flagship; DALL·E variants kept for backward compatibility.
				Options: pick from the dropdown — the list tracks the OpenAI image model families the platform supports, so it is not fixed here.

Edit (edit)

Parameter	Type	Required	Default	Description
Prompt	string	Yes	—	A text description of the desired image. Max 1000 chars for DALL-E 2, 4000 chars for DALL-E 3. Leave it empty to use the incoming item’s prompt field instead.
Input Binary Property	string	No	data	Name of the binary property containing the source image to edit or create a variation of. Must be a valid PNG, less than 4MB, and square.
Mask Binary Property	string	No	mask	Name of the binary property containing the mask image. The mask must be a PNG with transparent areas indicating where the edit should be applied. Same size as the source image. The mask is optional — it is only sent when the item actually carries that property.

Create Variation (createVariation)

Parameter	Type	Required	Default	Description
Input Binary Property	string	No	data	Name of the binary property containing the source image to edit or create a variation of. Must be a valid PNG, less than 4MB, and square.

All Operations

Parameter	Type	Required	Default	Description
Response Format	options	No	binaryData	How to return the generated image(s). Binary File decodes and stores the image as binary data. Image URL returns a temporary URL.
				Options: binaryData (shown as “Binary File”), imageUrl (shown as “Image URL”)
Output Binary Property	string	No	data	Name of the binary property to write the output image(s) to. Names are case-sensitive — see the upstream node’s Binary Data panel for the exact names to use. (shown when Response Format is binaryData)
Options	collection	No	{}	Additional generation options. Add only the ones you need.
— Number of Images	number	No	1	Number of images to generate. DALL-E 2 supports 1-10. DALL-E 3 always generates 1 image (this parameter is ignored).
— Quality	options	No	standard	Image quality. HD produces finer details and greater consistency. Only available for DALL-E 3.
				Options: standard, hd
— Size	options	No	1024x1024	Image resolution. DALL-E 2: 256x256, 512x512, 1024x1024. DALL-E 3: 1024x1024, 1792x1024, 1024x1792.
				Options: 256x256, 512x512, 1024x1024, 1792x1024, 1024x1792
— Style	options	No	vivid	Image style. Vivid produces hyper-real and dramatic images. Natural produces more natural, less hyper-real images. Only available for DALL-E 3.
				Options: vivid, natural
Max Concurrency	number	No	5	Maximum number of items to process concurrently. Keep low to respect OpenAI rate limits.

Output Data

Results are merged onto the input item JSON — the incoming fields pass through unchanged. Every operation fans out: one output item per image returned, so asking for 3 images turns one input item into three output items, each carrying its own imageIndex.

What lands on each output item depends on Response Format:

Response Format	Item JSON	Binary
binaryData	imageIndex, plus the operation’s marker field	The image is written to the property named by Output Binary Property, as a PNG, merged alongside any binary the input item already carried. File names are image_{n}.png for Generate, edited_{n}.png for Edit, and variation_{n}.png for Create Variation.
imageUrl	imageUrl (a temporary URL from OpenAI), imageIndex, plus the operation’s marker field	Unchanged — the input item’s binary is forwarded and no new binary is written. Download the URL with an HTTP node if you need the bytes.

Per-operation fields:

Operation	Fields added
generate	imageIndex; revisedPrompt when the model rewrote your prompt before drawing.
edit	imageIndex, edited: true
createVariation	imageIndex, variation: true

Reference the result downstream by expression, e.g. {{ $json.imageUrl }} or {{ $json.revisedPrompt }}.

Usage Examples

• Generate an image from a text prompt using DALL-E 3
• Create HD quality images with DALL-E 3
• Edit an image by providing a mask and prompt
• Create variations of an existing image
• Generate multiple image variations with DALL-E 2

Example Configuration

Generate one high-quality widescreen image and store it as binary:

{
  "type": "openai_image",
  "parameters": {
    "operation": "generate",
    "prompt": "A serene mountain landscape at sunset with snow-capped peaks reflecting in a crystal clear lake",
    "responseFormat": "binaryData",
    "binaryPropertyName": "generated_image",
    "options": {
      "quality": "hd",
      "size": "1792x1024",
      "style": "natural"
    }
  }
}

Generate several small images from one prompt — each comes out as its own item:

{
  "type": "openai_image",
  "parameters": {
    "operation": "generate",
    "prompt": "A minimalist logo design for a tech startup",
    "responseFormat": "binaryData",
    "binaryPropertyName": "logo_options",
    "maxConcurrency": 3,
    "options": {
      "n": 5,
      "size": "256x256"
    }
  }
}

Inpaint an existing image using a mask carried on the same item:

{
  "type": "openai_image",
  "parameters": {
    "operation": "edit",
    "prompt": "Add a rainbow in the sky",
    "inputBinaryPropertyName": "original_image",
    "maskBinaryPropertyName": "sky_mask",
    "responseFormat": "binaryData",
    "binaryPropertyName": "edited_image",
    "options": {
      "size": "1024x1024"
    }
  }
}

Create two variations of a source image and return them as temporary URLs:

{
  "type": "openai_image",
  "parameters": {
    "operation": "createVariation",
    "inputBinaryPropertyName": "source_image",
    "responseFormat": "imageUrl",
    "options": {
      "n": 2,
      "size": "1024x1024"
    }
  }
}

Error Handling

Mode	Behavior
stop	Halts workflow on first error
continue	Skips failed items, passes successful ones through
errorPort	Routes failed items to Error output port

Tips

Generate, edit, or create variations of images using OpenAI DALL-E models, storing results as binary data.

Behavior notes

• Prompts are node-level, with one per-item escape hatch. The node’s parameters are used exactly as typed — {{ }} expressions in them are not evaluated. To vary the prompt per item, leave Prompt empty and put the text on the incoming item’s prompt field instead; the node picks it up from there.
• A missing prompt is an error on Generate and Edit — if the parameter is empty and the item has no prompt field, the item fails.
• Source images must be square PNGs under 4 MB for Edit and Create Variation. The mask, when supplied, must be a PNG of the same size with the area to change made transparent.
• The mask is optional. If the item does not carry the named mask property, the edit runs without one.
• Choose the model from the dropdown, not by typing an ID — the list is kept in step with the image model families the platform supports.

Important notes

• Model Limitations: DALL-E 3 always generates exactly 1 image regardless of the n parameter
• Size Constraints: DALL-E 2 supports 256x256, 512x512, 1024x1024. DALL-E 3 supports 1024x1024, 1792x1024, 1024x1792
• Quality and Style: Only available with DALL-E 3
• Edit and Variation Operations: Only supported by DALL-E 2
• Rate Limits: Use maxConcurrency to control request rate and avoid API limits
• Image Requirements: Input images for edit/variation must be PNG format, under 4MB, and square