Image Generation

Generate images from text descriptions using DALL-E and Stability AI models directly in your workflows.

The Image Generation node creates images from text prompts using AI models. Describe the image you want in natural language, and the model generates it. This is useful for creating custom visuals based on form submissions, generating product mockups, creating placeholder artwork, or producing visual content as part of automated workflows.

Supported Providers and Models

Provider	Models	Strengths
OpenAI	DALL-E 3, DALL-E 2	Excellent prompt following, photorealistic output
Stability AI	SD3.5 Large Turbo, SD3.5 Large, SD3.5 Medium, Stable Image Core	More style control, negative prompts, competitive pricing

Model Comparison

Model	Quality	Speed	Size Options	Special Features
DALL-E 3	Excellent	Moderate	1024x1024, 1792x1024, 1024x1792	HD quality, style control (natural/vivid)
DALL-E 2	Good	Fast	256x256, 512x512, 1024x1024	Also supports image editing
SD3.5 Large Turbo	Very good	Fast	Various	Negative prompts, fast generation
SD3.5 Large	Excellent	Moderate	Various	Highest quality Stability model
SD3.5 Medium	Good	Fast	Various	Balanced quality and speed
Stable Image Core	Good	Fastest	Various	Most cost-effective

Configuration

Provider

Select OpenAI or Stability AI. The available models, sizes, and options change based on the provider.

Model

Choose the specific model. See the comparison table above for guidance.

Credential

Select a saved API key for the chosen provider. Only credentials matching the selected provider are shown. See Credential Management for setup.

Prompt

The text description of the image you want to generate. This field supports template variables, allowing you to create dynamic prompts based on form data.

Effective prompts include:

Subject: What is in the image (a person, object, scene)
Style: The visual style (photorealistic, watercolor, digital art, line drawing)
Composition: How the image is arranged (close-up, wide shot, aerial view)
Lighting: The lighting conditions (natural light, studio lighting, dramatic shadows)
Details: Specific attributes (colors, textures, expressions)

Example prompts:

A professional headshot of a friendly customer service
representative, studio lighting, neutral background,
photorealistic style

A modern minimalist logo for a company called "AquaFlow",
incorporating water droplet imagery, blue and white color
scheme, vector art style, clean lines

An isometric illustration of a cozy home office with a
standing desk, dual monitors, plants, and warm lighting,
digital art style

Negative Prompt (Stability AI Only)

Available only when using Stability AI models. Describes what you do not want in the image. This helps the model avoid unwanted elements.

Example negative prompts:

blurry, low quality, distorted, watermark, text overlay

people, faces, hands, photorealistic

Negative prompts are not available with DALL-E models.

Size

The dimensions of the generated image, in pixels. Available sizes depend on the model:

DALL-E 3:

Size	Aspect Ratio	Best For
1024x1024	Square (1:1)	Social media posts, profile images, icons
1792x1024	Landscape (16:9)	Blog headers, banners, presentations
1024x1792	Portrait (9:16)	Mobile screens, stories, posters

DALL-E 2:

Size	Notes
256x256	Fastest, lowest cost, good for thumbnails
512x512	Moderate quality
1024x1024	Best quality for DALL-E 2

Stability AI: Default sizes are 512x512, 768x768, and 1024x1024. Consult the Stability AI documentation for the full list of supported resolutions.

Quality (DALL-E 3 Only)

Quality	Description	Cost
Standard	Good quality, faster generation	Lower
HD	Enhanced detail and consistency	Higher (roughly 2x standard)

Use HD when the image will be displayed at large sizes or when fine details matter. Standard is sufficient for thumbnails, social media, and general-purpose images.

Style (DALL-E 3 Only)

Style	Description	Best For
Natural	Realistic, photographic appearance	Product mockups, realistic scenes
Vivid	Enhanced colors, more dramatic and stylized	Marketing materials, artistic content

Count

The number of images to generate per execution, from 1 to 4. Default is 1.

DALL-E 3 behavior: DALL-E 3 generates one image per API call. When you request multiple images, Buildorado automatically loops internally, making separate API calls for each image. This means generating 4 images takes roughly 4x the time and cost of a single image.

Stability AI V2 models (SD3.5, Stable Image Core): These also generate one image per API call and loop internally for multiple images, similar to DALL-E 3.

DALL-E 2: Can generate multiple images in a single API call, which is faster.

Output

The Image Generation node produces:

Field	Type	Description
`images`	array	Array of file reference objects for all generated images. Each object contains `url`, `key`, `mimeType`, `sizeBytes`, and `filename`.
`imageUrl`	string	URL of the first generated image (shortcut for the most common single-image case)
`revisedPrompt`	string	The prompt as revised by DALL-E 3 (OpenAI only, when the model rewrites the prompt)
`count`	number	The number of images actually generated
`model`	string	The model that was used
`provider`	string	The provider that was used

Generated images are uploaded to S3 and accessible via the returned URLs.

Use Cases

Dynamic Marketing Visuals

Generate custom images based on form submissions:

A user fills out a form describing their brand (colors, industry, style).
The Image Generation node creates a branded visual based on the description.
The image URL is included in a follow-up email or stored in a CMS.

Product Visualization

Create product mockups from descriptions:

An e-commerce form collects product specifications (type, color, material).
The prompt template dynamically describes the product based on form values.
Generated images are stored in Google Drive or attached to a product record.

Content Creation Workflows

Auto-generate blog post or social media images:

A content submission form collects the article title and summary.
The Image Generation node creates a header image matching the content theme.
The image is stored and linked in the content management workflow.

Custom Avatar Generation

Create personalized avatars:

A user selects preferences (style, colors, accessories) in a form.
The prompt builds an avatar description from the selections.
The generated avatar is displayed or stored in the user's profile.

Provider Comparison

Feature	OpenAI (DALL-E)	Stability AI
Prompt following	Excellent (DALL-E 3 rewrites prompts internally)	Very good, follows prompts literally
Negative prompts	Not supported	Supported
Quality options	Standard / HD	Model-dependent
Style control	Natural / Vivid	Via prompt and negative prompt
Image editing	Yes (DALL-E 2 via Image Edit node)	Yes (Stable Image Core via Image Edit node)
Multi-image speed	Sequential (DALL-E 3), Batch (DALL-E 2)	Sequential (V2 models)
Pricing model	Per image	Per image

Choose OpenAI when you want the highest prompt accuracy and do not need negative prompts. DALL-E 3 internally rewrites prompts to improve output quality.

Choose Stability AI when you need fine-grained control over what to include and exclude (via negative prompts), want more model variety, or prefer open-source model options.

Best Practices

Write detailed prompts. The more specific your description, the closer the output matches your intent. Include style, composition, lighting, and color details.
Use negative prompts with Stability AI to exclude unwanted elements. This significantly improves output quality.
Start with standard quality (DALL-E 3) and upgrade to HD only when needed. The cost difference is significant at scale.
Generate one image first to validate your prompt before generating multiples.
Store generated images promptly. Use a Google Drive or S3 action node to persist important images.
Template variables in prompts make each generation unique. Include form field values so the generated image reflects the user's submission.
Content policies apply. Both providers enforce content policies and may reject prompts that request inappropriate content. The node will fail if the prompt is rejected.

Limitations

DALL-E 3 generates exactly one image per API call. Multi-image requests are handled by looping, which multiplies time and cost.
DALL-E 3 may internally modify your prompt for better results. The revised prompt is returned in the revisedPrompt output field.
Stability AI V2 models also generate one image per API call and loop for multiple images.
The node does not support inpainting or outpainting. For editing existing images, use the Image Edit node.
Execution timeout scales with count: 60 seconds per image for sequential models (DALL-E 3, SD3.5, Stable Image Core), 120 seconds total for batch models (DALL-E 2).

On this page