Image Generation
Create professional images with AI using DALL-E
Image Generation
Image generation in TractionDesk uses OpenAI's DALL-E model to create custom images from text descriptions. Whether you need product photos, lifestyle scenes, abstract concepts, or social media graphics, the AI can generate professional-quality images in 10-15 seconds. The system supports various styles ranging from photorealistic to illustrative, and you can control composition, lighting, subject matter, and overall aesthetic through your prompts and presets.
Creating Your First Image
To generate an image, navigate to the Images section in your dashboard and click "Generate New Image." You'll see options to select a preset (optional) and a prompt (required). If you don't have any presets yet, don't worry—you can enter a custom description directly in the prompt field. Describe what you want to see in detail: "A modern office workspace with a laptop, coffee cup, and notebook on a wooden desk, with soft natural lighting from a window." The more specific you are, the better the results.
After entering your description, click "Generate Image." The process takes about 10-15 seconds, and you'll see a loading indicator while the AI works. Once complete, the generated image appears on screen with options to download, save to your library, or regenerate if you're not satisfied with the result. Each generation costs 5 credits. If you want variations, click "Generate Again" with the same prompt—the AI will create a different version using a new random seed.
For faster generation in the future, create image presets that define your brand's visual style. Click "New Preset" and describe your desired aesthetic in detail. For example: "Professional product photography style: clean white background, centered composition, soft diffused lighting, high contrast, commercial advertising quality, 4K resolution." Save this preset with a memorable name like "Product Showcase" and you can reuse it indefinitely with different prompts ("wireless headphones," "smartphone," "laptop") to generate a consistent series of product images.
Image Presets and Prompts
Image presets are powerful because they encode your brand's visual identity. A well-crafted preset ensures that every image generated for a specific use case maintains consistent quality and style. Create separate presets for different needs: "Social Media Graphics" (bright colors, bold text, 1:1 square format), "Blog Headers" (wide 16:9 format, subtle colors, professional), "Product Photos" (clean backgrounds, centered subjects), or "Lifestyle Imagery" (natural settings, authentic moments, warm tones).
Within each preset, you can create multiple prompts that represent specific content you generate frequently. For instance, using your "Social Media Graphics" preset, you might have prompts for "motivational quote design," "product announcement," "event promotion," or "customer testimonial." The combination of preset + prompt gives you both consistency (the preset's style) and variety (different prompts for different content needs).
Advanced users leverage reference images to guide generation. When creating a preset or generating an image, you can upload a reference image that shows the style, composition, or aesthetic you want to match. The AI analyzes the reference and incorporates its visual characteristics into the generation. This is particularly useful for maintaining brand consistency when you have existing visual examples of what "on-brand" looks like for your company.
Using Images in Campaigns
Images are one of the most common campaign deliverables. When you create a campaign that includes images, you can specify how many images to generate by including the quantity in your campaign description or deliverables configuration. The Campaign Orchestrator will generate the requested number of images based on the campaign's topic and any copy that was created earlier in the workflow. For example, if the campaign generated 5 social posts, the orchestrator might create 5 complementary images—one for each post.
Campaign-generated images are automatically linked to the campaign in your asset library, making it easy to find all visual assets associated with a specific marketing initiative. You can review these images after the campaign completes and download the ones you want to use. Any images you don't need can be deleted to keep your library organized.
Image Specifications
Generated images have the following technical specifications:
- Resolution: 1024x1024 pixels (square) by default, with support for other aspect ratios
- Format: PNG with transparency support
- File Size: Typically 500KB - 2MB per image
- Color Space: sRGB for web compatibility
- Quality: High-resolution suitable for both digital and print use
Images are stored in Supabase cloud storage and accessible via secure URLs. When you download an image, you receive the full-resolution file. The platform also generates thumbnails for previews in your asset library and campaign views.
Best Practices for Image Generation
Be Descriptive: The more detail you provide in prompts, the better the results. Instead of "a cat," write "a fluffy orange tabby cat sitting on a windowsill, looking outside at falling snow, soft afternoon light, cozy atmosphere."
Specify Style: Explicitly mention the style you want: "photorealistic," "watercolor painting," "3D render," "minimalist illustration," "vintage photograph," etc. This guides the AI's artistic interpretation.
Include Composition Details: Describe the layout and framing: "centered composition," "rule of thirds," "overhead view," "close-up shot," "wide angle landscape," "portrait orientation."
Mention Lighting and Color: These elements dramatically affect mood. "Golden hour lighting," "dramatic shadows," "soft diffused light," "vibrant saturated colors," "muted pastel tones," "high contrast black and white."
Reference Brand Elements: If you have specific brand colors or visual themes, include them: "using blue (#007AFF) and white color scheme," "modern tech aesthetic," "warm and approachable feel."
Iterate on Results: First generations rarely perfect. Review the output, identify what you like and what could improve, then regenerate with refined prompts. Most users find their ideal image within 2-3 generations.
Troubleshooting Image Generation
If images aren't generating as expected, try these solutions:
Content Policy Issues: DALL-E has content policies that prevent generating certain subjects (real people's faces, copyrighted characters, violence, etc.). If generation fails with a policy error, modify your prompt to remove restricted content.
Vague Prompts: If results are inconsistent or off-target, add more detail to your prompt. Vague descriptions like "nice background" or "professional look" leave too much to interpretation.
Style Mismatches: If the style doesn't match your brand, create a detailed preset with specific aesthetic instructions and use it consistently for all images.
Technical Errors: Occasional API timeouts can occur. Simply try again—these are usually temporary. If problems persist, check your credit balance and internet connection.