Skip to main content

Image Generation in Poppy

How to use Image generation

Updated over 2 months ago

Poppy's image generation feature lets you create custom images using AI, but with a powerful twist - you can use multiple sources as reference material, including YouTube videos, websites, competitor ads, and your own brand assets.


How to Use Image Generation

Basic Usage

Getting started with image generation is simple:

  • Look for the "Create Image" button above your chat text box

  • You can also use slash commands for quick access

  • Simply describe what you want: "give me a picture of a cat swimming on the beach"

  • Images generate within a few minutes with a Poppy loading screen

Advanced Features - What Makes Poppy Different

  • Connect multiple sources as reference: YouTube videos, websites, logos, competitor ads

  • Use context from non-image sources: Pull copy and text from websites to use in your image text

  • Everything connected shows up in the chat: When you connect sources, they automatically appear as reference material

Image Generation Interface


When you click the "Create Image" option, you'll see any reference photos that are already connected to your chat displayed in the image toolbar. You also have the option to add more images specifically for this image generation (without adding them to your broader knowledge base), and you can remove any images so the AI doesn't reference certain images even if they're connected to your chat. This gives you complete control over what the chat uses to create your image, and you can easily access and update these settings.


Practical Examples

Ad Recreation

Take a competitor's ad (like an Apple iPhone ad) + your logo + your website → Create "Content creation is easier on Poppy" ad in Apple's style, but with your branding and messaging.


Carousel Copying

Use a successful carousel (like Alex Hormozi's posts) + your branding → Recreate the same style for your content, maintaining the visual appeal that works.

YouTube Thumbnails

Connect a YouTube video → AI automatically pulls the thumbnail and lets you replace elements while keeping the same dimensions and style.


Brand Integration

Connect your brand voice materials (website, voice recordings, examples) → Generate on-brand infographics that match your tone and style.


Common Questions & Answers

Q: Does the AI model choice matter for image generation?
A: Yes, you can choose between GPT 1 and NanoBanana

Q: How do I get different aspect ratios for different platforms?
A: The system tries to match your reference image's aspect ratio, but you can request specific formats by saying "redo the image but make it vertical" or "make this for YouTube thumbnail dimensions."

Q: Can I use YouTube links to reference thumbnails?
A: Yes! When you connect a YouTube video, Poppy automatically pulls the thumbnail for reference, making it easy to model successful thumbnail styles.

Q: How much do images cost in credits?
A: Credit usage depends on how much content you have connected as reference material. More connections typically mean higher credit usage per image.

Q: Can it swap faces in thumbnails?
A: Yes, but the results won't look exactly like you due to current AI limitations. Future versions will allow training on your specific appearance for more accurate results.

Q: Why use Poppy instead of ChatGPT for images?
A: ChatGPT can't reference YouTube links, websites, or multiple sources simultaneously. Poppy's ability to use rich context from multiple sources makes it much more powerful for branded content creation.

Q: Can it create entire carousels at once?
A: Currently, it creates one image at a time. For multi-slide carousels, request each slide separately, or ask for prompts for all slides first, then generate them individually.

Pro Tips

  • Start with reference: Always connect examples of what you want to model - successful ads, thumbnails, or posts in your industry

  • Use your brand assets: Connect your website, logo, and brand guidelines to ensure consistent output

  • Be specific about format: Mention if you need the image for YouTube, Instagram, email, etc., to get the right dimensions

  • Iterate quickly: If the first result isn't perfect, ask for specific changes rather than starting over


Did this answer your question?