hi all - is there anythng in gamma generate api to optimise the output / export file sizes - ie. which image model is recommended for file size and does export as pptx or pdf make a difference
Hey Dan! 👋 Good question on file sizes. Short answer: there's no direct file size optimization parameter The export process is handled automatically on our end, so you don't have control over compression settings or output optimization. What affects file size in practice:
Number of cards - fewer cards = smaller file
Image source - "source": "noImages" or "source": "placeholder" will obviously be smallest
Image model - the docs don't specify file size differences between models, but generally the "basic" tier models (like imagen-3-flash or flux-1-quick) produce simpler images that may compress better than the ultra-detailed premium models
PDF vs PPTX Not documented which is smaller, but in my experience PDFs tend to compress better since they're flattened. PPTX files retain editable layers which can add overhead. Worth testing with your specific content to see what works best for your use case. Practical suggestion If file size is a priority, I'd run a few test generations with different image models and compare the export sizes. The basic models at 2 credits/image are a good starting point. Docs for reference: https://developers.gamma.app Let me know if you want me to dig into anything else!
Thanks for this link to my other direct message. I suppose I want to know if there's a good way to prompt to include more images and force accent images rather than AI-generated images. Not sure if you saw that one but is there anything I can do in the prompt?
I'm assuming if I don't generate AI images, then the Gamma would also be faster to work with. But really I just want to pull Accent Images—no AI-generated images or minimal AI-generated images—and then export as a PDF. I feel like that should make the process quite fast and optimized. What do you think?
Hey Dan! 👋 Perfect timing on this question. We actually just shipped a new feature that does exactly what you're looking for. Use "source": "themeAccent" in your request: json
{
"inputText": "Your content here",
"textMode": "generate",
"themeId": "your-theme-id",
"imageOptions": {
"source": "themeAccent"
}
}How it works: The API pulls from your theme's configured accent backgrounds instead of generating AI images. Each card that uses an accent image layout gets a random selection from your theme's accent images, and it's seeded by card ID so regenerating the same deck produces consistent results. You're right about speed and optimization: Since no AI image generation occurs, this should be noticeably faster. No waiting on image models, no credits spent on image generation. Combined with PDF export, you're looking at a pretty streamlined workflow. Requirements: Your themeId needs to have accent images configured. You can set these up in the theme editor under Theme Settings → Accent Images. A lot of the built-in themes already have accent backgrounds, so you can test with those too. The model and style parameters in imageOptions get ignored when using themeAccent since there's nothing to generate. If your theme has no accent images configured, cards will fall back to blank layouts instead. We're updating the docs to include this shortly. Let me know how it goes!
cheaper and faster
my middle name
