How many references or Elements does each image model support?
Image generation models in LTX Studio support a different number of reference images and Elements per prompt – up to 14. Note that saved Elements count toward the same limit as uploaded references.
Reference images: All plans can use reference images in text-to-image prompts. Which image models are accessible depends on your plan — see plans for details.
Elements: Available on Standard, Pro, and Enterprise plans.
How many references or Elements can each model support?
| Image model | Max references or Elements |
|---|---|
| GPT Image 2 | 14 |
| Nano Banana 2 or Pro | 14 |
| Flux 2 Pro | 8 |
| Z-Image | 10 |
To see which models are available on your plan, visit the pricing page.
Elements are saved, reusable reference images, so they draw from the same pool as uploaded references. You can use up to the model's limit across any combination: uploaded references, Elements, or both.
Which prompts support reference images and Elements?
Reference images (and saved Elements) work in any text-to-image prompt, whether you're generating in Gen Space, Canvas, or the Storyboard. Anywhere you can write a prompt and generate an image, you can add references or tag Elements to define characters, styles, objects, locations, and more.
Why is there a limit?
Each model can only give so much focus to reference inputs. Spreading attention across too many reduces output quality and consistency, so limits are set to keep results optimized.
Frequently asked questions
Can I use references and Elements in video prompts?
Elements and image references — used to define characters, styles, objects, or locations — aren't supported in video generation prompts.
Video generation does support image inputs, but they work differently: as a start or end frame, or as a style reference for Video-to-Video motion control. These are a distinct input type with a different purpose.
I need to use more references or Elements than my model allows. What can I do?
Check which model you're on. GPT Image 2 and Nano Banana both support up to 14 inputs — the highest available. Flux 2 Pro caps at 8, Z-Image at 10.
If you're using color Elements, a workaround is to drop them from your prompt and use hex codes in your text prompt directly instead. This frees up slots for other references or Elements.
Beyond these options, adding more inputs isn't currently supported as doing so would reduce output quality.
Does the limit apply to all Element types combined?
Yes. The limit is a total across any Element type (character, location, color, font etc.) and uploaded reference, or any mix. For example, you could max out your limit with 10 Character Elements, 2 Color Elements and 2 references images with Nano Banana Pro.