Guide to Multiple Image References Combinations
Explore the 120 possible reference combinations in AI Generator. Master the use of Multiple Image References to create varied and unique images.
This feature is available in AI Generator's SD mode (in Essential mode, you can use up to one Image Reference). It only works with Stable Diffusion XL-based models, such as JuggernautXL X, RealVisXL V4.0, and many others.
What reference combinations are (or aren’t) allowed?
Duplicating reference types is not allowed. For example, you can’t select two Image to Image references, or a Character, Style, and then Character again.
You can mix references from different categories (Image to Image, IP Adapters, and ControlNet) or use multiple references from the same category (e.g., three varied ControlNet models).
Therefore, 36 two-reference combinations and 84 three-reference combinations are available, for a total of 120 possible reference combinations.
The order in which the references are set makes no difference. For example, the combination of Character in the 1st position and Style in the 2nd will work the same as Style in the 1st and Character in the 2nd.
We recommend reading the guide on the basics using Multiple Image References, which is full of useful tips!
2-reference combinations
The following reference combinations work well when using the same image for both references to better preserve its structure:
Soft Edges + Pose,
Hard Edges + Pose,
Depth + Normal Map,
Depth + Pose,
Normal Map + Pose.
Image to Image combinations
Image to Image combinations are useful for generating similar variations of an existing image:
Image to Image + Style,
Image to Image + Content,
Image to Image + Character,
Image to Image + Hard Edges,
Image to Image + Soft Edges,
Image to Image + Depth,
Image to Image + Normal Map,
Image to Image + Pose.
When using ControlNet models with Image to Image, it essentially becomes an Image to Image process that focuses on preserving the chosen structure.
If you'd like to make more significant changes to the source image, you can write a prompt that differs from what's visible in the picture and set a low reference strength. This applies to all Image to Image combinations.
For more information on using Image to Image references effectively, please refer to our Image to Image guide.
Style + Content
Style + Character
Style + Hard Edges
Style + Soft Edges
Style + Depth
Style + Normal Map
Style + Pose
Content + Character
Content + Hard Edges
Content + Soft Edges
Content + Depth
Content + Normal Map
Content + Pose
Character + Hard Edges
Character + Soft Edges
Character + Depth
Character + Normal Map
Character + Pose
Hard Edges + Soft Edges
Hard Edges + Depth
Hard Edges + Normal Map
Soft Edges + Depth
Soft Edges + Normal Map
3-reference combinations
As with 2-reference combinations, some 3-reference combinations work best when using the same image for either 2 or 3 of the references to preserve the original structure better.
For example, the following combinations work best when using the Style IP Adapter to transfer your desired style to another image used as the remaining two references:
Style + Soft Edges + Pose,
Style + Depth + Normal Map,
Style + Depth + Pose,
Style + Normal Map + Pose,
Style + Hard Edges + Pose.
You could also transfer content with the following combinations:
Content + Hard Edges + Depth,
Content + Hard Edges + Normal Map
Content + Hard Edges + Pose
Content + Soft Edges + Depth,
Content + Soft Edges + Normal Map,
Content + Soft Edges + Pose,
Content + Depth + Normal Map,
Content + Depth + Pose,
Content + Normal Map + Pose.
Or character's pose and general structure of the image:
Character + Soft Edges + Pose,
Character + Depth + Pose,
Soft Edges + Depth + Pose,
Soft Edges + Normal Map + Pose,
Depth + Normal Map + Pose,
Hard Edges + Depth + Pose,
Hard Edges + Normal Map + Pose,
Hard Edges + Soft Edges + Pose.
Image to Image combinations
Once again, Image to Image combinations are useful for generating similar variations of an existing image:
Image to Image + Style + Content,
Image to Image + Style + Character,
Image to Image + Style + Hard Edges,
Image to Image + Style + Soft Edges,
Image to Image + Style + Depth,
Image to Image + Style + Normal Map,
Image to Image + Style + Pose,
Image to Image + Content + Character,
Image to Image + Content + Hard Edges,
Image to Image + Content + Soft Edges,
Image to Image + Content + Depth,
Image to Image + Content + Normal Map,
Image to Image + Content + Pose,
Image to Image + Character + Hard Edges,
Image to Image + Character + Soft Edges,
Image to Image + Character + Depth,
Image to Image + Character + Normal Map,
Image to Image + Character + Pose,
Image to Image + Hard Edges + Soft Edges,
Image to Image + Hard Edges + Depth,
Image to Image + Hard Edges + Normal Map,
Image to Image + Hard Edges + Pose,
Image to Image + Soft Edges + Depth,
Image to Image + Soft Edges + Normal Map,
Image to Image + Soft Edges + Pose,
Image to Image + Depth + Normal Map,
Image to Image + Depth + Pose,
Image to Image + Normal Map + Pose.
If you'd like to make more significant changes to the source image, you can write a prompt that differs from what's visible in the picture and set a low reference strength. This applies to all Image to Image combinations.
For more information on using Image to Image references effectively, please refer to our Image to Image guide.
Style + Content + Character
Style + Content + Hard Edges
Style + Content + Soft Edges
Style + Content + Depth
Style + Content + Normal Map
Style + Content + Pose
Style + Character + Hard Edges
Style + Character + Soft Edges
Style + Character + Depth
Style + Character + Normal Map
Style + Character + Pose
Style + Hard Edges + Soft Edges
Style + Hard Edges + Depth
Style + Hard Edges + Normal Map
Style + Soft Edges + Depth
Style + Soft Edges + Normal Map
Content + Character + Hard Edges
Content + Character + Soft Edges
Content + Character + Depth
Content + Character + Normal Map
Content + Character + Pose
Content + Hard Edges + Soft Edges
Character + Hard Edges + Soft Edges
Character + Hard Edges + Depth
Character + Hard Edges + Normal Map
Character + Hard Edges + Pose
Character + Soft Edges + Depth
Character + Soft Edges + Normal Map
Character + Depth + Normal Map
Character + Normal Map + Pose
Hard Edges + Soft Edges + Depth
Hard Edges + Soft Edges + Normal Map
Hard Edges + Depth + Normal Map
Soft Edges + Depth + Normal Map
Choosing the right combination of references can make all the difference in achieving your desired outcome. We hope this guide provided a wealth of inspiration to get you started!