Persona Slots

Persona slots are the visual knowledge library of personas.

Persona Slots

Persona slots are the visual knowledge library of personas.

Slot Number
1-6	Visual training slots from 1 to 6. This slots make fine-tuning for your persona's visual style. Note: 1 is master slot and using for Style Transfer with pid, it is style_transfer parameter for generators. If you pass style_transfer=persona to the generate endpoint, it will use Slot 1 for Style Transfer
7	Main image for persona to manipulate, it is image parameter for generators

Curl Example

# Store image in slot 1 (style reference)
curl -X POST https://api.luw.ai/v2/personas/123/trains \
-H "Authorization: Bearer LUW_API_TOKEN" \
-d "photo=https://example.com/style.jpg" \
-d "slot=1"

# Store image in slot 7 (main image)
curl -X POST https://api.luw.ai/v2/personas/123/trains \
-H "Authorization: Bearer LUW_API_TOKEN" \
-d "photo=https://example.com/room.jpg" \
-d "slot=7"

Using Slots in Generation

To use slots in generation, provide the persona ID using the pid parameter:

curl -X POST https://api.luw.ai/v2/generate \
-H "Authorization: Bearer LUW_API_TOKEN" \
-d "model=interior" \
-d "pid=123" \
-d "style_transfer=persona"  # Uses slot 1 as style reference

This will automatically:

• Use slot 1 for style reference if style_transfer=persona

• Use slot 7 as the input image if no explicit image URL provided

When configuring persona slots in Interior AI, providing a pid (Persona ID) parameter eliminates the need to specify the image and style_transfer parameters.

For instance, in the Fluw implementation, once you establish a persona and assign an image to slot 7, subsequent operations require only the pid parameter. The system will automatically utilize the slot's designated image as the starting frame for video generation, with processing continuing from that initial state. This persona slot functionality is not unique to Interior AI, or Fluw - it can be implemented across all Luw.ai generators, allowing for consistent style application and streamlined parameter management.