CreateInstanceDiffusionTracking

The CreateInstanceDiffusionTracking node receives a JSON string containing object bounding box coordinates and, combined with user-defined categories, IDs, and prompts, constructs the tracking data structure required by InstanceDiffusion.

The CreateInstanceDiffusionTracking node is a specialized tool for generating instance tracking data. Its core function is to encapsulate object coordinates, category information, and prompts into a specific format, enabling subsequent InstanceDiffusion nodes to identify and process specific instances within an image. This node serves as a crucial bridge connecting object detection with refined image generation.

Node Function

This node receives a JSON string containing object bounding box coordinates and, combined with user-defined categories, IDs, and prompts, constructs the tracking data structure required by InstanceDiffusion. Its internal mechanism adheres to the specific format of InstanceDiffusion, for example, organizing data in the form "class_id.class_name": "prompt", ensuring compatibility with downstream generation nodes.

Node Parameter Description - CreateInstanceDiffusionTracking

Control Parameters (Parameters)

Parameter Name	Data Type	Required	Default Value	Range/Options	Description
`coordinates`	STRING	Yes	-	-	coordinates (Input Parameter): A JSON string containing object bounding box coordinates, serving as the foundation for generating tracking data. The format must meet the requirements.
`width`	INT	Yes	512	16 - 4096 (Step: 1)	width (Input Parameter): Passes through the input image width value.
`height`	INT	Yes	512	16 - 4096 (Step: 1)	height (Input Parameter): Passes through the input image height value.
`bbox_width`	INT	Yes	512	16 - 4096 (Step: 1)	bbox_width (Input Parameter): Passes through the input bounding box width value.
`bbox_height`	INT	Yes	512	16 - 4096 (Step: 1)	bbox_height (Input Parameter): Passes through the input bounding box height value.
`class_name`	STRING	Yes	class_name	-	class_name (Input Parameter): Specifies a category name for the tracked instance, e.g., "person" or "car".
`class_id`	INT	Yes	0	0 - 255 (Step: 1)	class_id (Input Parameter): Assigns a numeric ID to the instance for unique identification, ranging from 0 to 255.
`prompt`	STRING	Yes	prompt	-	prompt (Input Parameter): Passes through the input prompt text.
`size_multiplier`	FLOAT	No	[1]	-	Optional list of float values for size multiplier.
`fit_in_frame`	BOOLEAN	No	True	-	fit_in_frame (Input Parameter): Whether to fit the instance within the frame.

Output

Parameter Name	Data Type	Description
tracking	TRACKING	Tracking data conforming to the InstanceDiffusion format requirements.
prompt	STRING	prompt (Output Parameter): Directly passes through the input prompt text.
width	INT	width (Output Parameter): Directly passes through the input image width value.
height	INT	height (Output Parameter): Directly passes through the input image height value.
bbox_width	INT	bbox_width (Output Parameter): Directly passes through the input bounding box width value.
bbox_height	INT	bbox_height (Output Parameter): Directly passes through the input bounding box height value.

Usage Scenarios

This node is highly useful in workflows for creating animations or generating consecutive video frames. For example, you can first use an object detection node to obtain the coordinates of "person heads" in each frame of a video, then feed this sequence of coordinates into this node, specifying the category as "head" and the corresponding prompt. This provides the InstanceDiffusion node with precise, frame-by-frame instance control information.

Notes

Ensure the input coordinate string is valid JSON format. Additionally, the output data format of this node is fixed and must strictly match the input requirements of the InstanceDiffusion node; otherwise, workflow execution may fail.