GLIGENTextBoxApplyBatchCoords Node Documentation

Overview

The GLIGENTextBoxApplyBatchCoords node is designed to enhance ComfyUI workflows, especially when integrated with AnimateDiff-Evolved, by allowing scheduling of GLIGEN text box positions within a batch. This node is particularly useful for users who want to control text box positioning dynamically over a sequence of images, enabling advanced prompt manipulation and placement.

Functionality

The node leverages GLIGEN models to position text boxes over a batch of images according to specified coordinates. Users can define the size and position of these text boxes, which will be scheduled throughout the batch, offering a high level of customization in workflows that involve compositional diffusion processes.

Inputs

1. latents: A latent input tensor used to determine the batch size. This tensor informs the node of the number of coordinates needed.
2. conditioning_to: The conditioning input, typically connecting from or to other nodes that handle prompt conditioning.
3. clip: The CLIP text encoder, which should match the one used for the main prompt text.
4. gligen_textbox_model: The GLIGEN model, which must be loaded via an appropriate GLIGEN Loader node.
5. coordinates: A JSON string that contains a list of coordinate points for positioning, compatible with outputs from nodes like the Spline Editor.
6. text: The text to be positioned using the GLIGEN model.
7. width: The width of the GLIGEN text box in pixels.
8. height: The height of the GLIGEN text box in pixels.
9. size_multiplier (optional): A list of multipliers for dynamically adjusting the size of each text box.

Outputs

1. conditioning: A conditioned tensor that integrates the GLIGEN text box positions. This output is typically connected to a sampler node in the workflow.
2. coord_preview: An image preview showing the planned placements of text boxes as defined by the coordinates, which helps users visualize the coordination of text box positioning.

Usage in ComfyUI Workflows

This node is primarily intended for use within workflows that incorporate AnimateDiff-Evolved. It empowers users to:

• Dynamically schedule the movement and transformation of text boxes across a batch of images.
• Integrate with spline editor nodes or other coordinate-generating nodes to create complex animations or transitions.
• Enhance visual storytelling or presentation by allowing text elements to follow or interact with image content.

In a typical workflow, you'd use this node by connecting latents, clip encoders, and a GLIGEN model. Then, provide coordinate strings from a node like the Spline Editor to control movement paths for text boxes, rendering the box's appearance consistently across the batch.

Special Features and Considerations

• GLIGEN Model Requirement: This node requires the GLIGEN model, which can be downloaded via the Manage's "Install Models" menu or directly from available repositories.
• Selection of Inputs: Proper synchronization of text box size and positions is crucial. Ensure the input size settings (width/height) and coordinate lists match your desired output specifications.
• Visualization: The node's coord_preview output is particularly useful for pre-visualizing text positionings before final processing, enabling adjustments and refinements as needed.
• Pairing: It’s worth considering pairing with other control nodes like the Spline Editor for seamless integration. The GLIGENTextBoxApplyBatchCoords node shines in workflows involving complex spatial and temporal adjustments of text elements in multimedia presentations.