UI Improvement for DragDiffusion: Correct Display for Non-Square Images

[Rubenacp]

Hi everyone,
If you've tried using DragDiffusion with a rectangular (non-square) image, you've probably noticed that the image gets cropped in the "Draw Mask" panel. This makes it difficult to draw an accurate mask.
Here's a simple fix for that.
The Problem:
The UI forces all images into a square container (480x480px), which crops any image that doesn't match this aspect ratio.
The Cause:
In the drag_ui.py file, the Gradio gr.Image components have a hardcoded height and width (height=LENGTH, width=LENGTH).
The Solution (100% Safe):
The fix is to simply remove these fixed-size parameters, allowing the UI to automatically adapt to the shape of your uploaded image.
This is a visual-only change. It does not affect the backend processing, LoRA training, or the final result in any way. The program always uses the full, original image behind the scenes.
How to Apply the Fix:
Open the drag_ui.py file.
Find the lines where the gr.Image components are defined (e.g., canvas, input_image, output_image).
Remove the height=LENGTH and width=LENGTH parameters from these definitions.
Example - Before:

canvas = gr.Image(type="numpy", tool="sketch", label="Draw Mask", show_label=True, height=LENGTH, width=LENGTH) # for mask painting

Example - After:

canvas = gr.Image(type="numpy", tool="sketch", label="Draw Mask", show_label=True) # for mask painting