How to generate images through dialogue?

Hello, thank you for your excellent research work. The model's performance is impressive.
I saw on your website's homepage that you can generate or edit images through dialogue and maintain contextual connections. However, it seems that currently, this repo only allows single-image generation through Gradio interface. How can I generate images through dialogue?
Looking forward to your reply. ^^ ^^

The issue mentions only the Gradio interface and its current single-image generation behavior; begin there and trace the existing generation entry point. Determine whether dialogue and contextual image generation are already exposed, then establish the required scope and verification path. Done means a reproducible dialogue-based image workflow is available.

Written by the indexing model from the issue text.