Progressive Text-to-Image Diffusion with Soft Latent Direction

TL;DR

This paper presents a progressive text-to-image synthesis method that uses a Large Language Model to decompose complex descriptions and a novel SRF framework for entity manipulation, improving handling of multiple entities and complex instructions.

Contribution

It introduces a new progressive synthesis and editing framework with SRF, enabling better multi-entity handling and complex text inputs in text-to-image generation.

Findings

Enhanced object synthesis with complex and lengthy descriptions

Achieved new benchmark performance in text-to-image tasks

Effective manipulation of multiple entities respecting constraints

Abstract

In spite of the rapidly evolving landscape of text-to-image generation, the synthesis and manipulation of multiple entities while adhering to specific relational constraints pose enduring challenges. This paper introduces an innovative progressive synthesis and editing operation that systematically incorporates entities into the target image, ensuring their adherence to spatial and relational constraints at each sequential step. Our key insight stems from the observation that while a pre-trained text-to-image diffusion model adeptly handles one or two entities, it often falters when dealing with a greater number. To address this limitation, we propose harnessing the capabilities of a Large Language Model (LLM) to decompose intricate and protracted text descriptions into coherent directives adhering to stringent formats. To facilitate the execution of directives involving distinct semantic operations-namely insertion, editing, and erasing-we formulate the Stimulus, Response, and Fusion (SRF) framework. Within this framework, latent regions are gently stimulated in alignment with each operation, followed by the fusion of the responsive latent components to achieve cohesive entity manipulation. Our proposed framework yields notable advancements in object synthesis, particularly when confronted with intricate and lengthy textual inputs. Consequently, it establishes a new benchmark for text-to-image generation tasks, further elevating the field's performance standards.