SVG-T2I: Scaling Up Text-to-Image Latent Diffusion Model Without Variational Autoencoder

TL;DR

SVG-T2I introduces a novel approach for high-quality text-to-image synthesis directly within the Visual Foundation Model (VFM) feature space, demonstrating competitive results without using Variational Autoencoders.

Contribution

The paper extends the SVG framework to support direct text-to-image generation in the VFM domain, enabling scalable and effective diffusion-based synthesis without VAEs.

Findings

Achieves 0.75 on GenEval and 85.78 on DPG-Bench benchmarks.

Validates the representational power of VFMs for generative tasks.

Provides open-source tools and models for further research.

Abstract

Visual generation grounded in Visual Foundation Model (VFM) representations offers a highly promising unified pathway for integrating visual understanding, perception, and generation. Despite this potential, training large-scale text-to-image diffusion models entirely within the VFM representation space remains largely unexplored. To bridge this gap, we scale the SVG (Self-supervised representations for Visual Generation) framework, proposing SVG-T2I to support high-quality text-to-image synthesis directly in the VFM feature domain. By leveraging a standard text-to-image diffusion pipeline, SVG-T2I achieves competitive performance, reaching 0.75 on GenEval and 85.78 on DPG-Bench. This performance validates the intrinsic representational power of VFMs for generative tasks. We fully open-source the project, including the autoencoder and generation model, together with their training, inference, evaluation pipelines, and pre-trained weights, to facilitate further research in representation-driven visual generation.