Kosmos-G: Generating Images in Context with Multimodal Large Language Models

TL;DR

Kosmos-G is a multimodal model that enables zero-shot, subject-driven image generation with interleaved multi-image and text inputs, advancing the goal of treating images as a language for generation.

Contribution

It introduces Kosmos-G, a novel approach that aligns MLLMs with CLIP for flexible, zero-shot image generation without test-time tuning or decoder modifications.

Findings

Achieves zero-shot subject-driven generation with interleaved multi-image and text inputs

Seamlessly integrates with various U-Net techniques and personalized decoders

Requires no modifications to the image decoder during training

Abstract

Recent advancements in subject-driven image generation have made significant strides. However, current methods still fall short in diverse application scenarios, as they require test-time tuning and cannot accept interleaved multi-image and text input. These limitations keep them far from the ultimate goal of "image as a foreign language in image generation." This paper presents Kosmos-G, a model that leverages the advanced multimodal perception capabilities of Multimodal Large Language Models (MLLMs) to tackle the aforementioned challenge. Our approach aligns the output space of MLLM with CLIP using the textual modality as an anchor and performs compositional instruction tuning on curated data. Kosmos-G demonstrates an impressive capability of zero-shot subject-driven generation with interleaved multi-image and text input. Notably, the score distillation instruction tuning requires no modifications to the image decoder. This allows for a seamless substitution of CLIP and effortless integration with a myriad of U-Net techniques ranging from fine-grained controls to personalized image decoder variants. We posit Kosmos-G as an initial attempt towards the goal of "image as a foreign language in image generation." The code can be found at https://aka.ms/Kosmos-G