#Beijing #Multimodal AI #HiDream.ai #ACM MM

HiDream.ai's Outstanding Achievement at ACM MM 2025

On November 6, 2025, HiDream.ai made headlines by winning the Best Demo award at the prestigious 33rd ACM International Conference on Multimedia (ACM MM 2025). This accomplishment marks a significant milestone, as HiDream.ai becomes the first Chinese startup in the field of multimodal generative AI to receive this esteemed recognition, signifying its superior research capabilities and innovative prowess in the industry.

The ACM MM conference, hosted by the Special Interest Group on Multimedia (SIGMM) under the Association for Computing Machinery (ACM), is recognized globally as a leading event in the multimedia sector. It promotes the advancement of research innovation and industrial applications related to multimedia technologies and attracts top scholars and industry leaders from around the world. Winning the Best Demo award reflects the high international recognition of HiDream.ai’s research outcomes and its team's exceptional competence in multimedia technology.

At the core of HiDream.ai’s offering is the HiDream-Agent, a groundbreaking multimodal agent designed to simplify and enhance the complex processes involved in visual content generation. This innovative agent transforms the process into a conversational experience, allowing users to engage intuitively with visual creation. HiDream-Agent integrates multiple functions—text-to-image generation, instruction-based image editing, and text/image-to-video generation—into a single cohesive interface. This effectively addresses a significant challenge in the industry: achieving consistent cross-modal semantic alignment.

The HiDream-Agent is built on the advanced HiDream-I1 model, which boasts a sparse Diffusion Transformer (DiT) structure paired with a dynamic Mixture-of-Experts (MoE) architecture. This fusion achieves remarkable performance on several international benchmarks, including HPS and GenEval. For its instruction-based image editing functionalities, HiDream.ai has fine-tuned the HiDream-I1 with robust in-context visual conditioning, enabling users to execute precise modifications with greater accuracy.

The implications of this technology go beyond just efficiency; it paves the way for a new paradigm of interactive visual storytelling and collaborative content creation within the realm of multimodal generative AI. By combining content generation and editing into a seamless dialogue-driven workflow, HiDream-Agent significantly reduces the effort needed to create high-quality visual materials. This technology allows creators to experience a