11 papers
NitroFusion2: Bringing Adversarial Loss to the Text Stream of Multimodal Diffusion Transformers
Dar-Yen Chen, Yi-Zhe Song
Adversarial diffusion distillation has conventionally relied on image features for discrimination. We revisit this choice for multimodal diffusion transformers (MMDiTs), whose text…
Context-Matched Distillation: Teacher Causality for Autoregressive Video Distillation
Hmrishav Bandyopadhyay, Xuanchi Ren, Zijian Huang +7
Interactive autoregressive video generation demands both low-latency rollouts and precise online control. Few-step distillation accelerates generation by reducing denoising steps,…
CaricHarmony: Contrastive Diffusion Paths for Identity-Preserving Caricature Synthesis
Dongyu Wang, Dar-Yen Chen, Yi-Zhe Song
Sketch-based caricature synthesis suffers from a fundamental failure mode: when identity and shape conditions are combined in diffusion models, they create destructive interference…
Block Cascading: Training Free Acceleration of Block-Causal Video Models
Hmrishav Bandyopadhyay, Nikhil Pinnaparaju, Rahim Entezari +3
Block-causal video generation faces a stark speed-quality trade-off: small 1.3B models manage only 16 FPS while large 14B models crawl at 4.5 FPS, forcing users to choose between r…
FrontierGS: Progressive View-Space Frontier Expansion for Sparse 3D Gaussian Splatting
Zijian Wu, Mingfeng Jiang, Zidian Lin +6
3D Gaussian Splatting (3DGS) enables efficient, high-fidelity novel view synthesis, yet its performance degrades severely under sparse-view supervision due to extreme ill-posedness…
SD3.5-Flash: Distribution-Guided Distillation of Generative Flows
Hmrishav Bandyopadhyay, Rahim Entezari, Jim Scott +3
We present SD3.5-Flash, an efficient few-step distillation framework that brings high-quality image generation to accessible consumer devices. Our approach distills computationally…
MotionFlow:Learning Implicit Motion Flow for Complex Camera Trajectory Control in Video Generation
Guojun Lei, Chi Wang, Yikai Wang +3
Generating videos guided by camera trajectories poses significant challenges in achieving consistency and generalizability, particularly when both camera and object motions are pre…
SketchAnimator: Animate Sketch via Motion Customization of Text-to-Video Diffusion Models
Ruolin Yang, Da Li, Honggang Zhang +1
Sketching is a uniquely human tool for expressing ideas and creativity. The animation of sketches infuses life into these static drawings, opening a new dimension for designers. An…
Normalized Attention Guidance: Universal Negative Guidance for Diffusion Models
Dar-Yen Chen, Hmrishav Bandyopadhyay, Kai Zou +1
Negative guidance -- explicitly suppressing unwanted attributes -- remains a fundamental challenge in diffusion models, particularly in few-step sampling regimes. While Classifier-…
NitroFusion: High-Fidelity Single-Step Diffusion through Dynamic Adversarial Training
Dar-Yen Chen, Hmrishav Bandyopadhyay, Kai Zou +1
We introduce NitroFusion, a fundamentally different approach to single-step diffusion that achieves high-quality generation through a dynamic adversarial framework. While one-step…
FlipSketch: Flipping Static Drawings to Text-Guided Sketch Animations
Hmrishav Bandyopadhyay, Yi-Zhe Song
Sketch animations offer a powerful medium for visual storytelling, from simple flip-book doodles to professional studio productions. While traditional animation requires teams of s…