WAN 2.1
PulseAugur coverage of WAN 2.1 — every cluster mentioning WAN 2.1 across labs, papers, and developer communities, ranked by signal.
- 2026-06-27 product_launch Alibaba's Wan team released Wan 2.1, an open-source video generation model suite. source
2 day(s) with sentiment data
-
User seeks best video inpainting model for HUD removal
A user on Reddit is seeking recommendations for the best video inpainting models available today, specifically for removing HUD and UI elements from videos while preserving background quality. They recall using Wan 2.1 …
-
AI-generated "Cyber Slayer" trailer mimics 1995 action movie aesthetic
A video creator has produced a 1995-style action movie trailer titled "Cyber Slayer" using a variety of AI tools, including MiniMax H3, Wan 2.2, and ComfyUI. The project began as a personal anniversary project but evolv…
-
NVIDIA and Hugging Face launch NeMo Automodel for scalable diffusion model fine-tuning
NVIDIA and Hugging Face have collaborated to release NeMo Automodel, an open-source PyTorch library designed to streamline the fine-tuning of diffusion models. This new tool allows users to train and fine-tune models in…
-
Krea 2 VAE Comparison: User Tests Four Models for Image Generation
A user on Reddit compared four different variational auto-encoders (VAEs) for Krea 2, a tool likely related to image generation. The VAEs tested were Qwen Image, WAN 2.1, Krea 2 HD, and Krea 2 Real. The user found minim…
-
MeanFlowNFT brings forward-process RL to average-velocity generators
Researchers have introduced MeanFlowNFT, a novel framework that adapts reinforcement learning (RL) techniques to MeanFlow generators for faster and more efficient content generation. This method bridges the gap between …
-
ACID method accelerates video generation with dynamic caching
Researchers have developed ACID, a novel adaptive caching method designed to accelerate video generation from diffusion models. Unlike existing methods that use a fixed threshold for caching, ACID dynamically adjusts th…
-
Wan Ai releases Wan-Dancer for long-duration dance video generation
Wan Ai has released Wan-Dancer, a new method capable of generating long-duration, high-quality dance videos from music. This framework reportedly surpasses conventional duration barriers, producing stable 720p/30fps vid…
-
New research refines diffusion model noise for better video generation control
Two new research papers propose novel methods for improving controllability in diffusion-based video generation by manipulating the initial noise input. The first paper, WINRO, focuses on text-to-motion generation by re…
-
CineMobile enables on-device cinematic video generation with 40x speedup
Researchers have developed CineMobile, a novel approach for on-device image-to-video generation focused on cinematic motion effects. The system utilizes a distillation-guided pruning strategy to create a compact model, …
-
LTX-2.3 text-to-video model struggles with car interior realism
Users are reporting that the LTX-2.3 text-to-video model struggles with accurately depicting car interiors and driver interactions. Compared to earlier versions like Wan 2.1/2.2, LTX-2.3 appears to have a diminished und…
-
OrbitQuant enables data-agnostic quantization for diffusion transformers
Researchers have developed OrbitQuant, a novel method for post-training quantization of diffusion transformers (DiTs). This technique allows for efficient inference by quantizing in a normalized, rotated basis, eliminat…
-
Alibaba releases open-source Wan 2.1 video generation suite
Alibaba's Wan team has released Wan 2.1, an open-source video generation model suite that aims to make high-quality video generation more accessible. The suite includes capabilities for text-to-video, image-to-video, an…
-
New framework evaluates AI video generation for physical plausibility · 3 sources tracked
Researchers have developed a new evaluation framework called Physics Question Scene Graph (PQSG) to assess the physical plausibility of videos generated by AI models. PQSG uses a hierarchical question-based approach, le…
-
Elden Ring Fan Trailer Created Using AI Tools
A user on Reddit shared a fan-made movie trailer for the game Elden Ring, created using AI tools. The trailer was generated by processing game screenshots through various AI models including Flux Klein and Wan 2.1/2.2, …
-
ReCache optimizes diffusion model caching for better image generation
Researchers have developed ReCache, a novel method for optimizing caching schedules in diffusion models to improve image and video generation efficiency. This technique uses policy gradients to learn a recomputation sch…
-
User Compares 62 Samplers and 16 Schedulers for Image Generation Models
A user has conducted extensive tests comparing various samplers and schedulers for image generation models Z-Image Turbo and WAN 2.1. The results, presented in a color-coded table, indicate performance differences acros…
-
New methods boost video diffusion model efficiency and quality
Researchers are developing new methods to improve the efficiency and quality of video diffusion models. Several papers introduce techniques to optimize attention mechanisms, such as sparse attention (LVSA, Veda) and lin…