Tag
RefineEdit is a training-free prompt-to-prompt image editing method that uses a generative refinement network to enhance edit localization and background preservation, achieving top benchmark scores.
SCTab-Diff is a semantics-consistent tabular diffusion framework that uses weak semantic priors to generate high-fidelity synthetic tabular data, improving distributional fidelity and semantic consistency over existing methods.
Repackaged model files for Qwen-Image-2.1 optimized for ComfyUI workflows, including text-to-image and image edit capabilities.
Qwen-Image-2.1 is an open-source unified text-to-image and image editing model with 7B parameters, featuring efficient architecture, transparency support, and versatile editing capabilities.
Dynin-Robotics is an omnimodal unified diffusion model that integrates vision, language, and action for language-conditioned robot control, improving adaptation and success through joint denoising and test-time scaling.
An independent audio researcher has trained a model called Foundation-1 that generates infinite one-shot sounds for music production and converts text prompts into playable synths, releasing the model along with documentation and inferencing tools.
Srijika is a system that generates installable OpenType fonts for nine Indic scripts by restyling glyphs using a latent diffusion model while preserving layout consistency.
DiDrive is a risk-aware hierarchical diffusion framework for safe offline reinforcement learning in autonomous driving that improves performance in complex traffic scenarios through integrated representation learning and distribution correction optimization.
Celeris-1 Magnus is a hybrid diffusion model optimized for agentic work, achieving a 41.2% solve rate on the τ³-bench banking benchmark at a 55-second median time, outperforming models like GPT-5.6-sol.
The paper presents Puppeteer, a diffusion-based co-speech gesture model that uses causal latent tokens and object geometry to generate temporally coherent, posture-aware, and physically grounded gestures.
LayerRecall improves long-video consistency in diffusion models by selectively routing historical memory to specific layers, supervised by cross-horizon prediction matching.
I trained a 1.2B DiT model for instrumental game music generation, using Stable Audio's VAE and aiming to cover diverse styles. The project is open-source with a WebUI and samples available on HuggingFace.
ADAPT is a physics-aware conditional diffusion model for HVAC control that reduces energy consumption and occupant discomfort, with robust transferability to unseen environments.
Block3D accelerates text-to-3D generation by using block-wise diffusion with confidence-guided correction to reduce inference time while preserving geometric fidelity, achieving a 5.15x speedup.
An individual trained a diffusion model to generate 32x32 pixel images on a Shrike lite microcontroller with only 264KB RAM, experimenting with FPGA acceleration that hit memory bottlenecks, resulting in noisy but sometimes interesting outputs.
DFlash 2 is a block-diffusion draft model for speculative decoding that improves inference speed for the Qwen3.8-27B language model, offering higher acceptance length and throughput in benchmarks.
Molei Tao introduces FLARE, a diffusion language model that achieves near GPT5 performance with significantly faster inference speed.
UniSwap is a new framework for joint audio-visual identity swapping in talking videos, using a unified streaming audio-visual diffusion transformer to replace appearance and vocal timbre while preserving source content and dynamics.
Surg-UniWorld is a unified surgical world model with multimodal control experts, enabling controllable generation of coherent instrument-tissue interaction videos using edge, depth, and optical-flow inputs. It introduces a new benchmark (Cholec80-SurgWAM) and outperforms existing controllable video generation methods.
This article provides instructions for placing repackaged MiniMax-Music-3 model files into ComfyUI directories for music generation.