sDMD adds adaptive Sinkhorn correction to diffusion distillation
With 2–4 steps on FLUX.1-dev, Qwen-Image-20B, and Wan2.1-14B, authors report gains over DMD2 and other baselines.
Stories per day · 30 days
1 RSS
Relatedblack-forest-labs3huggingface3image-generation3comfyui2diffusion2distillation2image-to-image2open-weights2concept-erasure1consumer-gpu1diffusers1flow-matching1image-synthesis1model-support1sinkhorn1snr1spatial-grounding1text-to-image1
18 stories, newest first · one card per story
Stories tagged #flux
Few-step image models raise unsafe output; DICE cuts it
On FLUX Schnell, DICE lowers the unsafe-generation rate to 0.046, versus at least 0.161 for prior erasure methods tested.
High-noise SNR training lifts layout adherence in flow-matching models
Shifting training toward high noise raised synthetic layout adherence from 25.9% to 80.8% and COCO Average Precision from 1.0 to 57.3.
Black Forest Labs' open-weights image model FLUX 3 Image Klein runs on consumer GPUs with just 8 GB of VRAM
Black Forest Labs released FLUX 3 Image Klein, a 5.6B open image model that runs on 8 GB GPUs.
Welcome FLUX.3 Image Klein in Diffusers
Hugging Face adds FLUX.3 Image Klein support to the Diffusers library.
black-forest-labs/flux-3-image-klein-comfy — new model trending #30 on Hugging Face
Black Forest Labs' small FLUX image-to-image model is trending on Hugging Face with only 39 downloads.
black-forest-labs/flux-3-image-klein — new model trending #29 on Hugging Face
Black Forest Labs' FLUX.3 Image Klein image-to-image model is trending on Hugging Face with 3,241 downloads.
InfoVis: Unified Slide Image Generation and Editing
InfoVis, a 9B slide model, nearly matches Nano-Banana-2 on a 1,527-instance benchmark.
Iris-3B: Going Beyond the Latent with Pixel-Space Diffusion Training, Conversion and Fine-Tuning
Iris-3B shows 3B pixel-space pretraining can match latent image models, without better downstream detail.
Black Forest Labs launches Flux 3 Image with multi-step editing that leaves the rest of your picture alone
Black Forest Labs launched Flux 3 Image with localized multi-step editing, 4K output, and licensable weights.
[AINews] Pi 1.0, Pi Durable, and AIE NYC
Latent Space recaps Pi 1.0 and Pi Durable plus Gemini 4 Argon, GPT-6.1 Sol, FLUX 3, and other AI launches.
Post-Training Frontier Text-to-Image Models by Composing Preference and Rubric Rewards
Composing preference and rubric rewards lifts post-trained Flux2dev and Ideogram-4 on the Arena text-to-image leaderboard.
Unifying Distributional Training for One-Step Visual Generation
MGFlow unifies one-step visual training and beats FD-Loss on ImageNet and four-step FLUX.2.
AutoRef: Harness Optimization for Agentic Multi-Reference Image Generation
AutoRef automatically rewrites image-generation harnesses, lifting FLUX.2 klein 4B from 5.72 to 7.37 on MultiBanana.
Black Forest Labs Releases FLUX 3 Action: A 7B Open-Weights World Action Model That Tops RoboLab-120
Black Forest Labs released FLUX 3 Action, a 7B open-weights robot model topping RoboLab-120 at 42.92%.
6
Black Forest Labs launches FLUX 3 Action, an open robotics AI model
Black Forest Labs released FLUX 3 Action, a 7B open robotics model topping RoboLab-120.
Paint-Anything: Unified Any-Color Control for Image Generation and Editing
Paint-Anything learns a shared hex-prompt interface for any-color image generation and editing on FLUX.2-4B, improving ACBench scores by 85.3% and 28.3%.
TransNormal-2: Geometry-Grounded Rectified Flow with Edge-Aware Decoding for Precise Normal Estimation
TransNormal-2 improves monocular surface-normal estimation by fixing VAE edge degradation with geometry-aware losses and refinement, matching MoGe-2 with 1.4% of annotations.