I Built a Latent Diffusion Model That Fits in 5.5MB and Runs Without a GPU
TinyDiffusion 128 is a 1.45M-parameter Diffusion Transformer that runs on a plain CPU. Here are the ablations, benchmarks, and the caveats.
#Image Generation
708 articles tagged with Image Generation
🧠 Ho scritto “Agents Read More Than They Use” e sviluppato #tracetrim: imparare dalle tracce quali campi dei tool passar...
🧠 Ho scritto “Agents Read More Than They Use” e sviluppato #tracetrim: imparare dalle tracce quali campi dei tool passare agli agenti #AI, misurando il risparmio e dove il filtro s...
🎨 Where open-source stands in image generation:#15 FLUX 3 Image — the best model you can actually download142 ELO points...
🎨 Where open-source stands in image generation:#15 FLUX 3 Image — the best model you can actually download142 ELO points behind GPT Image 2.6 Sunburst (max) (OpenAI)Blind human pre...
I trained a diffusion model and put it in a stylesheet.No JavaScript. Press the button and CSS turns a random seed into ...
I trained a diffusion model and put it in a stylesheet.No JavaScript. Press the button and CSS turns a random seed into noise, then denoises it into a 16×16 sprite in 12 steps. 202...
Adafruit、AI基板設計のFluxが示す利用者110万人を虚偽広告と主張 https://fed.brid.gy/r/https://fabscene.com/new/news/adafruit-flux-lawsuit-amende...
Adafruit、AI基板設計のFluxが示す利用者110万人を虚偽広告と主張 https://fed.brid.gy/r/https://fabscene.com/new/news/adafruit-flux-lawsuit-amended-complaint-false-advertising/?utm_source=rss&utm_medium=rss...
Stop Dirty Datacenters 2026: contributi locali e internazionali dalle lotte@0 https://radioblackout.org/2026/11/stop-dir...
Stop Dirty Datacenters 2026: contributi locali e internazionali dalle lotte@0 https://radioblackout.org/2026/11/stop-dirty-datacenters-2026-contributi-locali-e-internazionali-dalle...
Creativity is more important than productivity.Make your own art.#AI #GenAI #Midjourney #StableDiffusion
Creativity is more important than productivity.Make your own art.#AI #GenAI #Midjourney #StableDiffusion
🎨 Where open-source stands in image generation:#13 FLUX 3 Image — the best model you can actually download110 ELO points...
🎨 Where open-source stands in image generation:#13 FLUX 3 Image — the best model you can actually download110 ELO points behind GPT Image 2.5 Sunburst (max) (OpenAI)Blind human pre...
NVIDIA Developer: Speed Up Image Generation with FLUX 3 Image Klein on NVIDIA Blackwell GPUshttps://ai-news.ypetit.net/?...
NVIDIA Developer: Speed Up Image Generation with FLUX 3 Image Klein on NVIDIA Blackwell GPUshttps://ai-news.ypetit.net/?a=0olt8au1qj6&utm_source=mastodon&utm_medium=social&utm_camp...
How I Automated 3D Game Assets with FLUX, Blender and the Roblox API
The problem I build Roblox games on my own, and good-looking 3D models were always my...
Show HN: A decent text-to-image model trained for ~$20K (Apache 2.0 Weights)https://www.linum.ai/field-notes/pjit-21k#ai...
Show HN: A decent text-to-image model trained for ~$20K (Apache 2.0 Weights)https://www.linum.ai/field-notes/pjit-21k#ai #apache
DawidNezo/local-image-gen-benchmark: CLI for benchmarking text-to-image diffusion models locally with Hugging Face Diffusers: load and generation time, throughput, CPU/RAM/GPU usage.
CLI for benchmarking text-to-image diffusion models locally with Hugging Face Diffusers: load and generation time, throughput, CPU/RAM/GPU usage.
fujc683/gpt-image-2-6-alternatives: GPT Image 2.6 vs FLUX 2, Seedream 5.0 and Nano Banana 2.
GPT Image 2.6 vs FLUX 2, Seedream 5.0 and Nano Banana 2.
Google cambia le regole per l'AI a scuola: per gli account degli studenti K-12 gestiti dalle scuole, Gemini, Gemini Note...
Google cambia le regole per l'AI a scuola: per gli account degli studenti K-12 gestiti dalle scuole, Gemini, Gemini Notebook e Gemini in Classroom saranno disattivati di default e ...
🎨 Where open-source stands in image editing:#15 FLUX 3 Image — the best model you can actually download76 ELO points beh...
🎨 Where open-source stands in image editing:#15 FLUX 3 Image — the best model you can actually download76 ELO points behind GPT Image 2.5 Sunburst (max) (OpenAI)Blind human prefere...
Campfire7/graphis: Uncensored text-to-image, entirely on your Mac. A native macOS app that runs MOSS-Image locally with mflux and MLX.
Uncensored text-to-image, entirely on your Mac. A native macOS app that runs MOSS-Image locally with mflux and MLX.
Show HN: A decent text-to-image model trained for ~$20K (Apache 2.0 Weights)
Show HN: A decent text-to-image model trained for ~$20K (Apache 2.0 Weights)
picontapLab/prompt-seen: Open-source AI image prompt library with tested GPT-Image-2.5, GPT-Image-2.6 examples and reusable templates to adapt for Nano Banana, FLUX, Midjourney and Stable Diffusion. Discover prompts for image generation, editing, product photography, posters, portraits and illustration.
Open-source AI image prompt library with tested GPT-Image-2.5, GPT-Image-2.6 examples and reusable templates to adapt for Nano Banana, FLUX, Midjourney and Stable Diffusion. Discov...
Linum-AI/pjit-21k: 2B-parameter pixel-space text-to-image model (Pyramid-JiT) trained for $21K
2B-parameter pixel-space text-to-image model (Pyramid-JiT) trained for $21K
ml-postech/EPIC: Official Implementation of EPIC: Efficient Predicate-Guided Inference-Time Control for Compositional Text-to-Image Generation [NeurIPS 2026]
Official Implementation of EPIC: Efficient Predicate-Guided Inference-Time Control for Compositional Text-to-Image Generation [NeurIPS 2026]
🎨 Where open-source stands in video generation:#3 FLUX 3 Max — the best model you can actually download7 ELO points behi...
🎨 Where open-source stands in video generation:#3 FLUX 3 Max — the best model you can actually download7 ELO points behind HeyGen Video 1 (HeyGen)Blind human preference, not market...
DreamX: Video Diffusion Model is a Universal Latent Action Learner
Learning a universal action representation from video requires modeling dynamics that are consistent across different executions of the same action yet invariant to changes in embo...
"Revolutionize Multimodal Modeling with LARC: Unleash Powerful Reasoning for Text-to-Image Generation."https://telegra.p...
"Revolutionize Multimodal Modeling with LARC: Unleash Powerful Reasoning for Text-to-Image Generation."https://telegra.ph/Technomantic-Breakthrough-Liquid-Neural-Network-Continuous...
JayLee1002/ALD-Font: Official Implementation of ``Disentangle to create: Architecture-level disentangled diffusion model for few-shot Chinese font generation''
Official Implementation of ``Disentangle to create: Architecture-level disentangled diffusion model for few-shot Chinese font generation''