Üretim4
görsel3D·reddit-stablediffusion·5 saat önce
Quick update on Continuity, my ComfyUI node pack for video and stills. A lot has landed since I last posted at 3.0: Screens (WIP, on main): put a picture or clip on a phone, laptop or TV in the shot. The model only sees a placeholder, then Continuity tracks it through every frame and composites your file on. It's still a work in progress, so expect some shots where the tracking slips or the edges aren't clean. Feedback and failure cases are very welcome. Image to 3D (3.2): turn a picture into a textured mesh from the tools dashboard (Pixal3D / TRELLIS.2). Qwen Image 2.1: a new model family…
görsel·reddit-stablediffusion·3 saat önce
40 year tech gap? No problem! On My Tandy 1000 TL/3 (10 MHz 286) I can now ask for a picture and view it a few seconds later. This is Desk Mind, a DOS program that talks over PicoMem WiFi to a small Python server on my PC. The server talks to Krea 2 in ComfyUI on my (4090), has a smoking fast ninfer Qwen3.8-27B (5090) rewrite the prompt, and then does the hard part: making a modern image look good in the Tandy's fixed 16-colour RGBI palette at 640x200 (an unsupported graphics mode BTW). - Prompt enhancement tuned for dithering. Qwen rewrites "a cat in space" into a prompt asking for one large…
görselmodel·arxiv-cv·20 saat önce
We present Multimodal Flow, a fully continuous generative model of language and vision. Most unified multimodal models either model both language and quantized images as discrete tokens or combine discrete language prediction with continuous image generation. The former introduces a visual quantization bottleneck. The latter requires modality-dependent objectives and sampling procedures. Fully continuous modeling avoids these trade-offs and enables a shared generative process, but remains underexplored for multimodal pretraining. Multimodal Flow introduces a unified continuous architecture…
modelvideo·arxiv-cv·20 saat önce
Video world models are expected to predict how the physical world evolves, yet they often produce visually plausible videos that violate basic physical principles. Existing approaches commonly assume that natural language is insufficient to represent the physical knowledge required for reliable generation, and therefore introduce additional visual, latent, numerical, or planning-based signals. We revisit this assumption and introduce Physis-Lang, a self-evolving framework that treats physical language as a shared and optimizable representation across data curation, model training, and video…
Modeller & Araştırma4
↑ yükselişte · yenidenHN'de GPT-6.1 0→8 hikâyemodel·openai·2 gün önce
Meet GPT-6.1 Sol: near-Astra intelligence for coding, computer use, and professional work at one-fifth of Astra’s standard API input and output token prices.
↑ yükseliştemodel·hf-papers·2 gün önce·158 oy
Modern multimodal models bring generation and understanding into a single unified system, which enables them to provide and learn from their own feedback. Motivated by this unified capacity, we introduce UniEvo-VL, a self-evolving framework for multimodal models to learn from this constructive self-correction feedback during test-time compute. Instead of relying on a separate, often larger, teacher, we leverage their self-critiques as privileged information and ask a single multimodal model to act as both teacher and student with different contexts. The student only sees the vanilla question…
↑ yükselişte84 oy/saataraçmodelşirket·hackernews·18 saat önce·1495 oy
model·hf-papers·2 gün önce·69 oy
Terminal agents act through stochastic model generations, yet the ability to generate a useful action does not ensure its reliable execution. A poor command (e.g., wrong package install) can change the environment in ways that hinder subsequent progress, even when the model could generate a better alternative. We investigate whether allocating test-time compute at the model-harness boundary can improve action reliability and trajectory success, and what makes this allocation effective. To study these questions, we introduce Mid-Harness, which samples and verifies candidate actions before…