Even_AdderCameraNoise: Enabling Faithful Camera Control in Video Diffusion through Geometry-Flow-Guided Noise Warping
RayDer: Scalable Self-Supervised Novel View Synthesis from Real-World Video
RayDer: Scalable Self-Supervised Novel View Synthesis from Real-World Video - Divisions by zero
# Abstract >Self-supervised novel view synthesis (NVS) remains challenging to scale, despite the abundance of video data, largely due to the brittleness of training on realistic videos and the hard-to-predict scaling behavior of multi-network system designs. We introduce RayDer, a unified, feed-forward transformer that consolidates camera estimation, scene reconstruction, and rendering into a single backbone, turning self-supervised NVS into a well-posed single-model scaling problem. A minimal dynamic state, treated as a nuisance factor, absorbs time-varying content and enables stable training on unconstrained real-world video. Importantly, RayDer keeps static-scene NVS as its target task: dynamic content is leveraged purely as scalable supervision, not reconstructed as in dynamic-scene (4D) NVS. Across multiple model sizes and orders of magnitude in data, RayDer exhibits clean power-law scaling with data and compute, and outperforms static-scene data mixtures. On a large number of benchmarks, RayDer achieves strong zero-shot open-set performance competitive with state-of-the-art supervised approaches. Project Page: this https URL [https://compvis.github.io/rayder] Paper: https://arxiv.org/abs/2605.31535 [https://arxiv.org/abs/2605.31535] Code: https://github.com/CompVis/rayder [https://github.com/CompVis/rayder] Weights: https://huggingface.co/CompVis/rayder [https://huggingface.co/CompVis/rayder] Project Page: https://compvis.github.io/rayder [https://compvis.github.io/rayder] Your browser does not support playing HTML5 video. You can download a copy of the video file [https://compvis.github.io/rayder/static/videos/video_grid_2_h264.mp4] instead.
Bringing Native Support for 3D Gaussian Splats into ComfyUI with TripoSplat
infinition/Pixal3D-pipeline: A batch automation layer that turns images into 3D assets (.glb) with no manual intervention.
nvidia/Cosmos3-Super-Text2Image
nvidia/Cosmos3-Super-Text2Image - Divisions by zero
# Abstract >We introduce Cosmos 3, a family of omnimodal world models designed to jointly process and generate lan- guage, image, video, audio, and action sequences within a unified mixture-of-transformers architecture. By supporting highly flexible input-output configurations, Cosmos 3 seamlessly unifies critical modalities for Physical AI—effectively subsuming vision-language models, video generators, world simulators, and world-action models into a single framework. Our evaluation demonstrates that Cosmos 3 establishes a new state-of-the-art across a diverse suite of understanding and generation tasks, demonstrating omnimodal world models as scalable, general-purpose backbones for embodied agents. Our post-trained Cosmos 3 models were ranked as the best open-source Text-to-Image and Image-to-Video models by Arti- ficial Analysis, and the best policy model by RoboArena at the time the technical report was written. To accelerate open research and deployment in Physical AI, we make our code, model checkpoints, curated synthetic datasets, and evaluation benchmark available under the Linux Foundation’s OpenMDW-1.1 License at github.com/nvidia/cosmos [http://github.com/nvidia/cosmos] and huggingface.co/collections/nvidia/cosmos3 [http://huggingface.co/collections/nvidia/cosmos3] . The project website is available at research.nvidia.com/labs/cosmos-lab/cosmos3 [http://research.nvidia.com/labs/cosmos-lab/cosmos3] . Paper: https://research.nvidia.com/labs/cosmos-lab/cosmos3/technical-report.pdf [https://research.nvidia.com/labs/cosmos-lab/cosmos3/technical-report.pdf] Code: https://github.com/nvidia/cosmos [https://github.com/nvidia/cosmos] Model Collection: https://huggingface.co/collections/nvidia/cosmos3 [https://huggingface.co/collections/nvidia/cosmos3] Project Page: https://research.nvidia.com/labs/cosmos-lab/cosmos3/ [https://research.nvidia.com/labs/cosmos-lab/cosmos3/]
FLUX.2 [klein] Schematic LoRA
FLUX.2 [klein] Schematic LoRA - Divisions by zero
Model: https://huggingface.co/nomadoor/flux-2-klein-9B-schematic-lora [https://huggingface.co/nomadoor/flux-2-klein-9B-schematic-lora] Dataset: https://huggingface.co/datasets/nomadoor/flux-2-klein-9B-schematic-dataset [https://huggingface.co/datasets/nomadoor/flux-2-klein-9B-schematic-dataset]
Bernini: Latent Semantic Planning for Video Diffusion
Bernini: Latent Semantic Planning for Video Diffusion - Divisions by zero
# Abstract >Multimodal large language models (MLLMs) and diffusion models have each reached remarkable maturity: MLLMs excel at reasoning over heterogeneous multimodal inputs with strong semantic grounding, while diffusion models synthesize images and videos with photorealistic fidelity. We argue that these two families can be unified through a simple division of labor: MLLMs perform semantic planning, while diffusion models render pixels from high-level semantic guidance and low-level visual features. Building on this idea, we propose Bernini, a unified framework for video generation and editing. An MLLM-based planner predicts the target semantic representation directly in the ViT embedding space, and a DiT-based renderer synthesizes pixels conditioned on this plan, augmented by text features and, for editing, source VAE features for detail preservation. Because semantics serve as the interface, the planner and renderer can be trained separately and only lightly co-trained, preserving the pretrained strengths of both components while keeping training efficient. To better handle multiple visual inputs, we introduce Segment-Aware 3D Rotary Positional Embedding (SA-3D RoPE), and further incorporate chain-of-thought reasoning in the planner to better transfer understanding into generation. Bernini achieves state-of-the-art performance across a wide range of video generation and editing benchmarks, with the MLLM’s pretrained understanding translating into strong generalization on challenging editing tasks. Paper: https://arxiv.org/abs/2605.22344 [https://arxiv.org/abs/2605.22344] Model: https://huggingface.co/ByteDance/Bernini/tree/main [https://huggingface.co/ByteDance/Bernini/tree/main] Project Page: https://bernini-ai.github.io/ [https://bernini-ai.github.io/] [https://lemmy.dbzer0.com/pictrs/image/068bf059-ab0d-4a44-bd78-21e06442b8cd.webp]
HabrAI-компаньон в проде на третьем месяце — 5 архитектурных решений и инфра-тюнинг
Каждый, кто пробовал собрать AI-чат по типовой схеме — chat-completions API, OpenAI Memory, один эндпоинт Stable Diffusion — рано или поздно упирается в одни и те же стены. Бот забывает разговор через десять реплик. Иногда сервер бодро отвечает HTTP 200, а внутри пустая строка: ни ошибки, ни таймаута, модель просто отказалась говорить и сделала это молча. Один и тот же запрос рисует двух разных персонажей. А одеть нарисованного персонажа в конкретное платье из каталога не получается вообще.
https://habr.com/ru/articles/1042280/
#ИИ #LLM #OpenRouter #Stable_Diffusion #LoRA #ComfyUI #Redis #ChromaDB #Python #чатботы
quinteroac/ComfyUI-AnimaFastTrain: Experimental Custom Node for Comfyui to Fast Training Anima Model in Memory
JpAndreBTA/Nexus-BTA: A local AI image, video, workflow and 3D experiment studio built around an embedded ComfyUI runtime
