Bone-Studio

Creating this together with the @iesebschool team has been a wonderful adventure and an honor! #IESENYC #IESEGAR #VirtualReality #vr #photogrammetry @RealityCapture_ @_3Dflow_ #zephyr @b3d #VisualInnovation #BoneStudio @UnrealEngine #newyork #madrid #barcelona

Creer que porque estás usando IA ya eres alguien sin habilidad, no has dibujado en tu vida, etc. Te tienes que reir. Muchos hemos dibujado (y modelado en 3D, etc.) durante muchos años (y profesionalmente) y estamos probando nuevas tecnologías. ¿Se entiende ya el asunto?

Hace 2 días presentamos MatPlus en el canal de @blendtuts con @Odin_3D y @BoneStudioAnim. El directo ya está disponible.: youtube.com/live/a68Q7o6_o… Gracias por el espacio y el debate técnico. #Blender3D #b3d #MatPlus #Texturing #substancepainter

@TencentHunyuan Sadly as long as you don’t change your open source license to allow EU, UK and South Korea, this pass as nothing interesting, even when the tech behind is amazing I’m sure But your license makes it useless for a pretty big part of the world. Embrace real Apache 2.0 or MIT please

Today, we introduce HunyuanImage 3.0-Instruct, a native multimodal model focusing on image-editing by integrating visual understanding with precise image synthesis! 🚀 It understands input images and reasons before generating images. Built on an 80B-parameter MoE architecture (13B activated), it natively unifies deep multimodal comprehension and high-fidelity generation. 🧠 A "Thinking" Model with Native CoT & MixGRPO: The model doesn’t just execute commands, it processes them through a Native Chain-of-Thought (CoT) schema. Enhanced by our self-developed MixGRPO algorithm, it reasons through complex instructions to achieve flawless intent alignment and human-preference consistency. 🎨 Precise Editing & Multi-Image Fusion: The model enables accurate image editing by adding, removing, or modifying elements while keeping non-target areas perfectly intact. It also excels at seamless multi-image fusion, synthesizing complex scenes by extracting and blending elements from multiple sources into a unified, consistent output. 🏆 SOTA Performance: HunyuanImage 3.0-Instruct sets a new benchmark in visual quality and alignment, delivering performance that matches leading proprietary models. We aim to enable the community to explore new ideas with a state-of-the-art foundation model, fostering a dynamic and vibrant image generation ecosystem. 🛠️🎨 💻Try it at (PC only): hunyuan.tencent.com/chat/HunyuanDe…

Si eres de los que empiezan a sentirse un poco estafados por los malos consejos anti-IA de tu artista favorito y empiezas a pensar que de esa manera no te vas a ganar la vida como artista, échale un ojo a la colleja de realidad que nos da @BoneStudioAnim youtube.com/live/m2aqfh0Ms…

@BrentLynch @Alibaba_Wan Yeah, but it seems no local/offline open source model, so right now it’s a no go sadly

@Alibaba_Wan Is pretty rad! x.com/BrentLynch/sta…

Today, we're officially launching Wan2.5-Preview! It's set to reshape the future of visual generation with a new architecture and powerful features. • Architectural Features: Native Multimodality, Deep Alignment ∘ Native Multimodal Architecture: Adopts a new, unified framework for both understanding and generation, flexibly supporting the input and output of text, images, video, and audio. ∘ Joint Multimodal Training: Achieves stronger modal alignment by jointly training on text, audio, and visual data—key to enabling audio-visual sync and greatly improved instruction following. ∘ Human Preference Alignment: Implements Reinforcement Learning from Human Feedback (RLHF) to continuously align with human preferences, enhancing image quality and video dynamics. • Video Capabilities: A/V Synchronization, Cinematic Quality ∘ Synchronized A/V Generation: Natively supports high-fidelity, high-consistency video generation with synchronized audio, including multi-person vocals, sound effects, and BGM. ∘ Controllable Multimodal Input: Supports text, images, and audio as input sources for limitless creativity. ∘ Cinematic Aesthetics: Features powerful dynamics and structural stability with an upgraded cinematic control system, generating 1080p HD 10s videos of cinematic quality. • Image Capabilities: Creative & Precise Control ∘ Advanced Image Generation: Greatly improved instruction following to support photorealistic quality, diverse artistic styles, creative typography, and professional-grade charts. ∘ Image Editing: Supports conversational, instruction-based image editing and pixel-level precision for tasks like multi-concept fusion, material transformation, and product color swapping, and more.

@Alibaba_Wan Do we have a model to download and use it locally offline? That’s what we need.

@giffboake @Alibaba_Wan And that is why I ask for the model to be native 1080p, something perfectly possible.

The point is if the current image models for a single image are nowhere in the same universe as 5s you ain't getting a video model to do that (at any decent resolution) for a very long time. Also you can't generate and length of video at 1080p with any amount of VRAM these are all 720p models

The wait is almost over! The WAN 2.5-Preview is coming. Take a look at our incredible lineup and tell us: Who are you most excited to see?

@giffboake @Alibaba_Wan I think that you believe I mean the model taking 5 seconds for the generation, I don’t mean that, it’s to be able to generate 10 seconds video in the amount of time it could require.

@giffboake @Alibaba_Wan It’s not silly, 5 seconds was a limit unreachable locally before, correct hands were impossible before, as many other things it’s perfectly possible with development. 2.2 is perfectly runnable yes, but I asked for 1080p in low vram, you cannot generate 5S at 1080p in 24Gb

@BoneStudioAnim @Alibaba_Wan Wan 2.2 14b is easily runnable locally on 24GB. Not even image models have 10s generations that's just silly

@Nekodificador Hay más eh, este solo es uno :)

@BoneStudioAnim PEEEEERDON?????

Ahora mismo daría medio brazo por poder usar una herramienta de gestión de ideas que funcionase con grafos multidimensionales. Esto se me empieza a quedar corto.

@Nekodificador ¿Has visto Think Machine? Thinkmachine.com

En realidad, un sistema de visualización 3D ya sería un paso enorme para organizar, y el resto de dimensiones se podrían representar de muchas otras formas, por ejemplo en la dimensión del tiempo, teniendo grafos con distintas posiciones según las dimensiones que estés visualizando. Factible es, desde luego.

@TecnoIA1 @TencentHunyuan Lee al principio de la licencia, excluyen EU y otros lugares, busca European Union en la licencia, en github

@BoneStudioAnim @TencentHunyuan ¿En que parte de su licencia lo especifican? Pasa la info

🚀 Hunyuan 3D 2.1 is here! The first fully open-source, production-ready PBR 3D generative model! ✅Cinema-grade visuals: PBR material synthesis brings leather, bronze, and more to life with stunning light interactions. ✅ Fully open-source: Model weights, training/inference code, data pipelines, and architecture—yours to fine-tune! ✅ Runs on consumer GPUs, empowering creators, devs, and small teams. Be the first to shape the future of 3D! Download now and build something epic. Model: huggingface.co/tencent/Hunyua… Demo: huggingface.co/spaces/tencent… Github: github.com/Tencent-Hunyua… Hunyuan 3D Creation Engine: 3d.hunyuan.tencent.com #Hunyuan3D #OpenSource #3DCreation

ComfyUI + Photoshop + @higgsfield_ai

@_BD3D Hey @_BD3D ! Cool nodes, I wonder what is the performance relative to native C++ nodes? That is what worries me the most in relation to this type of nodes.

What else would you want to be added to #b3d existing node editors? Give me ideas! #geometrynodes 🐍 Nodes with Custom Sockets colors are welcome ideas too.

@TXhunyuan Can you please clarify the license? Why do you leave European Union out? Please update the license for EU to full Apache 2 or MIT

We have deployed the accelerated models and inference code across all repositories. For instance, Hunyuan3D-2mv Turbo is now accessible in the Hunyuan3D-2mv repository here: huggingface.co/tencent/Hunyua…. If you find our models valuable, don’t forget to give it a star! 🌟

@Xianbao_QIAN @huggingface VRAM required? Also in the comfy node it says “remote server”, why?

StepFun Ti2V just got released on @huggingface - 30B - up to 102 frames, 5 s, 540P video - Both text and image control - Precise motion control Link below

@TXhunyuan Why banning European Union? And then you give a license to FreePik? Modify the license and make it a true Apache 2 or MIT please

We are thrilled to announce a ​30x acceleration in white model generation speed across the entire Hunyuan3D 2.0 family, reducing the processing time from 30 seconds to just ​1 second.