Tugg Speedman

SOMEONE VIBE CODED A TOOL WITH CLAUDE CODE THAT TURNS ANY PHOTO OR VIDEO INTO A CINEMATIC 3D SHOT WITH A MOVING CAMERA instead of a flat, static image, it adds a subtle 3d camera drift so the scene feels like a real cinematic shot pulling through it > drop in any photo or video and it instantly gets depth, like a real 3d scene > customizable camera controls, tilt, pan, zoom, roll, fov, so you can frame the shot exactly how you want > keyframe the whole thing on a timeline, set a start and end and it animates the camera move between them > add blur and depth of field, so parts of the image go soft like an actual camera lens > then export the finished cinematic clip, all in the browser its great for intros, product shots, or making any clip look way more premium doing this by hand sucks and takes hours, youre manually cutting layers, animating camera moves, and faking depth in after effects this takes that exact shot and turns it into a drag and drop

Neural Rendering (video to video) is very much the future of controllable story telling. Block out the motion with Cartwheel Swing in the Unreal Engine and then render with your favorite video model (we use seedance 2.0 here). The great thing is you can now fully edit the performance of each character and just re-render when you're ready. 3d motion IS the way forward, avoiding the AI gambling or painful reprompting to try and pull the performance out of the model

We just killed Higgsfield. You can now use Seedance 2.0 directly from Claude Code or Codex. No subscription required. Higgsfield: $129/mo, limited credits. Monid: $0.04/10K tokens, pay as you go.

Building PhoneClaw with GPT-5.6 in Codex this week it’s a mobile-native AI agent framework: local models, native iOS skills, permissions, voice/image/live surfaces, and real phone actions trying to make phones feel less like chat windows and more like agent runtimes github.com/kellyvv/PhoneC… devpost.com/software/phone…

Lucy 2.5 Realtime is now live on fal! Real-time video-to-video editing over WebRTC, from @DecartAI. - Restyle, swap backgrounds, and add, replace, or remove objects live on a webcam or streamed feed as it plays - Swap characters, change color and texture, and drop in VFX like fireballs, explosions, and smoke - Reference-image character-swap and virtual try-on at interactive latency - Sharper edit isolation and prompt adherence on small, off-center, non-human objects

【GPT 5.6 SOL × Blender × Seedance！「AI監督」が創る次世代の映像制作パイプライン】 👉 x.com/aiwarper/statu… 単一の曖昧な指示から「企画・構図設計・動画生成」までをワンストップで完結させる、極めて実用的なAI映像制作ワークフローが登場しました！🎬 この仕組みの面白いところは、LLM（GPT 5.6 SOL）が緻密に設計したカメラワークやカット割りのデータを、3DCGソフト「Blender」に落とし込んで空間検証（ブロッキングという簡易的な構図確認の工程）を挟んでいる点です。これによって、動画生成AI特有の「思い通りの絵が出ないガチャ」を回避し、狙い通りのカメラワークを確実に表現できるようになります！✨ 最終的な映像出力には、ByteDanceの強力な動画生成モデル「Seedance」を使用。プロンプト1つで、カメラアングルから演出タイミングまで完璧に制御されたプロレベルのアニメ演出動画が生成される仕組みは本当に革新的ですね！ 本来は多くのクリエイターで分担していた「脚本・3D設計・映像合成」という三層の工程を一人でコントロールできるこのパイプライン、個人の制作能力を劇的に引き上げる素晴らしい技術検証です！🚀 #AI映像制作 #Blender

Generates high-fidelity 3D facial animation from an audio source with accurate lip-sync and nuanced emotional expression. github.com/NVIDIA/Audio2F…

holy sht.. AI has automated Adobe After Effects you can now install this plugin and talk to agents to create motion graphic with layers, everything is customizable..

Google engineer explained how to fine-tune a tiny LLM from 46% to 90% accuracy on your phone in 21 minutes - better than $1500 on-device AI bootcamps. pick Gemma 270M -> generate synthetic task data -> fine-tune with LoRA -> quantize to int4 -> deploy to Pixel and hit 2000 tokens per second. That loop is how a 270M model beats a 70B one on your task, running fully offline in your pocket. Gemma 270M + synthetic data + LoRA + int4 quantization + on-device runtime - that's the stack. Watch and save it, then fine-tune your own tiny agent tonight.

Kimi K3 is looking absurdly strong in frontend design. The feature set may be similar, but Kimi seems to have better visual judgment, cleaner composition, and noticeably stronger 3D execution. The final result simply feels more polished and intentional. An open model performing at this level is a very big deal for the entire ecosystem. Cc: @filicroval

THE AI STOPPED TALKING TO THE ENGINE THROUGH A BRIDGE, IT MOVED INSIDE IT. for a year the pattern was the same. a model drives Blender or Godot through MCP. a socket relaying tool calls in from outside. this is different. the AI lives in the engine itself. one conversation builds a quiet, realistic bedroom scene from nothing. whitebox layout. asset import. materials. lighting. post. light bake. collision. all the things you used to click through panel by panel, now just described. no tool-switching. no external MCP. the whole workflow grew inside the editor. the speed isn't the story. an indie dev doing that prototype by hand is half a day, sure. the story is that native engine ops stopped being manual. the model isn't reaching in through a pipe, it's holding the controls directly. but a bedroom scene sitting still is the easy read. a screenshot can't tell you if the lighting holds up in motion, if collision feels right when you actually walk it. setup collapsed. the engine work you can describe just got cheap. deciding whether it plays right is still the part nobody can prompt.

yeah disney my thinking is literally the same as yours generative models were not about replacing creation, the whole point is making creation faster. auto rigging and motion gen that you can actually reuse, not one off outputs. been saying this for weeks and you just shipped it first as a blender plugin lol via disney research zurich / eth zurich, siggraph talks 26

No sound department? No problem. Our new Foley LoRA built on LTX-2.3 adds synced sound design to silent footage. Footsteps, impacts, materials, and ambience layered in to match the action. No music bed, no dialogue, ready to drop straight into your mix.

Highly-recommended overview of self-improving agentic systems. (bookmark it) Self-improving agents are moving from research demos into deployed systems. This survey frames a modern agent as a foundation model coupled with an operational scaffold, then formalizes self-improvement as a self-induced update operator that commits changes to either. Paper: arxiv.org/abs/2607.13104 Learn to build effective AI agents in our academy: academy.dair.ai

MoLingo, a text-to-motion (T2M) model that generates realistic, lifelike human motion by denoising in a continuous latent space. Recent works perform latent space diffusion, either on the whole latent at once or auto-regressively over multiple latents. This paper studies how to make diffusion on continuous motion latents work best. Paper Title: MoLingo: Motion-Language Alignment for Text-to-Human Motion Generation Project: hynann.github.io/molingo/MoLing…. Link: arxiv.org/abs/2512.13840

China open-sourced a model that reconstructs any scene in 3D from a regular video, in real-time. one camera. no LiDAR. 10,000+ frames without falling apart. just walk around with your camera and watch the entire world get rebuilt in 3D at 20 fps. → runs at ~20 FPS on a single GPU → Stable over 10,000+ frames → Beats optimization-based methods on benchmarks → Works on drone footage, driving videos, indoor walkthroughs 100% open source.

A guy in China plugged an LLM into Karpathy method + Claude Code and built a second brain that remembers everything forever. Not a note app, not RAG just 3 folders on a laptop. Folder 1 holds raw sources: articles, journal entries, chat logs. The LLM reads them but never touches them. Folder 2 is the wiki hundreds of markdown pages the LLM writes and maintains itself, where every person, idea and decision gets filed, cross-referenced and kept current. Folder 3 is one file: CLAUDEmd, the rulebook that turns a chatbot into a librarian. He drops in a new article and Claude reads it, updates 10-15 pages, flags what contradicts his old notes, logs the timestamp while he watches his memory grow node by node in Obsidian graph view. Ask it anything and it doesn’t dig through raw files like ChatGPT does the synthesis already exists, the connections are already drawn. The method comes from a gist Karpathy posted on April 4, now sitting at 5,000 stars and 5,000 forks, with 1 production team already scaled to 4,000 interlinked concepts in 6 months. The part nobody mentions: Vannevar Bush described this exact machine in 1945 the Memex, a personal memory store with trails between documents. Nobody built it for 81 years because someone had to do the maintenance. Now maintenance costs nothing the LLM does the filing while the human does the thinking. Every article makes the brain smarter, and it never forgets. His memory has a git history now.

i can't tell if i should love or be terrified of this startup lol > their end goal is the complete eradication of mosquitoes > they're building a $50/month pet drone that guards your home 24/7 so you never deal with a mosquito again > it listens for mosquito wingbeats, locks on, then shreds them mid-air with its propellers > every insect's wingbeat sounds slightly different, so it can tell a mosquito apart from a fly or a wasp > by their math, 10 drones keep a full square kilometer mosquito-free > the video below is their first ever air-to-air kill

Kimi K3’s performance here is honestly absurd. Purely speculating, but results like this make it feel as if the team studied the strongest models in the field, especially Fable 5, and pushed those lessons as far as possible. For an early output, this is an incredibly strong showing. Cc: @chetaslua

🚨Architects are going to hate this. Someone just open-sourced a full 3D building editor that runs entirely in your browser. No AutoCAD. No Revit. No $5,000/year licenses. It's called Pascal Editor. Built with React Three Fiber and WebGPU -- meaning it renders directly on your GPU at near-native speed. Here's what's inside this thing: → A full building/level/wall/zone hierarchy you can edit in real time → An ECS-style architecture where every object updates through GPU-powered systems → Zustand state management with full undo/redo built in → Next.js frontend so it deploys as a web app, not a desktop install → Dirty node tracking -- only re-renders what changed, not the whole scene Here's the wildest part: You can stack, explode, or solo individual building levels. Select a zone, drag a wall, reshape a slab -- all in 3D, all in the browser. Architecture firms pay $50K+ per seat for BIM software that does this workflow. This is free. 100% Open Source.