SpatialEdit just dropped as a fully open-source 16B model that gives you fine-grained geometric control over 2D images. Beyond shifting objects and rotating their angles, it lets you tweak camera yaw, pitch, and zoom with remarkable precision. The catch? The 32 GB weights mean you'll need a beefy GPU to run it locally.
Researchers just introduced Floorplan Markup Language (FML), a structured representation that turns vector floor plan design into a next-token prediction task. Think of it like HTML for houses: rooms, walls, and doors are encoded into tags so a Transformer can generate adaptable layouts. For now, it's just a CVPR paper with no open code yet.
OpenAI's next-gen image model, GPT-Image-2, has reportedly leaked onto the Arena platform under several tape-themed aliases. Developer Pieter Levels noted the mystery models demonstrate exceptionally strong world knowledge and text rendering, potentially beating Nano Banana Pro. OpenAI has yet to officially confirm the testing.
Researchers just unveiled Generative World Renderer, an AI system trained on 4 million continuous frames of G-buffer data extracted from AAA titles like Black Myth: Wukong. It lets creators restyle game environments, tweak lighting, and change weather using natural language prompts. Why it matters: by tapping into real geometry and normals rather than pure pixels, game visual overhauls just became wildly controllable.
Alibaba just launched Wan2.7-Image, a unified model that combines image generation, editing, and understanding. It renders print-quality text across 12 languages, offers precise facial and HEX color controls, and outputs up to 12 consistent images per prompt, solving a major headache for multi-panel creators.
Moonlake just unveiled a world-building agent that operates directly inside Blender. Instead of one-shot asset generation in chat boxes, it clicks, edits meshes, and adjusts lighting across long-horizon tasks while learning from expert workflows. Why it matters: It shifts 3D AI from isolated generation to executing end-to-end production pipelines in real software.
Alibaba just rolled out Wan 2.7 Image, a unified model that handles both generation and precision editing. It focuses on lifelike human faces with natural variation, lets creators lock brand palettes using up to eight hex codes, and renders multi-language text up to 3,000 tokens. Why it matters: Pumping out up to 12 consistent shots in a single prompt makes storyboards and marketing assets far easier to build.
Researchers just open-sourced Gen-Searcher, a search-augmented agent framework that scours the web for reference images and facts before generating visuals. By grounding diffusion models in real-time retrieval, it tackles visual hallucinations on obscure entities across both open and closed models.
ByteDance just introduced DreamLite, a compact 0.39B diffusion model that handles both text-to-image generation and image editing directly on smartphones. It can generate a 1024×1024 image in around 3 seconds on an iPhone, swap styles, and remove objects completely offline. While fine details still lag behind massive server models, it brings fast, private image manipulation right to local devices.
General Motors is embedding generative AI into vehicle design to slash timelines from months to hours. Partnering with startup Vizcom, GM turns hand-drawn sketches into 360-degree models and animations in under a day, while an AI virtual wind tunnel delivers aerodynamic drag estimates in real time instead of two weeks. GM says human designers still set the creative vision.
Bambu Lab's MakerWorld just integrated AI generator Meshy into its MakerLab hub, allowing users to turn any 2D photo into a print-ready 3D file in minutes. Driven by the Meshy-6 engine, the tool automatically maps colors to filament slots for Bambu's AMS multi-material system, skipping manual painting inside slicing software. It costs 2 MakerLab credits per export and bridges a major gap for printer owners without CAD skills.
Runway just launched Ad Concepter, a dedicated web app that turns a single prompt, a reference image, and a product shot into ready-to-test advertising ideas. Marketers and creators can now explore compositions and story beats without piecing together complex generative workflows.
Meta just dropped SAM 3.1, adding a shared-memory approach called Object Multiplex to its visual segmentation model. It delivers roughly 7x faster inference when tracking 128 objects on a single H100 without dropping accuracy, beating prior scores on 6 out of 7 video benchmarks. The model checkpoints are available to request on Hugging Face.
Babylon.js just launched version 9.0, calling it the framework's biggest update yet. The release introduces Clustered Lighting to render thousands of lights smoothly across WebGPU and WebGL 2, alongside a Frame Graph pipeline that cuts GPU memory by up to 40%. It also rolls out a Node Particle Editor and expands Gaussian Splat support built with Adobe.
Researchers just introduced PSDesigner, an automated framework that turns text prompts into multi-layered Photoshop designs instead of flat images. A specialized multi-agent workflow collects assets, plans layouts, and iteratively refines the canvas through tool execution. Why it matters: It bridges the frustrating gap between one-shot AI generation and professional PSD editing workflows.
Real Restorer just launched as a specialized open-source model designed to rescue ruined photos, from wiping out glare and noise to restoring scratched black-and-white shots. It matches top closed-source models in benchmarks, but you will need a beefy GPU to host its 42 GB weight package locally.
ComfyUI just introduced dynamic VRAM, a feature that streams model layers on demand and unloads them immediately after execution. Instead of crashing low-VRAM setups, mid-tier cards like the RTX 5060 can now run heavyweight models like Flux 2dev with render times cut in half. Running frontier generation locally just became a lot more accessible.
Lovart just launched Move Object, letting users reposition elements in an image without messing with masks or layers. Grab a lasso or box selection, drag the object anywhere, and optionally add a prompt tweak to get a clean, seamless output. It basically kills the annoying loop of manual inpainting and constant re-rolls.
RealMaster can convert 3D rendered game footage into photorealistic video without messing up underlying geometry. Unlike competing editors that distort details, it preserves original shapes, shadows, and motion logic, making it equally useful for upgrading synthetic autonomous driving simulation datasets.
Figma just launched its MCP server in beta, letting coding agents like Claude Code and Codex design directly on the canvas using your team's real components and design tokens. Teams can also steer agents with Markdown-based "skills" to prevent off-brand AI output. It is free during the beta, with usage-based pricing coming later.
