Vannarot Roeung
Recent posts by Vannarot Roeung
Trellis2.c: Local Image-to-3D With No Python
trellis2.c runs Microsoft's TRELLIS.2 image-to-3D model natively on CUDA or Vulkan with no Python, chaining four models into a full local pipeline from photo to rigged 3D asset.
Suno Hack Exposes Its AI Music Training Data
A hacker leaked Suno's training data manifests, revealing 2 million scraped YouTube clips plus Deezer, Genius and Pond5 catalogs, weeks before a July copyright ruling.
Real-Time AI Video Editing 2026: Lucy vs Aleph vs Local
Edit a video already playing? Decart Lucy 2.5. Refining a finished clip? Runway Aleph 2.0. A brand new shot? Seedance 2.5. No per-use cost? An open model in Velorn. Live-feed editing is now a third pillar of AI video.
Decart Lucy 2.5 Brings Real-Time AI Video Editing
Decart has launched Lucy 2.5, an upgrade to its real-time AI video model that edits live video from natural-language prompts with sub-40ms latency.
OpenAI Codex Now Encrypts Agent-to-Agent Instructions
OpenAI's Codex now encrypts the instructions its AI agents pass to one another, removing readable delegation text from local audit logs. Here is what changed and how developers can keep an audit trail.
Lemonade 11.0: AMD's Local AI Server Adds TTS and 3D
AMD's Lemonade 11.0 turns its local AI server into a full multimodal stack, adding text-to-speech and 3D generation alongside LLMs and image gen.
Branda: Open-Source AI On-Brand Ad Generator
Branda turns any website URL into ready-to-ship, on-brand LinkedIn and X ad creatives in seconds, pulling real logos and colors via the Context.dev Brand API.
Figmaboy: A Local-First Figma Clone With Codex Built In
Figmaboy is an open-source, local-first Figma clone with OpenAI's Codex CLI built directly into the canvas. Instead of generating flat images, its MCP server lets Codex inspect and edit real frames, layers, and components in your live document.
StyleSeed Stops AI Agents Building Generic UI
StyleSeed is a free MIT-licensed design engine that gives Claude Code, Cursor, and Codex real design judgment so AI-built UI stops looking generated.
Google Images Turns 25: Nano Banana Comes to Search
Google Images turned 25 on July 14, 2026, and Google rebuilt it around generative AI: Nano Banana image generation inside Search plus a personalized desktop gallery.
Bonsai 27B Runs a 27B AI Model on Your Phone
Bonsai 27B is a new open-weights multimodal model that PrismML compressed to as little as 3.9 GB, small enough to run entirely on a phone or laptop with no cloud connection.
Juggler: A Visual Workbench for AI Coding Agents
Juggler is a new open-source coding agent from JUCE creator Julian Storer that replaces the terminal with a visual, branching workbench for AI coding sessions.
Mozilla State of Open Source AI 2026: Key Findings
Mozilla released its inaugural State of Open Source AI report on July 14, 2026. Open-weight models have closed the capability gap with closed frontier systems to just 3.3 percent, while inference costs fell roughly 50 times in three years.
audio.cpp 0.3 Adds Supertonic 3 and 4 Local TTS Models
audio.cpp 0.3 adds Supertonic 3, IndexTTS2, Irodori-TTS, and MOSS-TTS to one C++/ggml engine, running local text-to-speech and voice cloning with no Python and no cloud fees.
Vivijure: Self-Hosted Open-Source AI Film Studio
Vivijure is a free, open-source AI film studio you self-host, orchestrating a dozen image, video, audio, and lip-sync models behind one web interface.
Gemini 3.5 Pro Reportedly Targets a July 17 Launch
Google's Gemini 3.5 Pro is reportedly targeting a July 17 general-availability launch, according to leaked launch plans and third-party reporting rather than any official Google post.
Grok Build CLI Uploaded Full Repos and Secrets to xAI
A security researcher has shown that xAI's Grok Build CLI was uploading entire git repositories, including unredacted secrets, to a Google Cloud Storage bucket, and that the tool's own opt-out toggle did not stop it.
Clawk: Sandbox Autonomous AI Coding Agents in a VM
Clawk gives an AI coding agent its own disposable Linux microVM, so Claude Code or Codex can run with full autonomy while your laptop stays isolated.
NoMac Ships iOS Apps From Your AI Agent, No Mac
NoMac is a new pipeline that lets an AI agent build, sign, and ship an iOS app to TestFlight and the App Store without a Mac or Xcode.
Claude Artifacts Get Multiplayer Editing and Public Links
Claude Code artifacts now support public share links and multiplayer editing, and can be created from Claude Tag in Slack.
How to Build an AI Short in ComfyUI With Velorn
Generate every clip, cut them on a real timeline, add captions, and export a finished AI short without leaving ComfyUI. A full step-by-step Velorn workflow, plus how to automate it with an AI agent.
Fudge MCP Gives AI Coding Agents Design Taste
Fudge MCP gives AI coding agents design taste grounded in real websites, letting them search nearly 10,000 sites by fonts, colors, and layouts.
Anthropic Launches Rupee Pricing for Claude in India
Anthropic has begun showing Claude subscription prices in Indian rupees across its website and mobile apps, its first move to localize pricing for its second-largest market after the US.
ChatCut Brings a Full AI Video Editor Into Codex
ChatCut has released a Codex plugin that turns OpenAIs coding agent into a full AI video editor, letting you describe an edit in plain language and get back an editable timeline.