Greetings! As I understand it ya'all use lots of javascript via node.js, There's a bunch of malware making the rounds hitting javascript and nodejs. I simply don't know how bad this wave is. Here's a ton of bleeping computer news and articles: https://www.bleepingcomputer.com/tag/javascript/ https://gizmodo.com/sprawling-malware-attack-reportedly-affects-zoom-cnn-fox-news-peacock-and-170-others-2000840831
>From the much less bad department of Ai. Holly hell! (while from the game world) ChatGPT and Anthropics coding models can help debug? oO Do you folks know if it's actually...helpful Oo I thought debugging also needed a human in the loop to betterr know: bug, bad ui/ux , or: oops. app crashes when I do (thing). And something that (might) be useful for ya'all to: https://github.com/microsoft/TRELLIS.2 : Pictures to 3Dish models. Can run on a home computer. and: "The official repository for TripoSplat is available at VAST-AI-Research/TripoSplat. It converts a single 2D image directly into a variable-density 3D Gaussian Splat (up to 262,144 Gaussians) in a single feed-forward pass. [1 <https://github.com/VAST-AI-Research/TripoSplat>, 2 <https://github.com/lightningpixel/modly-triposplat-extension>] How the TripoSplat Pipeline Graph Works Unlike traditional multi-view photogrammetry, TripoSplat generates 3D splats from one photo using a streamlined feed-forward graph: [1 <https://github.com/lyehe/lichtfeld-triposplat-plugin>, 2 <https://note.com/angework/n/n769fd9eb4d26?hl=en>, 3 <https://github.com/lightningpixel/modly-triposplat-extension>] - Input & Foreground Cutout. The graph takes a single RGB image and isolates the subject from the background (typically using BiRefNet in integrated workflows). [1 <https://github.com/lightningpixel/modly-triposplat-extension>, 2 <https://github.com/lyehe/lichtfeld-triposplat-plugin>] - Feature Extraction & Flow Sampling. A vision backbone (such as DINOv3) encodes the image features, which are passed through a Euler flow-matching sampler (10–25 steps) to generate a 3D latent representation. [1 <https://github.com/lightningpixel/modly-triposplat-extension>] - Variable Gaussian Decoding. The latent is decoded into a cloud of 3D Gaussians with controllable density (65k, 131k, or up to 262k Gaussians). You can re-decode the cached latent at a higher or lower Gaussian count without re-running the diffusion sampler. [1 <https://github.com/VAST-AI-Research/TripoSplat>, 2 <https://github.com/lyehe/lichtfeld-triposplat-plugin>, 3 <https://github.com/lightningpixel/modly-triposplat-extension>] - Optional Mesh Conversion. While the native output is a .ply or .splat Gaussian cloud, node graphs can route the splats through a density grid and Surface Nets extraction (such as ComfyUI's SplatToMesh node) to output a watertight, vertex-colored .glb mesh. [1 <https://github.com/lightningpixel/modly-triposplat-extension>] GitHub Repositories & Node Integrations Depending on which node-based environment you are using, you can grab the core code or pre-wired graph integrations from these GitHub repositories: - VAST-AI-Research/TripoSplat (Official Inference Repo). Contains the standalone Python implementation (triposplat.py and model.py), a local Gradio interface (run_gradio.py), and instructions to pull the ~3.8 GB weights from VAST-AI/TripoSplat on Hugging Face <https://huggingface.co/VAST-AI/TripoSplat>. - ComfyUI Official TripoSplat Workflow <https://docs.comfy.org/tutorials/3d/triposplat>. ComfyUI includes native node support and an official drag-and-drop workflow template for generating, previewing, and converting TripoSplat outputs to .glb meshes. - lightningpixel/modly-triposplat-extension <https://github.com/lightningpixel/modly-triposplat-extension>. A node extension for Modly that wraps the pure-PyTorch TripoSplat pipeline and automatically converts the resulting .ply/.splat into a smoothed, vertex-colored .glb mesh. - lyehe/lichtfeld-triposplat-plugin. A plugin for LichtFeld Studio that inserts generated TripoSplat assets directly into a live 3D scene for placement and training. [1 <https://github.com/Comfy-Org/docs/discussions/1580>, 2 <https://github.com/VAST-AI-Research/TripoSplat>, 3 <https://github.com/lyehe/lichtfeld-triposplat-plugin>, 4 <https://github.com/lightningpixel/modly-triposplat-extension>] Might be helpful to kind of knock together something...and or just mess around with. I don' think it's a complete mesh with this technique. my brain melts that it's a thing...
.- .-.. .-.. / ..-. --- --- - . .-. ... / .- .-. . / .-- .-. --- -. --. / ... --- -- . / .- .-. . / ..- ... . ..-. ..- .-.. FRIAM Applied Complexity Group listserv Fridays 9a-12p Friday St. Johns Cafe / Thursdays 9a-12p Zoom https://bit.ly/virtualfriam to (un)subscribe http://redfish.com/mailman/listinfo/friam_redfish.com FRIAM-COMIC http://friam-comic.blogspot.com/ archives: 5/2017 thru present https://redfish.com/pipermail/friam_redfish.com/ 1/2003 thru 6/2021 http://friam.383.s1.nabble.com/
