Developer builds offline Python tool that converts photos into 3D-printable meshes

·1 views

A developer has released PolyForge, a standalone Python package for CAD generation and offline photogrammetry that requires no cloud API or language model to function. The tool supports three CAD backends — OpenSCAD, FreeCAD, and Blender — and can generate parametric parts like brackets and enclosures from plain-text descriptions. A newly added feature, reconstruct-from-photos, uses COLMAP for sparse 3D reconstruction and OpenMVS for dense meshing, enabling full offline conversion of overlapping photos into STL files. Building the pipeline required patching OpenMVS source code three times to resolve compatibility issues with a newer OpenCV version on a rolling-release system. Testing revealed that camera coverage pattern matters more than photo count, with a minimum of three elevation rings needed for successful reconstruction.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Tool Uses Monte Carlo Simulation to Reveal True Lifetime Cost of Owning a Dog

A developer has built a web tool that estimates the lifetime cost of dog ownership by breed using a Monte Carlo simulation of 10,000 simulated dog-lifetimes per breed, totalling 300,000 lives computed in Snowflake SQL. Unlike typical articles that offer a single average figure, the tool displays a full probability distribution, highlighting the financial tail-risk most owners overlook. Among its findings, Golden Retrievers and Labrador Retrievers rank as the costliest breeds, with Golden Retrievers carrying nearly a 1-in-5 chance of exceeding $40,000 in lifetime costs, largely due to cancer risk. The tool also includes an insurance scenario lab with adjustable deductibles and coinsurance sliders, an adopt-versus-buy toggle, and a breed cost leaderboard. All figures are presented in uninsured, pay-as-you-go US dollars and are accessible via a live Streamlit app.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

DeepRead: Open-Source Tool Links AI Reading Summaries to Source Evidence

A developer has released DeepRead, an open-source AI reading tool designed to clearly separate an author's claims, supporting evidence, and model-generated inferences. Unlike conventional summarizers, DeepRead structures reports around claims, evidence, data, and relationships, with each key claim linked to its source location. The tool supports articles, books, PDFs, and document sets, and offers five analysis modes including deep-dive, knowledge mapping, and plain-language explanation. It can also compare between 2 and 10 documents to surface agreements, conflicts, and complementary evidence. Released under the MIT license at version 0.5.4, the project has passed automated checks for GitHub's Awesome Copilot list and is awaiting maintainer review.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Dev Community Publishes Best Practices Guide for Java-Selenium Automation Frameworks

A detailed best practices guide for building scalable Java-Selenium test automation frameworks has been published on DEV Community. The guide outlines a strict layered architecture separating UI page locators, business components, and test scripts into distinct responsibilities. It covers naming conventions, Page Object Model implementation, locator strategy priorities, and parallel execution configuration. The framework emphasizes reusability by requiring business components to extend shared base classes rather than calling Selenium WebDriver directly in test scripts. Additional topics include CI/CD integration, ExtentReports setup, and environment configuration via a centralized properties file.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Why AI Tools Keep Forgetting You: The Fragmented Memory Problem Explained

Developers using multiple AI tools like Claude, Cursor, and ChatGPT face a recurring frustration: each tool maintains its own isolated memory, forcing users to repeatedly re-establish context when switching between them. This fragmentation is not a technical limitation of context windows but a structural problem where proprietary memory silos cannot share user preferences, project details, or prior decisions. The constant need to re-explain context imposes both a cognitive burden and real token costs on developers. Proposed solutions like vector databases fall short because they lack temporal awareness, cannot resolve contradictory facts, and struggle to capture relationships between pieces of information. The article argues that true AI memory requires not just retrieval but an understanding of how facts evolve and relate to one another over time.

0 comments Read more at DEV Community