Modly¶

Development line: project:modly · thread modly-development
Last event: 2026-03-20 · 1 dated since 2026-03-20 · Researched: 2026-09-05 · confidence: medium

What it is¶

Modly is a local alternative to cloud image-to-3D tools such as Meshy for creators who run models on their own hardware.

• 3D mesh generation from an image or text prompt.
• Model and process extensions, including Hunyuan3D 2, TripoSG, and Trellis.
• Desktop app with node workflows, mesh export, and a local CLI.

Development line¶

• 2026-03-20 — Modly GitHub repository reference. On 2026-03-20, Modly’s development line linked to the public GitHub repository github.com/lightningpixel/modly. The link establishes a dated repository reference without identifying a particular release, commit, feature, or operational change.

What changed¶

• 2026-03-20 — Modly was recorded at its public GitHub repository; no first-party dated artifact for that exact day's feature scope was retrieved.
• 2026-03-26 — Modly Beta v0.2.1 revised the extension system, trusted-extension handling, and updater.
• 2026-04-03 — Modly Beta v0.3.0 moved the product toward node-based workflows and an extension-centered Models page.
• 2026-04-21 — Modly Beta v0.3.2 added headless /workflow-runs execution and embedded Python.
• 2026-04-22 — Modly Beta v0.3.3 added a remesh workflow node.
• 2026-06-21 — Modly Beta v0.4.0 added the agent CLI, local-extension installation, Apple Silicon work, transform controls, and Gaussian-splat viewing.
• 2026-07-16 — Modly Beta v0.4.1 added workflow iteration nodes, audio process outputs, a workflow browser, and Apple Silicon packaging.
• 2026-08-28 — Modly Beta v0.4.2 became the latest tagged beta, fixing extension validation and picker behavior, CLI JSON arguments, Windows UTF-8 subprocess I/O, and workflow-run collection placement.

How to use this¶

From 2026-03-20, use the referenced GitHub repository as the starting point for source-level verification of Modly; do not infer releases, capabilities, or usage instructions from this link alone.

1. Check that the target model and platform fit the machine; the official site recommends an NVIDIA CUDA GPU with at least 6 GB VRAM, then download the current beta installer. — https://modly3d.app/
2. Install an official model extension from the Models page by using “Install from GitHub,” then download its model variant. — https://github.com/lightningpixel/modly
3. Build a basic graph: Image → Generate Mesh → Add to Scene; select it in Generate, run it, and inspect Settings/Logs/Errors if it fails. — https://github.com/lightningpixel/modly
4. Preview the mesh and export GLB, OBJ, STL, or PLY for the target DCC, engine, or slicer. — https://modly3d.app/
5. For automation, launch the desktop app, run python tools/modly-cli/agent.py health, inspect models, then use generate --image ... --output ... and retain the returned run ID for status or cancellation. — https://github.com/lightningpixel/modly/blob/main/tools/modly-cli/SKILL.md

Best practices¶

• Start with the smallest connected workflow and verify every edge before generating; preserve the mesh view and read the inline error/log panel instead of rebuilding the graph after a failed run. — https://github.com/lightningpixel/modly
• Use a clean, isolated input image or a specific prompt, then select the installed model variant for speed or fidelity rather than assuming one model fits every task. — https://modly3d.app/
• For scripted use, prefer the canonical health, model, and workflow-run commands; treat /generate/* and experimental ComfyUI helpers as compatibility or experimental paths, not a silent fallback. — https://github.com/lightningpixel/modly/blob/main/tools/modly-cli/SKILL.md
• Do not treat the desktop installers or stock extension setup as Jetson support: the Jetson guide documents an unsupported path requiring a Jetson-native PyTorch build, NumPy pinning, and a background-removal workaround. — https://github.com/lightningpixel/modly/blob/main/docs/running-on-jetson.md

Superseded by this¶

• The pre-v0.3 model-page-centered workflow is superseded by the node workflow and extension architecture from 2026-04-03.
• The old /generate/* automation surface is compatibility-only; canonical automation is the workflow-run contract introduced by the v0.4 line, 2026-06-21.
• Modly Beta v0.4.1 is no longer the latest tagged beta; v0.4.2 superseded it on 2026-08-28.

Still unknown¶

• No retrieved first-party release note, tag, or archived repository snapshot fixes the exact Modly feature set on 2026-03-20, so later releases are kept as separate events.
• First-party hardware guidance is not yet a single support matrix: the product site recommends NVIDIA CUDA with 6 GB VRAM, the repository lists Apple Silicon macOS, and the Jetson guide explicitly describes an unsupported workaround.
• A targeted Chinese-language search found secondary coverage but no official Chinese Modly documentation or release record.

Sources¶

Agent brief¶

• Subject: project:modly, thread modly-development, 1 dated events 2026-03-20 → 2026-03-20.
• Practical note: From 2026-03-20, use the referenced GitHub repository as the starting point for source-level verification of Modly; do not infer releases, capabilities, or usage instructions from this link alone.
• Confidence: medium. Dated supersedes above are the authority for what is obsolete.