feat: add Preview3DAdvanced node

[jtydhr88]

A new Preview3D that emits live viewport state as outputs: model_file, camera_info, model_info, width, height. Reuses the Load3D capture widget and the PreviewUI3D side effect.
Input model_file is File3D-only (no String path) to keep the contract typed; width/height are pre-promoted to socket inputs.

Inputs win when connected; otherwise the viewport's live framing & transform carries through.
The same resolved values flow to the output (so chained Preview3DAdvanced nodes propagate in a single run, and headless execution still carries state) and to PreviewUI3DAdvanced (so the viewport applies the input when one is connected).

FE: Comfy-Org/ComfyUI_frontend#12527

2026-05-29.22-57-21.mp4

[coderabbitai]

📝 Walkthrough

Walkthrough

This PR introduces Preview3DAdvanced, a new experimental output node for handling 3D file previews. The node accepts a 3D model file, input image, optional camera parameters, and model metadata, along with configurable output dimensions. During execution, it persists the 3D asset to the output directory under a generated filename and returns the original model file, camera information, model metadata, dimensions, and a UI preview instance. The node is registered in the extension's public node list.

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately describes the main change: adding a new Preview3DAdvanced node, which is the primary focus of this PR.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Description check	✅ Passed	The PR description clearly describes the new Preview3DAdvanced node, its inputs/outputs, behavior, and design decisions.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 2

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@comfy_extras/nodes_load_3d.py`:
- Around line 152-177: The execute method in Load3D node (execute) currently
ignores the optional input sockets: it always reads camera_info and
model_3d_info from the widget-provided image dict, while the UI preview reads
kwargs.get("camera_info"), so wired inputs are dropped or inconsistent. Fix by
preferring connected inputs from kwargs when present: compute camera =
kwargs.get("camera_info", image.get("camera_info")) and model_3d =
kwargs.get("model_3d_info", image.get("model_3d_info", [])), use those same
values in the IO.NodeOutput and pass camera to UI.PreviewUI3D so the output and
preview are consistent and wired inputs are respected; alternatively remove the
unused input sockets if you intend to always use widget values.
- Around line 138-150: Preview3DAdvanced.execute() currently ignores the
explicit inputs IO.Load3DCamera.Input("camera_info") and
IO.Load3DModelInfo.Input("model_3d_info") and always returns
image['camera_info'] and image.get('model_3d_info', []); update execute() in
Preview3DAdvanced to prefer the provided inputs (kwargs.get("camera_info") and
kwargs.get("model_3d_info")) and fall back to values embedded in the loaded
image (image.get('camera_info') and image.get('model_3d_info', [])), or remove
the unused input nodes if you intend the image fields to be authoritative;
reference Preview3DAdvanced, its execute() method,
IO.Load3DCamera.Input("camera_info"), IO.Load3DModelInfo.Input("model_3d_info")
and IO.MultiType.Input("model_file", ...) when making the change.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: 908ae6a9-4055-46dc-9023-9d17641404cd

📥 Commits

Reviewing files that changed from the base of the PR and between ec1896a and eb7cd3b.

📒 Files selected for processing (1)

• comfy_extras/nodes_load_3d.py