Module pdf3d 

The live 3D model a sheet PDF can carry — a U3D stream (super::u3d) and the PDF 1.7 dictionaries that show it (ISO 32000-1 §13.6).

What the reader sees

One U3D stream holds the whole scene, shared by every 3D annotation in the file:

• each drawn solid as a MESH node named for the solid, and its edges as a dark LINE SET node beside it ({solid} edges), so the model reads as a CAD model and not a smooth blob;
• each enabled, resolved annotation of each saved view with a camera as a LINE SET node (PMI {view} {annotation}: dimension, extension and leader lines, frames, and the text STROKED with the PMI font — the same pmi_present layout the viewport and the STEP presentation use, sized as the viewport drew it when the view was saved), plus a small MESH node for its filled arrowheads (… fill).

Each saved view is a /3DView in the stream’s /VA list: its camera as the camera-to-world matrix, its projection, and a /NA node list that shows THAT view’s annotations and hides every other view’s, and hides the solids the view hides. A reader’s view menu is therefore the document’s PMI view list, and picking one shows exactly that view’s PMI.

Only readers that implement 3D annotations draw any of this — Adobe Acrobat / Reader (with 3D enabled) and Foxit. Every other reader shows the annotation’s appearance, which is why the sheet’s own vector drawing stays on the page beneath a placement, and why the full 3D page draws the default view as a hidden-line drawing under its 3D box.

The camera

A view’s /C2W maps camera coordinates to world coordinates: its columns are the camera’s x axis (right), y axis and z axis (the viewing direction, eye → target), then the eye. ISO 32000-1 §13.6.5 states the camera’s y axis points UP; CAMERA_Y_UP records which way this writer points it, because the claim is checked in a real reader, not taken from the text.

Structs

Body

One solid, gathered by the engine from the display scene.

Button

A view button on the 3D page.

DrawnAnnotation

One annotation laid out in world space as a 3D view draws it.

Model3d

The file’s 3D content: the U3D bytes, the node names in stream order, and every saved view that has a camera.

Page3d

The 3D page: a title, the 3D box, a button per saved view, a caption.

Pdf3d

Everything 3D a sheet PDF carries.

Placement3d

A placement drawn live: a 3D box over the placement’s drawing, opening on the placement’s own camera at the placement’s own scale, so activating it changes nothing until the reader turns the model.

Plan

The object numbers of every 3D object, assigned before any is written so the page dictionaries can name their annotations.

Summary

What a PDF carries in 3D, read back from its bytes: the count of 3D annotations, of views in the stream’s /VA, of view links, and the U3D stream’s decoded size — what the export door reports about the file.

View3d

A saved view as the reader will show it.

Enums

Projection3d

How a view projects.

Constants

CAMERA_Y_UP

The camera’s y axis is the view’s UP vector (true) or its down vector (false). See the module doc.

CAPTION

The paragraph under the 3D box — the truth about who can show it.

ISOMETRIC

The id of the view a document without saved views opens on.

Functions

ascii85

ASCII85 (ISO 32000-1 §7.4.3), ~>-terminated, in 76-character lines.

build

Build the 3D model: the solids, then every saved view’s annotations.

c2w

The 12-element camera-to-world matrix and the distance to the centre of orbit (the camera’s target).

camera_frame

The camera’s right, up and forward (eye → target) unit vectors; None for a degenerate camera (eye on target, or up along the view).

isometric_camera

Looking down the (−1, −1, −1) diagonal at the solids’ box centre, +Y up (the application’s), orthographic, the bounding sphere filling the height.

projection

A view’s projection: a perspective camera keeps its field of view; an orthographic one is scaled so its saved half height fills half of a 3D box box_height_pt tall, unless scale fixes it.

rect_pt

A 3D annotation’s rectangle in PDF points: centred on centre_mm (sheet millimetres from the top-left), half_mm either way, on a page page_height_mm tall.

stage_matrix

A solid’s explode stages as a /3DNode /M: the matrix’s three columns, then its translation, the order /C2W is written in. Every stage is affine, so the composition is read off where it takes the origin and the three unit points.

stream_object

The 3D stream: every saved view in /VA, the default the first, and the U3D bytes ASCII85-encoded so the file stays ASCII (the PDF writer’s contract).

summary

Summary of a file this module wrote; None when it carries no 3D.

view_annotations

Every drawable annotation of a saved view (rows, its report) laid out for a 3D model shown from camera: sized as the viewport sized them when the view was saved (the overlay’s pixel sizes times the saved camera’s world per pixel; a model-unit line cannot stay a constant size on screen, so this is exact at the view’s own framing and scales with the model when the reader zooms), a view-aligned row laid out in the view-parallel plane through its label, facing the camera, as the STEP presentation lays it out. The 3D PDF and the instructions player both draw these.

view_dict

A /3DView dictionary: name in the reader’s menu, the view’s camera, projection, its render mode, a CAD lighting scheme, and every node’s visibility.

world_per_pixel

Model units per viewport pixel under a saved camera, in the viewport the view was saved from — ViewCamera::world_per_pixel’s rule.