From a shop drawing to an assembly video.
The useful transformation is from a drawing someone has to decode to a sequence they can follow. A video is the final expression of that work.
Start with a product you can describe.
A good input identifies the components, shows their dimensions, and explains the intended connections. An illustrated instruction guide may already contain a sequence. A shop drawing may show only the finished positions. Those inputs require different amounts of interpretation.
For the current Assembly Animator, that information arrives as a PDF. Include readable dimensions, multiple views when needed, and notes that explain hardware or assembly order. A selected hardware category is useful context; it is not evidence of the number or location of fasteners.
Build the model before the movement.
The analysis becomes a structured component model: each part has dimensions, a final position, a material, and supported geometry. Pins, screws, bolts, washers and nuts are individual parts. Openings that receive them need to be represented, too.
This matters because a plausible outline is not enough. A pin passing through a solid panel may look convincing from one angle, but it describes the wrong construction. The model needs to explain the connection.
Make the sequence explicit.
Supported placements use straight paths. Related components can move together, and ordinary round fasteners can use illustrative axial turning. The current engine can also try to infer a clear placement order from final geometry when explicit steps are absent.
The planner checks modeled overlaps and obstructed paths. It cannot infer every real-world constraint: a support jig, a technician's grip or a required tightening order may need to come from the product expert.
Render, then compare with the source.
The video is rendered from the saved model, with consistent parts and labeled steps. That gives you a concrete artifact to review alongside the source: count the parts, inspect their orientation, confirm the hardware, and check the order.
A successful render proves that the modeled scene can be rendered. It does not prove that the interpretation matches the drawing.
A useful first trial.
Choose a small assembly with readable dimensions and standard hardware. Write down which details must be exact and which can be simplified for explanation. The goal is a result a product expert can review and a customer can understand.