2. Choose the stages

Decide: which stages does your lesson need, and what happens in each one?

2.1 Use XR where the real thing is risky, scarce or hidden

Do: go through these questions for your task.

If on the real object…	Then
mistakes are dangerous or costly	Practise in VR first, where mistakes cost nothing.
practice time is scarce (few devices, lab slots)	Practise in VR first, so every real attempt counts.
the important parts are hidden or move too fast to see	Use XR to make them visible: open the machine, slow it down, show the inside.
learners need help on the object itself	Use AR, with labels lying on the real parts.

In ALF: you choose VR or AR per scene in Scene Generation: VR or AR?. Lining up digital content with a real object in AR is done with Positioning Setup.

Why: XR pays off where practising on the real thing is risky, rare or doesn't show what matters. Simulation fidelity research (Alexander et al.), situated learning (Lave & Wenger).

2.2 Keep the same task in every stage

Do

• Use the same task, the same order and the same words in every stage. Only the amount of help changes.
• Build one scene per stage, and put them in order in the main menu.
• If your task is long, don't cut it into separate mini-lessons. Keep the whole task, and make the first version simpler instead.

In ALF

• Create one scene per stage with Scene Generation.
• Put them in order in the main menu, and give them names that show the order, for example 1 Practise, 2 On the real device.
• A scenario has at most 20 steps (see Add, reorder or remove steps). For a longer task, group the actions into phases, one step per phase action, or split the task into sections with their own scenes.

The learner experiences: the same task, recognisably, in a more and more real setting.

Avoid: a VR stage that teaches something different from what learners later do for real, for example a simplified order of steps.

Why: practising the whole task with decreasing support builds skills that transfer. Skills carry over best when practice and the real task share as many elements as possible. 4C/ID (van Merriënboer & Kirschner), identical elements (Thorndike), hugging (Perkins & Salomon).

2.3 Make the actions faithful, not the graphics

Do

• Put your effort into what learners do and how the system reacts: the right control, the right part, the right order, a visible result.
• Simple models are fine, as long as the parts learners work with are where they are in reality and work like they do in reality.
• Use hands when the real task is done by hand (see Let learners act).

In ALF: moving parts and their reactions are set up with the Kinematic Editor Tool, including sounds or lights when a part moves.

Avoid: spending your time on photorealism while the steps don't match the real procedure.

Why: transfer depends more on doing the right actions than on how real it looks. Beginners even learn better with simpler visuals, because there is less to distract them. Simulation fidelity (Hays & Singer; Alexander et al.), plausibility (Slater).

2.4 Plan what XR can't teach

Do

• Ask what the real task needs that a headset can't give: force and resistance (how tight is tight enough), heat, weight, fine finger work, sound or smell as a warning sign.
• Explain those parts in words in VR, and move the practice of them to the AR stage or the real stage.

In ALF: add a short note on the instruction panel where the feeling matters (for example "firm, but not as hard as you can"), and plan that step on the real object in the checklist.

Avoid: assuming learners can do something for real just because they did it in VR.

Why: every simulation leaves something out. Knowing what is missing tells you what the next stage must add. Simulation fidelity, motor learning stages (Fitts & Posner).

2.5 Which stages, at a glance

Your situation	Suggested stages
Risky or costly task, real object available	Orient → Practise in VR → Transfer in AR → Perform for real
Risky or costly task, AR not available	Orient → Practise in VR → Perform for real (supervised)
Real object rarely or never available	Orient → Practise in VR, and AR in the room if available. Check in VR or with a supervisor.
Safe task, real object available, help is needed on the object	Orient → Transfer in AR → Perform for real
Understanding a system rather than doing a task	Orient → Explore in VR or AR → Explain and apply