Generative massing sounds like it should be complicated — algorithms, parameters, AI-generated options — but the actual beginner workflow in Autodesk Forma is more approachable than the term suggests. You’re not writing code or configuring a machine-learning model. You’re defining a handful of real-world constraints (plot boundary, height limits, setbacks, program mix) and letting the tool generate and evaluate multiple massing arrangements against those constraints, so you can compare options side by side instead of designing one option and hoping it’s close enough.
This guide walks through what that process actually looks like for someone doing it for the first time, step by step, with the kind of beginner mistakes to watch for along the way.
According to Autodesk’s own product material, the feature now called Site Automation generates and evaluates multiple site design options based on parameters the user defines, rather than requiring every variant to be modeled by hand from scratch. The workflow explicitly supports three approaches: drawing massing manually yourself, letting the tool generate options automatically within your constraints, or combining both — sketching a rough direction and letting the tool fill in and vary the detail.
This matters for a beginner because it removes the biggest practical barrier to testing multiple options: the time cost of modeling each one by hand. In a traditional workflow, testing five genuinely different massing concepts means modeling five things from scratch. In Forma’s generative workflow, you define the rules once, and the tool produces multiple compliant variations for you to evaluate.
Before generating anything, a first-time user needs to define the constraints the generated options will have to respect. The core parameters, based on Forma’s documented workflow, are:
For a true first project, keep these simple and realistic rather than importing every nuance of a real client brief on day one. A straightforward rectangular plot with one height limit, one setback rule, and one FAR target is enough to learn the mechanics before layering in the complexity of a genuinely messy real-world site.
With parameters set, the next step is letting Forma generate massing options within those constraints. This is where the tool’s real value for a beginner becomes obvious: rather than staring at a blank canvas trying to imagine what a compliant massing could look like, you get multiple concrete starting points to react to, refine, or discard. Autodesk’s own course material describes generating options this way as a direct alternative to (or complement of) manually sketching massing from scratch.
A realistic first-project goal here is to generate somewhere in the range of 5–10 distinct options — enough variety to see genuinely different approaches to the same constraints, without producing so many that comparing them becomes overwhelming for a first attempt.
Once you have a set of generated options, the next step is evaluating them against real performance criteria rather than picking one on pure visual instinct. Forma’s environmental analysis tools — daylight, wind, outdoor comfort, noise (including from nearby roads and railways), and solar exposure — can be run across your generated options to see how each one actually performs, not just how it looks.
For a first project, don’t try to optimize for every metric simultaneously. Pick the one or two performance factors most relevant to your specific site and brief — sun exposure on a south-facing plaza, say, or wind comfort on an elevated terrace — and use those as your primary comparison criteria. Trying to simultaneously optimize daylight, wind, noise, and solar exposure on your very first generative study is a fast route to analysis paralysis rather than a useful comparison.
Autodesk’s course material specifically highlights a visual compare tool as part of this workflow — a way to see multiple generated (or manually drawn) options next to each other rather than toggling between them one at a time from memory. For a beginner, this comparison step is genuinely where the learning happens: seeing that Option 3 gets noticeably better morning sun on the plaza than Option 1, at the cost of slightly worse wind shelter, teaches you something real about how massing and orientation decisions trade off against each other — a lesson that’s much harder to internalize by modeling one option at a time in isolation.
A few patterns show up reliably with first-time users of generative massing tools:
Once a massing direction is selected from your comparison, Autodesk’s workflow moves it into further development — internally referred to as moving the design into 3D Sketch for more detailed refinement, and eventually handing off to Revit for full detailed design and documentation. The generative study’s job is done once it’s given you a defensible, comparison-backed reason to commit to one direction rather than another — it isn’t meant to replace the detailed design process that follows.
It’s worth being clear-eyed about what carries forward and what doesn’t at this handoff point. The selected massing’s overall form, orientation, and the performance reasoning behind choosing it all carry forward directly — that’s the whole value of the exercise. What doesn’t carry forward automatically is every structural, material, and systems decision that detailed design still has to make; the generative study answers “what shape and orientation” and “why,” not “what structure, what facade system, what MEP strategy.” Treating the handoff as the end of the easy, cheap-to-change phase and the beginning of the more committed, detailed phase helps set realistic expectations with the rest of the project team about what’s still genuinely open for discussion at this point and what’s effectively settled.
It’s worth being explicit about why the comparison step — not the generation step — is actually the valuable part of this workflow. Generating options is mechanically easy once your parameters are set; the tool does that part for you. The genuine design skill being exercised is in the comparison: noticing that one option performs meaningfully better on the specific criterion that matters most for this project, understanding why, and being able to articulate that reasoning to a client or team member who wasn’t in the room when you ran the study.
This is also where a beginner’s instinct to generate one option and move on actually works against their own learning, not just against the project outcome. Comparing even three or four meaningfully different massing arrangements against the same performance criteria teaches you, concretely, how orientation, height, and footprint shape affect real outcomes on your specific site — a lesson that doesn’t transfer nearly as well from reading about climate-responsive design in the abstract.
To make this concrete, consider a realistic first project: a mixed-use plot with a ground-floor retail frontage and residential floors above, constrained by a moderate height limit and a standard FAR target. A beginner working through this in Forma for the first time might reasonably do the following:
Notice that this workflow deliberately narrows in stages — broad generation first, broad comparison second, then a closer second-pass look at only the strongest candidates — rather than trying to deeply analyze every single generated option with equal effort. That staged narrowing is what keeps a first project manageable rather than overwhelming.
A single generative massing exercise teaches you the mechanics. Genuine fluency — knowing instinctively which parameters to set, which comparison criteria matter for a given site, and how to interpret results quickly rather than puzzling over them — comes from repetition across different site types and briefs. This is exactly where structured, project-based training earns its value over unguided practice: working through multiple realistic scenarios with an instructor who can explain not just how to run the tool but why a given result looks the way it does, compresses months of trial-and-error into a much shorter, much more deliberate learning curve.
| You Define (Inputs) | Tool Evaluates (Outputs) |
|---|---|
| Plot boundary | Multiple compliant massing arrangements |
| Height limits & setbacks | Daylight and solar exposure performance |
| Floor Area Ratio (FAR) | Wind behavior and outdoor comfort |
| Land use / program type | Noise exposure (roads, rail) |
| Comparison priorities (your choice) | Side-by-side visual comparison of all options |
No. The generative workflow is parameter-driven through the interface — you set constraints like height, setbacks, and FAR, and the tool generates options within them. No scripting is required for this core workflow.
Somewhere around 5–10 is a practical starting range — enough to see genuinely different approaches without making the comparison step overwhelming for a first attempt.
No — it’s a comparison-stage output meant to help you choose a direction with real data behind the decision. The selected option still moves into further development (3D Sketch, then Revit) before it’s construction-ready.
Manual drawing requires you to model every variant by hand; Site Automation generates multiple compliant variants automatically from your parameters. Forma explicitly supports combining both — sketching a rough direction manually and letting the tool generate variations around it.
Pick the one or two factors most relevant to your actual site and brief rather than trying to optimize everything at once. A plaza-focused project might prioritize sun exposure and outdoor comfort; a roadside site might prioritize noise. Trying to weigh all available environmental checks equally on a first attempt tends to produce a comparison that’s technically thorough but practically hard to act on, since no single option is likely to win on every metric simultaneously.
Forma’s generative workflow is built around the actual plot boundary you define, which can be any real shape — it isn’t restricted to simple geometry. For a genuine first attempt, though, starting with a simpler plot shape makes it easier to understand how the parameters and generated results relate to each other before tackling a more complex, irregular site where those relationships are harder to isolate.
Generative massing in Forma isn’t about handing design decisions to an algorithm — it’s about removing the time cost of testing multiple real options, so the decision you eventually commit to is backed by actual comparison rather than a single untested guess. For a beginner, the real skill to build isn’t operating the generator itself (that part is mechanically simple); it’s learning to set sensible constraints, pick meaningful comparison criteria, and interpret what the results are actually telling you — exactly the kind of practical judgment a structured course, with real project examples, builds faster than working it out alone.
Anyone ready to build this skill properly rather than piecing it together from scattered tutorials can start with Orbit’s Autodesk Forma course in Dubai, which walks through this exact generative workflow on real project scenarios.
Source: Autodesk Forma blog, “How to use generative design, AI, and 3D modeling for improved site planning”; Autodesk Forma on-demand course material, “Conduct a massing study in Forma.”
