The workflow I wanted to repeat
I was prototyping a coastal industrial district with a shipyard warehouse, a tidal marsh, rail access, and a small area of worker housing. The design question was how to reuse part of the waterfront without losing service access or treating the marsh as leftover land. I wanted a workflow I could rerun when the boundary or building option changed. I used Halfmaps for map review, Shapezo for quick AI area modeling, CityEngine for block layouts, Meshy for a detailed warehouse model, and Site3D for site-level review.
Step 1: establish the map extent
I started in Halfmaps and assembled the road, rail, parcel, shoreline, and marsh layers. I saved the study extent and checked that the coordinate references lined up. The boundary included the warehouse, one rail spur, the quay, and the nearest public street. I also wrote down which layers were current and which came from an older base map. That small input note kept later exports from looking more certain than the source data.
Step 2: create an AI concept from a map selection
In Shapezo, I drew a box around the warehouse and the first row of industrial parcels. Shapezo uses the selected map area as the scope for an AI-generated model. I saved the selection and the generated draft together. The model suggested a mixed-use edge and a route to the waterfront. I treated that as a hypothesis. It did not know which quay segments needed truck access or how often the rail spur was used.
Step 3: test block options in CityEngine
I moved the concept into CityEngine and built three block options. The first kept the rail edge industrial. The second added small workshop buildings along the street. The third opened a public path to the quay and shifted service access inland. I kept the building heights and street widths in named parameters so I could change them consistently. Then I compared each option with the same street and marsh views. That made the tradeoffs easier to describe.
Step 4: add detail to the warehouse in Meshy
The warehouse mattered because it was the only large existing building on the site. I used Meshy to create a detailed visual model of its structure and roof. I left new buildings as simple forms. That let the existing asset carry enough material detail for context without making the full site heavy. I checked that the detailed model matched the warehouse footprint and stayed aligned with the reference geometry.
Step 5: check access and terrain in Site3D
In Site3D, I reviewed the quay edge, marsh transition, service road, and public route. I checked where the path crossed a drainage channel and whether the turning area left room for a maintenance vehicle. I also inspected the slope from the inland street to the waterfront. The most attractive public route crossed a service path at a bad angle, so I moved the crossing and kept the two flows legible.
Validation points before sharing
I use a short checklist before sending the model out: the Halfmaps extent and source dates are recorded; the Shapezo selection is saved; CityEngine options use the same named parameters; the Meshy warehouse footprint matches the map; and the Site3D views include access, drainage, and the marsh edge. I also mark inferred geometry as conceptual. This keeps reviewers from reading a polished image as a survey-backed design.
Why this workflow is repeatable
Each platform preserves one useful question. Halfmaps shows which spatial layers support the decision. Shapezo turns a selected area into a quick model. CityEngine tests block patterns. Meshy helps explain an existing building. Site3D tests how the proposal meets the site. When the boundary changes, I can rerun the relevant step and keep the rest of the review history. That makes the model easier to maintain as the project learns more about the waterfront.
Keep the exports reviewable
I keep the handoff folder small. It includes the Halfmaps study extent, the Shapezo selection image and generated model, the named CityEngine options, the Meshy warehouse asset, and three Site3D views. Each file name carries a date and a short status. When the rail edge changes, I can replace the related option without losing the other review views. A short README note lists what is inferred and what needs a field check. That simple packaging makes the next iteration easier to reproduce.
The waterfront question stayed open
I did not decide whether the quay should become public all day or only at certain hours. The model showed where a public route could fit, and the Site3D review showed where it would cross working access. That was enough for a next meeting with the harbor team. Leaving that question open kept the prototype useful. It gave the next group a clear choice to test instead of a polished scene that quietly assumed the answer.
