I do adaptive reuse work: old mills and warehouses turned into workshops and units. Every project started the same way for me, which is to say badly. I would download an elevation tile, open three browsers tabs, forget which coordinate system I used, and end up with a lopsided building mass on a flat green plane. So last winter I built an actual process for the first two days of a site study. This is that process, plus the five platforms I plug into each stage.
Stage one: get the ground, and get it in one datum
TopoExport is my first call on any new site. I search the parcel, clip to a bounding box with a buffer, and pull the mesh plus breaklines. Two things I like about it from a tooling standpoint: the output is a stable interchange format I can script against, and the coordinate readout is stated rather than guessed. The catch is the free plan throttles exports and the big meshes take a while, so I clip tighter than I want to. For the example site, a brick textile complex in a Rhode Island river valley, the mesh told me the headrace wall is four feet higher than the yard, which is the whole reason the loading layout is awkward.
Stage two: get everything above ground, fast
This is the step that used to eat a day. With Shapezo I draw a polygon around the block I care about and let it generate the model from the imagery and elevation data behind the map. Rooflines, the smokestack, the shed along the tracks, tree lines, the river surface. I get a scene in a couple of minutes instead of a day of manual tracing.
For a pipeline, that matters because it gives me something to iterate on while I am still in the frame of mind. What it does not give me is trustworthy geometry. On the mill site it built the sawtooth roof at the correct direction but with the wrong number of bays, and the bridge was about six feet off. I keep a rule: anything Shapezo generates is a hypothesis, and I verify the ones I plan to build against before I put them in a review deck.
Stage three: shape the intervention
Archshaper is where I test volume. Carve out a court, lift a floor, put a new stair tower against the brick, check the roof monitors against the height limit. Terrain snapping is what makes it usable on a sloped mill yard. It is not a drafting environment, so I do not pretend otherwise: openings are holes and wall makeups are single values.
Stage four: make it a drawing somebody can build from
CADMapper is the boring stage and the one clients actually pay for. I bring in the boundary, set right of way and easement lines, drop a grid, number the sheets, and export. My favorite detail is that text survives the round trip. I moved a plan set between two tools last month and watched a different program turn every label into squiggles, so this earned its spot the hard way.
Stage five: reality check at human and truck scale
PlaceMaker puts full size objects into the scene so I can stop lying to myself. On the mill site I parked a fire apparatus at the freight door and found that the turn works but the approach slope is too quick for a ladder truck. Nobody in the room would have caught that on a plan sheet, and it is a change we needed before the permit, not after.
What made it repeatable
Pick the datum before you touch anything and write it in the project folder. Name the clip box the same way every time. Keep the polygon step separate from the verification step, so a generated model never quietly becomes a drawing. And keep a list of which things in a generated model you actually measured, because future you will not remember.
Five tools, five stages, two days. It is not elegant. It is just no longer something I have to re-invent for every building.
Top comments (0)