What this article is for
ZeroCount's original workflow read a drawing's legend for you. You marked out the parts of the sheet worth counting, pressed Submit, and the server extracted every symbol from the legend table and matched each one across the whole drawing in a single unattended run.
That pipeline still exists behind the scenes, and you may meet its language — the Stage 1, Stage 2, Stage 3 strip in the sidebar, or a drawing that was set up under it before you inherited the project. This article explains what it did, so those things make sense.
This is background reading, not a procedure you can follow today. Auto Takeoff from Legend is not shown in the New Takeoff menu in this build, and new drawings skip Stage 1 entirely — they open straight into the takeoff workspace. Use Auto Detect Symbols and Detect New Symbol instead.
The three stages
The sidebar shows Stage 1 › Stage 2 › Stage 3 above the marker list, with the stage you are in highlighted. Hovering each one describes it:
| Stage | What it was for |
|---|---|
| Stage 1 | "Amend the auto-detected areas of interest (AoIs) and legends, or create new ones. Submit to detect any legend symbols found within the AoIs." |
| Stage 2 | "Review and amend the detected legend symbols. Create new markers to highlight any missed components. Create a new group to manually mark any missed legend symbols. Or, rearrange existing markers assigned to the wrong group." |
| Stage 3 | "Focus on reviewing and organizing detections and detections groups ready for export." |
Today the strip mostly reads as a progress indicator. Stage 2 is the takeoff workspace where you count, and Stage 3 is Review mode — collapsing the sidebar moves you between them. See Takeoff and Review modes.
How the old flow worked
A drawing opened at Stage 1 rather than at the takeoff workspace. ZeroCount had already had a first look at the sheet and placed its guesses on it: one or more AoIs around the plan areas, and a rectangle around what it took to be the legend. You corrected those — moved an AoI off the title block, redrew the legend box — and pressed Submit.
The rest ran on the server, with a message reading "Reading the legend, then matching every symbol across the drawing. This can take a few minutes." That estimate was honest: on a large sheet the run took minutes rather than seconds, and the drawing was unavailable until it finished.
What the server did, in plain words:
- It read the sheet. The pale background was stripped out, then the text on the drawing was read so ZeroCount knew where the words were. From that it worked out where the plan areas were and where the legend table sat, and cut each symbol out of the legend as a template.
- It matched every template across the drawing. Each legend symbol was compared against the whole sheet, at every rotation, and every place it matched well enough was recorded.
- It sorted the results. The matches were grouped one group per legend symbol, and handed back as the Detection groups you saw at Stage 2.
You then reviewed the results — deleting false matches, adding the ones it missed, merging groups that turned out to be the same device — exactly as you review results from any capture today.
Why it was replaced
The pipeline did the whole drawing in one go, which was its appeal and its problem. If the legend was unusual, or the sheet was busy, you found out several minutes later and had little to adjust except the boxes and a re-run. A capture that takes a second and shows you the answer immediately is easier to work with, even when you have to run it a dozen times.
The current tools took the useful parts of it:
- Auto Detect Symbols does the whole-drawing pass, but shows you a list to pick from first.
- Drawing a legend does the legend-reading part, on a rectangle you draw yourself, and names each group from the legend code.
- Detect New Symbol does one symbol at a time, in about a second, with a live preview.
Between them you get the same result with far more control over it.
Messages you may still see
If a drawing does still run the old flow, three outcomes are worth recognising:
"No legend found"
"Couldn't detect a legend/symbol table on this drawing. Use Auto Count Symbol to draw around a symbol instead." Nothing on the sheet looked like a symbol table. Some drawings keep their key on a separate sheet, and some draw it in a way the reader does not recognise.
"No symbols in legend"
"A legend was found, but no usable symbols could be extracted from it." The table was located but nothing could be cut out of it as a template — usually a key that is all text, or one printed too small to read.
"Auto takeoff failed"
Something went wrong on the server. The message carries the detail. Count the drawing with Auto Count Symbol instead.
Drawings set up the old way
If you open a project that was counted under the old pipeline, everything it produced is ordinary marker data — Detection groups with markers in them, AoIs, and the legend it found. You can work on all of it with the current tools: rename the groups, top them up with Detect New Symbol, merge duplicates, price them, export them.
Two things to be aware of:
- The AoIs are still active. Anything you count now is restricted to inside them until you delete them. See Areas of interest.
- Resetting starts from scratch. Actions › Reset Drawing clears the markers, zones and AoIs, and the drawing is read fresh from then on. See Resetting a drawing.
Next steps
- Auto Detect Symbols — the current whole-drawing pass, with a list you choose from.
- Drawing a legend — mark the symbol key so the scan reads it row by row.
- Detect New Symbol — the everyday one-symbol-at-a-time tool.
- Areas of interest — what the Stage 1 AoIs still do on an old drawing.
- Takeoff and Review modes — how Stage 2 and Stage 3 map onto today's workspace.