Your first Drawing QA run
Run the analysis, read the findings, and know which ones to trust first.
This walks through one full run on one set, and explains what you are looking at when it finishes.
Run it
Open the set from Drawing QA/QC and choose Analyze.
Each sheet is read by a vision model, which pulls out what is on the page: the title block, grid lines, tags, dimensions, callouts, view titles. The checks then run over that extraction, first within each sheet and then across the set.
A large set takes time. You can leave the page while it runs.
The viewer
A set opens with the sheet on the left and its findings on the right.
Selecting a finding highlights the region of the sheet it came from. Use it. A finding you cannot see on the drawing is a finding you should not act on, and the highlight is how you tell a real omission from a mis-read.
Severity
Findings carry one of five levels.
| Severity | What it means |
|---|---|
| Critical | The set cannot be issued like this |
| High | A real error that will cause a question or a rework |
| Medium | Worth fixing before issue |
| Low | Minor, catch it when convenient |
| Cosmetic | Presentation only |
Severity is about consequence on site, not about how confident the check is. A missing sheet number is high because the sheet cannot be filed or referenced, not because Bimiq is certain.
What actually gets checked
The checks fall into two groups, and the second is the one worth your attention.
Within one sheet. Things that are wrong on the page by themselves. A sheet with no number in the title block (A1), no title (A2), or scaled content with no scale stated (A3). A grid line with no label (B2), a tag with nothing in it (B3), a dimension with no value (C1).
Across the set. Things that are only wrong in company, which is where a human reviewer runs out of patience first.
- A callout that points at a sheet that is not in the set (
D1) - The same sheet number used twice (
D3) - Grid labels that disagree between disciplines, so architectural grid C is structural grid D (
XD1)
Then, if you have standards loaded, each sheet is checked against the rules extracted from them. Every project starts with the built-in standard, so this runs even before you upload your own.
Where the false positives come from
Nearly all of them come from extraction, not from the rules.
A dimension flagged as having no value usually does have one, in a font or at an angle the model read poorly. A grid label flagged as missing is often just faint. This is why the highlight matters: check the region before you act, and dismiss what is wrong.
Findings on a scanned set are less reliable than on a plotted one, for the same reason.
Two things you can run afterwards
- Deep XD goes further on cross-discipline coordination between the sheets in the set.
- Reconcile with model compares the sheets against the Revit model they came from, catching sheets that no longer match the model. This one needs the Bridge and the model open.
Next
The full module reference is in Drawing QA/QC and Review.