In short
Drawing ballooning — the balloon drawing, or "bubbled print" as some shops call it — is the most thankless step of an inspection report: numbering every dimension on the drawing, one by one, so the measurement table and the print tell the same story. In most small machine shops, this is still done with a highlighter and a pencil, then retyped into a spreadsheet.
How much time does your shop spend drawing balloons on prints — and how many transcription errors does that manual work inject into your reports? This article covers the whole question: what a ballooned drawing really is, why the manual method breaks down, and what automatic dimension extraction changes.
What is a ballooned drawing?
When a customer requires a dimensional inspection report, they are asking a simple question: prove to me that every dimension on my drawing was measured and found conforming. The trouble is the word "every." An ordinary machining print carries thirty, fifty, sometimes over a hundred characteristics: dimensions, tolerances, threads, geometric tolerancing symbols, finish notes. To tie each one to a line of the report, you need an identifier.
That is exactly what ballooning does. You place a small numbered balloon — a circle with a digit — next to each characteristic to be checked. Balloon #12 on the print corresponds to line 12 of the measurement table. The customer can then verify, dimension by dimension, that nothing was skipped. Without balloons, an inspection report is an orphaned list of numbers: nobody can tie it back to the drawing without redoing the identification work.
The vocabulary varies from shop to shop: ballooned drawing, balloon drawing, bubbled print, "numbering the dimensions," "marking up the print." Behind all those words, the same operation — and the same chore.
The traditional method: highlighter, pencil, spreadsheet
In most of the shops I know — and I spent fourteen years in industrial quality in small manufacturers — ballooning is still done entirely by hand. The recipe is the same everywhere, give or take.
You print the drawing. You go through it with a highlighter, dimension by dimension, trying not to skip any. You draw a balloon next to each characteristic, incrementing the number in your head. Then you open a spreadsheet and retype: the balloon number, the nominal value, the tolerance, the minimum, the maximum — computed mentally or on a calculator. Finally you scan the annotated print to attach it to the report.
For a thirty-dimension print, count a good half hour of focused work. For a multi-page package, it's the whole morning. And that's only the preparation: the measurements themselves are still to come. I've detailed elsewhere what that reporting time really costs a shop once you multiply it by the parts shipped in a year.
Manual ballooning isn't hard. It's long, repetitive, and impossible to do perfectly every single time.
Where the manual method breaks down
The problem with manual ballooning isn't that it's complicated — any machinist or inspector-level controller can do it. The problem is that it stacks three fragilities that always end up costing something.
The skipped dimension. On a dense print, the eye tires. A slanted dimension, a stacked tolerance in a corner, a diameter marked "2X": one lapse of attention and a characteristic never gets a balloon. Nobody sees it — until the customer does. A report missing a dimension isn't an almost-complete report: it's a non-conforming deliverable.
The transcription error. Between the print and the spreadsheet, every value is retyped by hand. A 12.7 that becomes 12.1, a ±0.05 tolerance read as ±0.5, a minimum computed in your head on a Friday afternoon: the transcription error is silent, and it contaminates everything downstream. The measurement may be perfect; the report will be wrong.
The drawing revision. This is the costliest fragility. The customer sends revision B of the drawing: three dimensions changed, one added. All the ballooning has to be redone — or worse, the old table gets "adapted" in the hope the numbering holds. Homemade templates, however careful, break at the first revision change. Every revision resets the clock.
None of these fragilities is about competence. They are structural: hand-copying dozens of values from one document to another is a task humans do poorly — not for lack of diligence, but by nature. It is exactly the kind of work that should go to a machine, so the human stays where they are irreplaceable: measuring, judging, signing. That, after all, is everything an inspection report tells those who know how to read one: the rigor of the shop that produced it.
What automatic dimension extraction changes
The good news is that this chore is precisely the kind of problem software can solve today. The principle: instead of spotting dimensions by eye and retyping them by hand, you import the PDF print, and the tool reads the drawing itself — it detects the dimensions, reads values and tolerances, computes minimums and maximums, and builds the measurement table with its numbering. The ballooned drawing and the report are born together, from the same gesture, and stay linked: clicking a line of the table shows the dimension on the drawing.
Concretely, that changes four things.
Time. Where manual preparation of a report took around 45 minutes, extraction brings the whole process down to about 7 minutes. It isn't only the reading that goes faster: it's the whole chain — no more retyping, no more hand-computed tolerances, no more renumbering.
Completeness. A good tool detects the vast majority of dimensions outright — in our case, about 95% — and above all, it shows what it couldn't read instead of making it disappear. The doubtful dimension is highlighted, set aside, submitted to the human. That is the exact opposite of the highlighter-skipped dimension, which leaves no trace of its absence. I've written elsewhere about why reading a dimensioned print has nothing to do with ordinary OCR, and why silent loss is the real test of any extraction tool.
Consistency. Minimums and maximums are no longer computed in your head: they come from a verifiable rule, the same for every dimension — including the general tolerances in the title block (ISO 2768 and friends) that are so easily forgotten on untoleranced dimensions.
Revisions. The new revision of the print gets re-imported. You compare, adjust, and move on — without redrawing fifty balloons in pencil.
And for teams measuring with micrometers and calipers — which is to say the vast majority of small machine shops, with no CMM and no scanner — nothing changes in the gesture: you measure as before, you write down as before. Only the paperwork around the measurement disappears. Some tools go as far as reading the values pencilled on the printed report, so even the final transcription goes away.
The ballooned drawing and the report are born from the same gesture, and stay linked: balloon #12 on the print is line 12 of the table, with no retyping in between.
Manual ballooning vs. automatic extraction: the honest comparison
Let's be frank: automatic extraction isn't magic, and a shop producing three reports a year probably needs nothing more than its highlighter. Here is the comparison the way I would run it for my own shop.
| Criterion | Manual ballooning (highlighter + spreadsheet) | Automatic extraction |
|---|---|---|
| Report preparation time | 30 to 60 minutes depending on print density | A few minutes, human verification included |
| Risk of a skipped dimension | Real, and invisible until the customer finds it | Unread dimensions are flagged, not lost |
| Transcription errors | Inevitable at scale | No retyping: the value read is the value in the table |
| Title-block general tolerances | Applied from memory, often forgotten | Applied by rule, systematically |
| Drawing revision (rev B, C…) | Everything redone | Re-import and compare |
| Cost | "Free" — paid for in controller hours | A software subscription |
| Worth it under 5 reports/month? | Entirely defensible | Rarely justified |
That last line deserves emphasis, because you never read it from software vendors: below a certain volume, the manual method is a reasonable choice. It's between ten and a hundred reports a month that the equation flips decisively — there, every hand-ballooned print is half an hour of a controller's time burned on work a machine does better. And if your instrument tracking also still lives in a binder, the same logic applies to calibration.
Questions to ask before choosing a tool
If you're shopping for ballooning or inspection-report software, a few questions quickly separate real trade tools from seductive demos.
What does it do with dimensions it couldn't read? This is the most important question. A tool that displays "47 dimensions detected" without saying how many it missed reproduces, only faster, the highlighter's flaw: silent loss.
Does it read tolerances, or only numbers? Detecting "12.7" is useless if the tool doesn't understand the ±0.05 next to it, the H7 fit, the position symbol, or the title-block general tolerance that applies by default.
Do the print and the table stay linked? A real ballooned drawing is navigable both ways: from line to balloon, from balloon to line. A frozen export where the link is lost sends you back to the original problem.
Where are your prints hosted? Your customers' drawings are often covered by confidentiality agreements. For a Quebec shop, the question of where prints are hosted, and Law 25, is worth asking before signing, not after.
Is the trial run on your own prints? A demo on a drawing chosen by the vendor proves little. It's on your prints — the old scans, the dense sheets, the slanted dimensions — that the tool has to earn its keep.
In conclusion: the balloon is not the work
The ballooned drawing is indispensable: it's what turns a list of measurements into evidence tied to the print. But drawing balloons was never the work that matters. The work that matters is the measurement itself, and the conformity judgment that follows it. Everything else — spotting, numbering, retyping, computing minimums — is mechanics, and mechanics belongs to machines.
The question I would put to any shop quality manager is this: out of an hour spent producing an inspection report, how many minutes actually go to measuring and judging — and how many to highlighting, numbering and retyping? If the second share exceeds the first, the problem is not your team. It's the tooling.
At Asterion Solutions, automatic dimension extraction from the PDF print — ballooning and measurement table linked, title-block tolerances applied by rule — is at the heart of our trade software for small manufacturers. The trial runs on your own prints.