Annotated drawing set
Flattened source pages and visible markup overlays. A signed package can include a sign-off summary sheet.
A practical, illustrated guide for importing AI-generated drawing sets, marking up issues, checking AEC standards, documenting reviewer decisions, and signing off.
Use the table of contents or search to jump directly to a task. Select any application screenshot to view it at full size.
Version 3 runs at markupvalidate.com. Log in with a verified email to use the secured drawing station. The illustrated screenshots below show the same Version 2 workspace controls with the new Version 3 account actions described here.
Registration starts on the Free plan. Confirm the link sent to your email address before entering the review station.
The top toolbar displays this month's credits. Free: 2 files; Starter $5: 10; Professional $14: 50; Team $49: 200. A PDF or image counts as one file; reopening identical bytes during the same UTC month is free.
Original PDF/image file bytes stay on your computer during the normal workflow. The page count and file hash are sent for credit accounting. The maximum file size is 50 MB.
Text and coordinates extracted from PDF sheets are sent securely to the PHP validation engine. The application displays detected findings and separate manual verification tasks.
Click ☁ Save to store markup data and findings without uploading the original drawings. ☷ Reviews lists saved records. To restore them, first load the matching drawing files in your browser.
After resolving all markups and dispositioning every finding, sign off. The server checks the validation receipt and records the authenticated reviewer name, time and approval. This is not a professional seal.
Select Account in the top toolbar to see usage, upgrade through Stripe Checkout, manage billing, or invite coworkers on the Team plan.
The human sign-off layer for AI-generated PDF and image drawing sets.
Version 2 adds an editable 15-rule AEC standards profile, explicit human-verification tasks, a rule coverage view, named reviewer evidence, and stronger sign-off records. The drawing viewer, markups, and exports remain in one browser application.
Multi-page PDF, PNG, JPG/JPEG drawings. Cloud or annotate issues, inspect sheet standards, disposition findings, and export a documented review record.
It does not parse native RVT or IFC, certify BIM geometry, apply an engineering seal, or cryptographically authenticate a reviewer.
Use the same sequence for concept drawings, coordination packages, and AI-generated deliverables.
Load PDF and image sheets; confirm source files and page order.
Cloud, call out, highlight, and document drawing concerns.
Confirm profile, expected sheets, sheet ID pattern, and required fields.
Run enabled rules; examine findings and human-verification tasks.
Resolve, verify, or waive each finding with name and evidence.
Complete review gates; preserve annotated PDF and review records.
The app uses a source navigator, central canvas, and review panel.
Enter the reviewer name, add drawings, run validation, open this manual, or export the review package.
On the left, browse imported sheets and find Restore review JSON.
In the center, choose a drawing tool, adjust ink/weight, zoom, or pan.
On the right, switch between Markups, Validation, Sign-off, and Standards.
Start by importing all source sheets you intend to review.
Use the top-bar button, Add PDF / images on the left, or drag files into the app.
Supported inputs: PDF, PNG, JPG and JPEG. Multi-page PDFs are separated into reviewable pages.
Compare thumbnails, source names, and page counts with the project issue list.
Keep the imported source drawings beside exported review records. These source bytes are not saved in review JSON.
Move quickly between sheets and inspect drawing details.
Select a source thumbnail or use the left/right arrow keys when not typing.
Use the +, − and Fit controls in the canvas toolbar.
Choose the pan tool or hold Space and drag on the sheet.
Select a markup or finding to go to its sheet, and an approximate location when extractable.
Select a tool, then draw directly over the source page.
Enter the reviewer name before creating markups to attribute each record.
Choose Ink color and line Weight from the toolbar.
Select a tool, drag across a location, and provide callout text when prompted.
Return to Select to move or edit a selected item. Use Undo, Redo, or Delete as appropriate.
Track markups across all sheets and update their status.
Choose a sheet and/or open/resolved status.
Select a record or its locate control to jump to that drawing.
Close a markup only after its underlying issue or action is addressed.
Reopen a resolved markup if the issue returns or verification changes.
Export the review JSON or CSV to retain a portable audit of the review.
Version 2 includes the AEC Drawing Set QA default profile, version 1.0.
Open the Standards tab in the right review panel.
Review Profile name and Version; use project-specific versions for different standards sets.
Confirm naming pattern, expected sheets, title-block fields, text height and dimension notation.
Use Save standards profile, then Run validation after any changes.
Set office- and project-specific inputs before running validation.
| Setting | Version 2 baseline | What to confirm |
|---|---|---|
| Sheet ID regex | ^[A-Z]{1,2}-\d{3}$ | Edit to match your firm's sheet numbering. |
| Expected sheets | User-defined, initially blank | Enter actual sheet IDs, one per line. |
| Title block fields | Eight fields | Project name/number; sheet number/title; date; scale; drawn/checked by. |
| Minimum text height | 0.125 plotted inches (9 pt) | Confirm applicable plotting/text standards. |
| Dimension notation | Mixed-notation screen or defined format | Verify drawing units and dimension style visually. |
Rule definitions remain editable, but modifying text does not invent new detection logic.
Select a rule title in the Rule definitions area.
Choose enabled/disabled and error, warning, or info.
Adjust What to check, Data source and Fail condition to describe office requirements.
Select Save standards profile.
Use Import JSON, Export JSON, or Restore 15-rule default as needed.
A searchable, expandable catalog of the supplied version 1.0 profile.
The list below reproduces each rule's exact intent, data source, and failure condition from the provided profile. The method tag describes the Version 2 application's practical assessment capability, not a guarantee of compliance.
Every sheet carries a filled title block with all required fields: project name, project number, sheet number, sheet title, issue date, scale, drawn by, checked by.
Text extracted from the title block region of each plotted PDF page.
Any required field is missing, blank, or a placeholder (e.g. 'TBD', '---'). One finding per sheet per missing field.
Every sheet number matches the office numbering pattern in sheet_naming_pattern, and no two sheets share a number.
Sheet numbers from the title blocks and the sheet list.
A sheet number breaks the pattern, or a duplicate number exists in the set.
Every sheet on the user's expected sheet list exists in the set.
User-defined expected sheet list vs. sheet numbers found in the set.
An expected sheet number has no matching sheet in the set.
All annotation text meets the minimum plotted text height (1/8 inch at the sheet's stated scale).
Text runs on each page, font size, and the sheet's stated scale.
Any text that plots smaller than 0.125 inches. Lists the sheet and the text string.
Keynotes, tags, and symbols match the office legend and one finished reference sheet in format, placement logic, and symbol choice.
Annotation extracted from each page vs. the office legend the user provides.
A keynote format, tag style, or symbol deviates from the legend. Cite the sheet and element.
No element is tagged twice with the same tag type.
Tag instances extracted per page, keyed by target element.
The same element carries two identical tags. Cite the sheet and the element.
Every tag points at an element that exists in the set.
Tag leader endpoints and tag target references vs. elements found on the sheets.
A tag's target element cannot be found. Cite the sheet and the tag ID.
Every dimension string agrees with the geometry it describes.
Dimension strings and the geometry they annotate on each page.
A dimension string disagrees with the measured geometry. Cite the sheet and the dimension.
One dimension style (arrow type, text size, precision, units) across the whole set.
Dimension style samples from each sheet.
More than one dimension style detected. List the sheets using each.
Values on element tags (door, window, room tags) match the corresponding schedule rows.
Tag values vs. schedule pages the user marks.
A tag value disagrees with its schedule row. Cite the tag sheet, the schedule sheet, and the element.
Every revision cloud on every sheet has a matching entry in the revision register: number, date, description, affected sheets.
Revision clouds and delta numbers on the sheets vs. the revision register the user provides.
A cloud or delta number has no register entry, or the register lists an affected sheet that shows no change.
No sheet carries two different revision numbers for the same change.
Revision numbers and descriptions across the sheets and register.
The same change is numbered differently on different sheets or in the register.
Plans, sections, and elevations agree with the modeled geometry: plan dimensions match the model, section cuts reference existing views, elevations match the current design option.
The plotted set plus the model data the user provides; where model data is unavailable, cross-sheet agreement is checked instead.
Any disagreement between drawing and model (or between sheets when no model data is given). Quote both sides: what the drawing says and what the model says. This is the highest-severity category; these errors cost money in the field.
Layer names, view naming, symbol usage, and text styles follow the office standard. The standard is a human document; the app only reports who followed it.
Layer lists, view names, and symbol catalogs extracted from the set vs. the office standard the user provides.
Any layer, view name, symbol, or text style outside the standard. One compliance report per sheet.
The set does not issue until the QA first pass is clean and every finding is dispositioned by a named reviewer: fixed, accepted with reason, or not an issue.
The findings list and the reviewer's disposition log in the app.
Sign-off is attempted while any finding is undispositioned.
Detected findings and manual tasks must be kept distinct.
Click Run validation at the top or from the Validation tab.
Open detected items for potential nonconformances supported by available PDF text.
Open Verify manually items where PDF evidence cannot conclusively establish compliance.
HUMAN CHECK, FINDING, NO FINDING, and GATE ACTIVE indicate review capability or condition.
Check unresolved items after automatic checks and manual review.
Every issue includes a scope, severity, status, description, and action.
Use page, status and severity filters to focus your review.
Error, warning and info indicate impact; VERIFY marks human-review requirements.
Click the issue or Locate to jump to the relevant sheet or approximate region.
Compare the reported observation to the actual drawing and applicable standards.
Resolve/Verify or Waive the finding with a reviewer name and written documentation.
Human verification tasks are not automatically passed.
Open an item labeled VERIFY in the Validation findings list.
Provide the person responsible for the technical check.
Include the sheet, element or location, comparison source, and what was checked.
Enter a meaningful note and close the manual-review task.
A waiver is acceptance of a known exception, not evidence of technical compliance.
Choose Waive for an issue eligible for documented acceptance.
Cite the sheet, item, approval basis and consequence or limitation.
Use a reviewer authorized to accept that exception.
Waived items remain in the review record and can be revisited.
Version 2 blocks sign-off while any required review item remains open.
Rerun validation after profile or drawing changes.
Resolve each open annotation in the Markups register.
Resolve, verify, or waive every finding and manual-review task.
Provide the named responsible reviewer in the Sign-off tab.
Approve the dialog, then verify the recorded date/time, counts, profile and waiver summary.
Use all four complementary formats to preserve drawings and review evidence.
Flattened source pages and visible markup overlays. A signed package can include a sign-off summary sheet.
Sources, markups, profile settings, findings, coverage, reviewer dispositions and sign-off history.
Spreadsheet-ready markup and validation rows with status, reviewer evidence and acceptance reasons.
Reviewer, date/time, drawing scope, counts and dispositions for issue records.
Review JSON restores structured review history, not the underlying drawing file bytes.
Launch the HTML review station in your browser.
Reimport original PDFs/images using the matching source names and page numbers.
Use Restore review JSON at the bottom of Sources.
Confirm source mappings, markups, findings and evidence.
Revalidate if drawings or standards changed.
Do not assume a restored session carries a live authenticated approval.
A sample workflow for a two-sheet architectural set.
Import A-101 and A-102 images, check thumbnails, and enter the reviewer.
Cloud a door swing on A-101 and add a callout requesting clearance and door schedule verification.
Configure the actual expected sheets, save project profile version and review title-block requirements.
Image-only sheets produce manual tasks for checks that need text or element evidence.
Inspect title blocks, compare dimensions, schedules and relevant revision register.
Resolve the cloud after correction. Verify manual tasks with locations, names and evidence; explain accepted waivers.
Confirm the review gate shows zero open markups and findings, then sign.
Save annotated PDF, review JSON, CSV and sign-off TXT alongside originals.
A practical quick reference for reviewers.
Check internet connectivity for the PDF.js CDN, reopen the HTML and retry.
Use a searchable properly exported PDF and provide office standards, legend, schedules and model evidence.
Enter actual expected sheet numbers; example values in the profile are not enforced automatically.
PDF text detection is heuristic. Inspect the specific field and record the location before resolving.
Run validation and disposition all open markups and findings, including manual tasks.
Reopen source files, restore previously exported review JSON and sign again where required.
Download the manual HTML beside the app HTML, using the original filenames. Or use the combined ZIP package.
Return to the application with your manual available in a second tab.