The revision cloud is not the revision
Every drawing that comes back to site carries clouds and a revision note saying what changed. The clouds are drawn by a person, at the end of a job, usually in a hurry, and they are the only record most people ever check. They are also wrong more often than anyone admits, in both directions at once.
Changes get missed, because the person clouding is not always the person who made the edit. And clouds get left behind from the previous revision, so an area gets marked that has not moved since. Read the cloud as a statement of intent, not as a record of fact. The only thing that knows what actually changed is the geometry.
That is all this tool does. It draws both revisions at the same size, on the same grid, and colours every pixel by whether it appears on the old sheet, the new one, or both. What comes out is not an opinion about the drawing, it is the drawing.
What the colours mean
- Red was on the old revision and is not on the new one. It went.
- Blue is on the new revision only. It arrived.
- Grey is on both, unchanged, and deliberately faint so it stays out of the way.
A line that moved shows as both: red where it was, blue where it is now. That reads better than it sounds, because the two sit next to each other and the offset is the thing you want to see.
The one thing that makes the result meaningless
If the two revisions were plotted at different sheet sizes, or one has been cropped, then every line on the drawing is in a different place and the overlay reports that all of it changed. This is the failure mode worth knowing about, because it does not look like an error. It looks like a catastrophic revision.
The tool compares the proportions of the two sheets before it compares the content, and it tells you when they do not match rather than handing you a red rectangle and letting you draw the wrong conclusion. A sheet that changed from A1 to A0 is fine and gets overlaid normally, because the A series is one shape all the way up. A landscape sheet reissued portrait, or a PDF cropped to the drawing frame, is not fine and you will be told.
Where this earns its keep
Checking a reissue before you fabricate off it. If you have already cut steel to revision C and revision D lands the same afternoon, the question is not what the revision note says, it is whether anything you have already made has moved. Ninety seconds with an overlay answers that properly, and the PNG goes straight into the email asking the question.
It is also worth running on a drawing that comes back marked no change, for information only. That happens a great deal, it is usually true, and when it is not true it is expensive.
One honest limit. Identical line work means nothing on this sheet moved. A revision can change a note in a schedule, a specification clause, the title block or a completely different sheet, and none of that shows up here. That is why the result says identical line work rather than no changes: one is a fact about the drawing in front of you, and the other is a promise about the whole revision that no overlay can make.
Common questions
How do I compare two revisions of a drawing?
Drop the older revision and the newer one in, pick the sheet in each, and overlay them. Anything only on the old revision comes out red, anything only on the new one comes out blue, and everything present on both goes grey. You get a PNG you can send to whoever issued it.
Why not just look at the revision clouds?
A cloud tells you where the drafter says it changed. The overlay tells you where it did. Clouds get missed on a rushed reissue, and clouds get left on from the revision before, so a clouded area is evidence of intent rather than of change. Check the clouds against the overlay and you find both the changes nobody clouded and the clouds that mark nothing.
The whole drawing has come out red and blue. What went wrong?
Almost always that the two sheets are not the same shape. If one revision was re-plotted at a different size, or cropped, every line lands somewhere different and everything reads as changed. The tool checks the sheet proportions and says so when they differ. Get like for like copies and run it again.
Does it work on a scan?
Yes, but the scan has to be square on and at the same crop, or the skew shows up as change along every line. A scan against a native PDF plot is usually not worth running. Two scans from the same machine usually are.
Why red and blue rather than red and green?
Because red and green is the common colour blindness axis, and around one man in twelve cannot separate them reliably. On a construction drawing that is not an edge case. Red and blue also come out as visibly different densities on a mono photocopy, and these get printed.
It says identical line work. Does that mean nothing changed?
No, and the wording is deliberate. It means nothing on this sheet moved, arrived or went. A revision can change the title block, a note in a schedule, a specification clause or another sheet entirely, and none of that shows here. Identical line work is a fact about the drawing. Nothing changed is a claim about the revision, and this tool cannot make it.
Are the drawings uploaded?
No. Both drawings are rendered and compared in your own browser. Nothing is sent anywhere, which matters on drawings that are commercially sensitive or under an NDA.