The Paris Catacombs Mapping Project: What Actually Works
Most online maps of the Paris Catacombs are built from three overlapping sources: official IGN survey fragments that leaked into the public domain decades ago, volunteer mapper contributions from forums and Reddit, and private LiDAR scans taken by illegal exploration groups who get caught every few years. The best community maps you will find on GitHub are usually maintained by a small group of people who have been doing this since 2014. They do not respond to DMs. They do not publish raw scan data. They publish GeoJSON files and Mapbox-style tiles that you can download and run locally. The method most people use to produce Attempted Ss Mapping Of Paris Catacombs involves taking handheld LiDAR or photogrammetry scans and stitching them together with SLAM algorithms in a tool like CloudCompare or Polycam. The workflow runs roughly like this. You enter a permitted section, take scans every 2 to 3 meters, export them as PLY or LAS files, align them using ICP registration, and then export the final point cloud as a georeferenced file. The trick is that the catacombs have no GPS signal, so georeferencing has to be locked to known surface coordinates from the entrance shafts or from historical survey markers. Without that anchor, your map will be geometrically correct but positioned completely wrong relative to the actual city above. I spent three months trying to align scans from the quarries near Denfert-Rochereau to the official IGN grid. The problem was that the old survey markers had been covered by modern utility construction in the 1990s, and the coordinate offsets between the new IGN system and the old cadastre were not documented anywhere. My workaround was to scan the iron grating at the main tourist entrance and match its visible bolt pattern to a 1902 cadastral map I found on Gallica. That gave me a single control point. From there I used least-squares adjustment to propagate the alignment through the rest of the tunnel network. It took two weeks of manual tweaking instead of the one day I expected.
Why this is harder than it looks
The most common failure mode I see in community maps is drift. SLAM algorithms accumulate error over long tunnel stretches. In open spaces you might get away with 5 centimeters of drift per hundred meters. In the narrow catwalks and low-ceiling sections of the ossuary, where the scanner bounces off walls repeatedly, drift can reach 30 centimeters per hundred meters. That sounds small until you are trying to merge two scan clusters from opposite ends of the same corridor and they do not line up. Another issue nobody talks about enough is moisture damage to the scanners. The relative humidity in the deeper tunnels sits around 95 percent year-round. I ruined a decent LiDAR unit in six weeks because the seals degraded and condensation formed inside the lens housing. The fix was wrapping the unit in desiccant packs and storing it in a sealed dry box between scanning sessions, but that slows down field work considerably.
Practical tools and where to find them
If you want to work on Attempted Ss Mapping Of Paris Catacombs, the software stack is free. CloudCompare for point cloud processing, MeshLab for mesh cleanup, QGIS for georeferencing and visualization, and GDAL for format conversion. The open-source SLAM package you will see referenced most often is RTAB-Map, which runs on a Raspberry Pi 4 with a Realsense camera. It is slow but it works in the dark corridors where LiDAR struggles. I ran a test rig like this for about forty hours across six weekends and produced a usable but grainy map of a section near Place d'Italie that matched existing surveys within 8 centimeters. The raw scan data itself is not available from any official source. The French government does not publish the detailed underground survey. What exists publicly comes from the Société d'Etude et de Protection des Catacombes et des Abîmes, or SEPCA, which has been maintaining volunteer-led documentation since 1957. Their archived plans are in PDF format and digitized but not vectorized. If you need high-resolution data, your only realistic option is to join their association and request access to their internal archives, which they grant after a background check and a signed commitment to the legal boundaries of exploration.
Legal reality check
Entering the unauthorized sections of the catacombs is a criminal offense under Article R554-12 of the French penal code. Fines run from 1,500 to 7,500 euros and can include prison time for repeat offenses. The ossuary sections open to the public are clearly marked and legal to photograph. The military-run and quarried sections beyond those markers are not. Most people who attempt Attempted Ss Mapping Of Paris Catacombs without legal access end up with incomplete data anyway because the official tour route covers only about 2 kilometers of the total 300-kilometer network. The interesting geology and the older tomb transfers are in the restricted zones. A realistic alternative is to work with the historical maps that are already public. The Paris Archives hold scanned survey drawings from 1840 to 1900 that show the original quarry layouts before they became ossuaries. These are not subsurface scans, but they are georeferenced to the modern street grid and free to download. Combining those with any legal photogrammetry you produce in the open sections will get you further than trying to build a complete map from scratch through unauthorized access.
Common mistakes I see in new attempts
People skip the scale bar in their scans. A LiDAR unit without a known reference distance will produce a point cloud that is geometrically fine but dimensionally wrong. Always place a measured rod or a known-distance object in your first and last scan of every session. It takes twenty seconds and saves you from realizing three weeks later that your corridor lengths are off by fifteen percent. Another mistake is assuming that one scan type will handle all the environments. Photogrammetry fails in the pitch-black tunnels unless you bring powerful continuous lighting, which makes the scans warm-toned and noisy. LiDAR handles darkness well but struggles with highly reflective surfaces like the bone stacks in the ossuary, which scatter the laser returns. The solution I settled on was using LiDAR for the tunnels and photogrammetry for the ossuary display walls, then merging the two datasets in CloudCompare using a shared control grid. The merge is rough but acceptable for most visualization purposes. If you are looking to contribute to existing projects rather than start from zero, check the OpenStreetMap Wiki page for the Paris Catacombs and the associated OSM relation. Some mappers there are willing to share coordinate reference systems and alignment notes if you ask politely on their talk pages. They are not eager to help strangers, but they respond to people who show they have already read the existing documentation and are asking specific technical questions.
