Doo Interactive · EDF & CEA · 1999–2000 · Case study

Nuclear sites on film

Some 450 spheres, one slide at a time

1999–2000: some 450 360° spheres shot on slide film inside EDF nuclear power plants and CEA research centres, scanned one circle at a time, between the Biron CD-ROM and the Accor platform.

Between the Biron and Lascaux CD-ROMs and the Accor platform, there were two years when 360° photography became an industrial job, on film, in some of the most controlled places in France: EDF nuclear power plants and the research centres of the CEA.

My customer was Abstract Interactive, in Pessac, who built image-based databases for EDF, the CEA and GIE INTRA. My team went on site with the cameras, then scanned and assembled the spheres. Some 450 of them, from 1999 to 2000, every one of them on slide film.

Rebuilt in 2026 from the original 1999 and 2000 film scans. Drag to look around.

~450

spheres on film, 1999–2000

56

rolls of slide film at Civaux alone

163

spheres across five CEA centres

2 min

to scan each fisheye circle

Why film

No digital camera in a controlled zone

Digital cameras existed by then. We would not take one inside. Radiation could damage the sensor, and anything found contaminated on the way out stays inside the plant. A camera body is easier to lose than a roll of film.

So we went in with film cameras and a big bag of film. Every visit came with access passes, dosimeter films and radiological surveillance. At Civaux, two of us spent five days on site and came out with 56 rolls of slide film.

The method

From slide to sphere, one step at a time

Each sphere meant four fisheye views, one every 90°, and most of the time two or three exposures of each. That is eight to twelve circles to scan for a single sphere, at about two minutes each, before any stitching could start. Multiply by 450 and you get thousands of scans and well over a hundred hours at the scanner.

We learned to work from mounted slides rather than film strips: the scanner could be fed in batches, and picking the right exposure was quicker. Then every image went through Photoshop: from fisheye to the six faces of a cube, then to an equirectangular sphere, with the tripod painted out. Ten minutes per image if the picture had been taken properly, up to an hour if not. All of it on Windows 98 and Windows 2000 machines, on the processor alone. There was no GPU. A heavy process.

STEP 1
Shoot
Four views, one every 90° on a rotator machined to ±0.5°, usually bracketed at two or three exposures. Fuji 100VS slide film.
STEP 2
Develop
Films developed at the lab and mounted as slides. Slides, not strips: easier to sort, easier to pick the right exposure, and the scanner could be fed in batches.
STEP 3
Scan
Every fisheye circle scanned on its own, about two minutes each, before any stitching could start.
STEP 4
Retouch
Each image reworked in Photoshop, from fisheye to cube faces, then to equirectangular, tripod removed. Ten minutes per image when the shot was right, up to an hour when it was not.
STEP 5
Deliver
Equirectangular spheres at 3000×1500, or a cube of six faces for VRML; iPIX and LivePicture files; CD-ROMs.
Inside the files

Two circles, one master, one sphere

In a research reactor hall, one fisheye looks down into the reactor pool and its blue glow, the other up at the dome and the polar crane. Assembled, they made the editable master that went to the client; in 2026 the same two slides became the sphere at the top of this page.

Looking down into the reactor pool. Slide scan, 1999.
Looking down into the reactor pool. Slide scan, 1999.
Looking up at the dome and the polar crane.
Looking up at the dome and the polar crane.
The editable master: both hemispheres remapped for retouching before delivery.
The editable master: both hemispheres remapped for retouching before delivery.
Inside a hot cell, 2000.
Inside a hot cell, 2000.
The opposite view, out through the cell door.
The opposite view, out through the cell door.
The sites

EDF power plants and CEA research centres

The CEA centres were delivered in December 1999 on ten CD-ROMs. Civaux and GIE INTRA followed in 2000. GIE INTRA is the robotic intervention group of the French nuclear industry; its database was used by maintenance teams and by the operators of emergency robots.

Dec 1999
CEA Saclay
41 spheres
Dec 1999
CEA Grenoble
36 spheres
Dec 1999
CEA Cadarache
38 spheres
Dec 1999
CEA Valrhô
31 spheres
Dec 1999
CEA Fontenay-aux-Roses
17 spheres
1999–2000
Civaux nuclear power plant (EDF)
123 spheres and 45 hemispheres in buildings BAN and BK, from 56 rolls of slide film
Mar 2000
GIE INTRA
60 spheres of the BAS building, for a maintenance and intervention database
Mar 2000
CEA Orsay
A half-day shoot, 7 spheres
Dec 1999
Blaye and Golfech (EDF)
Two week-long sessions on site with the teams
The bridge

Between Biron and Accor

1998–1999
Biron & Lascaux →
Film, a few dozen spheres for two CD-ROMs. The method is born.
1999–2000
Nuclear sites
Film at industrial scale: some 450 spheres, thousands of slides, scanned one by one.
2000–2001
Accor →
100% digital: Nikon cameras, an online platform, 750+ hotels.

The nuclear years are where the protocol was hardened: nothing left to chance, every room documented, every frame accounted for. When digital cameras became reliable, we carried the same discipline straight into the Accor hotels.

Credits

Doo Interactive for Abstract Interactive, 1999–2000. End clients: EDF, the CEA and GIE INTRA.
Photography, scanning and assembly: Stéphane Jacob, with the Doo Interactive team.