Light is getting into your interior through a gap you cannot see.

A bright line along the skirting. A glow in the corner of the ceiling. A room that should be dark at night and is not. The usual advice is to make the walls thicker, add a bevel, stack a Solidify on everything — and render again to see if it helped. Nobody tells you where the light gets in, or how wide the opening is.

LeakProof finds every gap between the parts of your scene that lets outside light into the room your camera sees, names its width, closes it, and proves with two renders how much of the light in your shot came through.

What actually leaks — measured, not assumed

A closed room of six separate planes, a sun outside, a camera inside, rendered in Cycles:

• Planes meeting edge to edge: 0% of the frame lit. A butt joint does not leak.
• The same room with a 1 mm gap at every seam: 7.2% of the frame lit.
• With a 1 cm gap: the whole frame — light bounces carry it everywhere.
• Thick walls without gaps: 0%.

So thickness is not the cure and a butt joint is not the problem. A real gap is. LeakProof looks for exactly that.

Width, place, and whether it really leaks

Every open edge and every sharp edge is checked against the parts around it. Where two parts nearly touch, the gap is measured — "Light leaks between Wall_N and Ceiling — a gap 1 mm wide" — and the 3D cursor goes to it on "Show Me". A gap only counts as a leak when one side of it opens to the sky and the other side lets light onto surfaces your camera sees.

That last part is what keeps it quiet where it should be quiet:

• A box standing a millimetre above the floor is a gap, but the room is on both sides of it: named as a gap that does not let light in, not as a leak.
• A window is wider than your gap line: it is a window.
• A chair pushed against a wall with a real crack behind it does not become a second leak: light that reaches it through the wall's crack is that crack's leak.
• Foliage and scatter — thousands of separate pieces — block light in the measurement, and are named as not checked for gaps inside them.

Seal it — and prove it

"Seal The Leaks" closes each leaking gap with a seal in the material of the wall beside it, all in one object in its own collection, and measures again: leaking gaps 12 → 0 on the test room. "Prove It With A Render" then renders your camera shot without the seals and with them, with the same settings and seed, and tells you what share of the light in the shot came through the gaps — next to the render noise, so a change smaller than noise is never called a leak. In a closed test room with 1 mm gaps, over 90% of the light in the shot was leak. With a window and the sun coming through it, 2 mm gaps were 0.6% of the shot's light — and the window's light stayed: the seals close gaps, not openings. The map shows where the leaked light fell, in red.

Your render settings are put back exactly as they were. "Remove The Seals" deletes only what LeakProof made — an object of yours moved into that collection stays.

Why the numbers are trustworthy

68 automated checks, zero problems, on Blender 4.2.9, 5.1.0, 5.2.0, 5.2.1 and 5.2.2. Deliberate breakages of the measuring code were introduced one at a time, and the checks caught every one. Every one of 45 real CC0 production models was put into a closed room and into the same room with 1 mm gaps: 0 leaks in every closed room, exactly the 12 real seams in every gapped one — not one model added a false leak or hid a real one. The same answer at a scene scale of 0.01, 1 and 100. Modifiers are measured as rendered: a wall shifted by a Displace modifier is found where it is, not where its mesh is.

Blender 4.2 and newer. Cycles. Sidebar tab: LeakProof.

License

GPL-3.0.