Skip to main content
Getly

Blender Add-ons

BakeProof — Where Your Bake Lies, and the Cage That Fixes It

Measures where every bake ray actually lands, says what share of your texture was sampled from the wrong place, finds the smallest cage that fixes it — and proves the fix with a second measurement

$59.00

or $14.75 x 4 installments

Product Description

Your normal map looks complete. A third of it can still be wrong.

A bake never fails loudly. When the cage is too small, the ray from the low-poly misses the high-poly and keeps going until it hits something — the far side of the model, or nothing at all. The texture still fills every island. It looks finished. It is wrong exactly where the low-poly sits inside the high-poly, and you find out in the engine, or from a client.

BakeProof bakes where every ray actually lands, with your own bake settings, and tells you what share of the texture was sampled from the wrong place — then finds the smallest cage that fixes it, sets it, and measures again so the fix is proven rather than assumed.

One number, and where it is

Measured on a real CC0 barrel with a reduced low-poly and the default extrusion of 0: 34.74% of 33,257 texels come from the far side of the model. "Show The Map" lays that out on your low-poly UVs: red from the far side, yellow skewed, green right.

The cage that fixes it — and the proof

"Find The Right Cage" searches for the smallest extrusion that is as good as any extrusion can be. On the same barrel it finds 2.049 cm. "Use This Cage" sets it in your bake settings and measures again: 34.74% → 0.00% from the far side, written down as a before-and-after line, not as a promise.

Smallest matters. A cage far bigger than needed catches the wrong surface in cavities and between parts; the search is checked to return an extrusion at which the error is zero while 10% less is not.

It says what a cage cannot fix

  • Skew. A bigger cage removes the far-side hits but makes rays land at a slant on steep walls. On a test sphere with a raised cap the far side goes from about 11% to 0.00% — and 4.09% of texels become skewed. BakeProof reports that as its own finding and says the fix is in the low-poly shape (an edge loop or bevel), not in the cage.
  • Holes. Where the ray finds nothing at any extrusion — a gap in the high-poly, or the low-poly poking through it — the remainder is named separately, with the texel count, instead of sending you back to search again.
  • Overlapping UVs. Texels covered by two islands are baked twice and the last island wins; that is reported with its share of the texture.

It measures your bake, not a default one

Your Extrusion, Max Ray Distance and cage object are the ones measured. The skew line is yours to set, as a percentage of the model size. Lengths are shown in your scene units.

Your scene is left as it was

The measurement runs on temporary copies with their own materials and images. Afterwards your objects, materials, images, bake settings, render engine, selection and active object are exactly as they were — that is checked, not assumed. The only new thing is the map. The only button that changes your scene is "Use This Cage", and it can be undone.

What it refuses to do

  • No UV map on the low-poly, no high-poly selected, an active object that is not a mesh: refused, with the reason in words.
  • A UV map that covers no texels at the chosen resolution is "not measured", never "0% wrong".
  • A UV map that covers only a handful of texels gets its share flagged as too few to trust, with what one texel is worth.
  • A cage found for one pair is not applied to another.

Why the numbers are trustworthy

98 automated checks, zero problems, on Blender 4.2.9, 5.1.0, 5.2.0, 5.2.1 and 5.2.2. The numbers are checked against geometry whose answer is known in advance. Deliberate breakages of the measuring code were introduced one at a time, and the checks caught every one — a suite that cannot fail proves nothing. Run on 45 real CC0 production models: at the extrusion saved in each file, 40 of them took part of their texture from the far side — median 7.9%, worst 57%. At the cage it found, 43 of the 45 come down below 0.4%, and what stays is named as a place where no cage helps. The two that do not: a chain-link fence of thin wires keeps 4.5%, and a tree whose UV map covers 7 texels is reported as too few texels to trust. No model's materials or slots were changed.

Blender 4.2 and newer. Cycles (the measurement uses Blender's own bake). Sidebar tab: BakeProof.

License

GPL-3.0.

What you get

1 file · 75.1 KB

  • bake_proof_v1_0_0.zipZIP · 75.1 KB
Package checkPassed
PublishedOct 2, 2026
File size75.1 KB
File formatZIP
Versionv1.0

Tags

bakingbakenormal mapcageextrusionray distancehigh polylow polygame assettexturevalidationproof
N
NeonAnvil

Frequently asked questions

Do I get access instantly?

Yes. This is a digital product - you can download it right after checkout, straight from your library.

Can I use it for commercial projects?

Licence terms are set by the seller. Check the product description before you buy - it states any limits on personal, commercial or client use.

What's your refund policy?

You're covered by a 30-day money-back guarantee. If something isn't right, request a refund from your order and the Getly team reviews it.

What file formats and sizes will I get?

You'll download 1 file - ZIP - 75.1 KB in total.

Do I get free updates?

Yes - when the seller ships a new version, you can re-download the latest files from your library at no extra cost. Check the changelog for recent updates.

Price

$59.00