Why does a normal map make a 3D model look wrong?
Disconnect the normal map first. If the defect disappears, check its color space, normal-map node, UV selection, and tangent-space convention before editing geometry. Blender's glTF exporter expects a tangent-space normal map with positive Y and a Non-Color image input. If the defect remains without the map, investigate mesh normals, face orientation, material settings, or lighting instead.
Compare shading with the normal map disconnected
Normal maps change the surface directions used for lighting; they do not rebuild the silhouette. Compare the same mesh, camera, and lighting with only the map connection changed. A dent that becomes a bump suggests a convention mismatch, but it is not proof on its own. A dark patch that remains without the map points away from texture interpretation. Keep face orientation and custom split normals as separate diagnostic branches rather than treating every shading artifact as one normals problem.
Check data color space, tangent space, and convention
In Blender, inspect the Image Texture feeding the Normal Map node. For a glTF handoff, set the image to Non-Color, use Tangent Space on the Normal Map node, and connect its output to the Principled BSDF normal input. Confirm that the map uses the UV layout it was authored for. Compare the receiving shader's expected convention with the source. If a channel conversion is genuinely required, make it once in the documented pipeline rather than flipping channels both in the image and importer.
Separate surface detail from geometry defects
The documented glTF setup does not establish every engine's shader convention. Changing image color space cannot repair a map baked against incompatible geometry or UVs. UV revisions, triangulation, and differing tangent calculations can affect the result, so preserve the mesh and material used for the original bake when available. Increasing normal strength can exaggerate a mismatch. No specific Objectify export artifact was tested for this guide, and an attractive viewport is not evidence of identical shading across tools.
Retest the same material in the receiving renderer
Inspect the same raised and recessed detail under a neutral light from several directions. Compare the source and destination with matching camera framing, then toggle the map in both. Look at UV seams and hard edges as well as the central surface. Export and reimport the corrected candidate to ensure material interpretation survives delivery. Record image color space, normal-space convention, any channel conversion, and importer version so another correction does not accidentally reverse the fix.
Evidence and review
Documentation-based guide · Documentation review ·
Documentation-based guidance, not a tested Objectify output or benchmark. Importer support and settings depend on your tool version; validate your actual exported file before production use.