Free baked-lighting detector
Remove shadows from a texture and estimate neutral base color
Detect baked lighting in a photo texture for free. Inspect shadow and highlight heatmaps, estimate neutral base color, preview relighting, and create a cleaner albedo map.
Free: score, shadow and highlight heatmaps, neutral estimate, source warnings, relighting proof, seam preview, and detector report. Starter: editable correction, exclusion masks, and full-resolution corrected PNG export.
Analyze a texture · Read the complete Material Delighter guide
How do you remove baked lighting from a photo texture?
Estimate broad illumination separately from the surface's smaller color and detail changes, lift likely shadow regions, compress likely highlights, and verify the corrected texture under new neutral lighting. PLAYTEX AI keeps the diagnosis inspectable with separate heatmaps, a 0–100 baked-lighting score, a neutral base-color estimate, editable protection masks, and a 2 × 2 seam test.
A result from one RGB photo is an estimated base color, not measured physical albedo. Surface reflectance, geometry, exposure, diffuse light, and specular reflection can produce similar pixels; clipped highlights and crushed shadows may require a better capture.
Transparent photo-to-albedo workflow
- Choose a photo or scanned texture. Use a PNG, JPEG, or WebP image. The browser creates a working analysis copy while keeping the original source available for an unlocked full-resolution export.
- Read the baked-lighting detector. Review the 0 to 100 score, dominant direction, shadow coverage, highlight coverage, exposure quality, and reflectivity warning before changing pixels.
- Inspect shadow and highlight heatmaps. Confirm that broad illumination was detected rather than real paint, grain, stone variation, decals, or damage that belongs in the surface color.
- Set correction strength and protect exceptions. Balance shadow recovery and highlight rolloff, then paint an exclusion mask over logos, colored stains, emissive details, or material variation that must remain unchanged.
- Check relighting and repeated seams. Compare the corrected preview under Studio, Overcast, and Courtyard neutral light, then inspect the 2 x 2 repeat and seam-delta measurement.
- Export the neutral base-color texture. Starter and higher plans can apply the analyzed illumination field to the original dimensions and download a PNG for the next PBR authoring step.
Correct the texture or recapture it?
| Source condition | Likely detector signal | Recommended action |
|---|---|---|
| Diffuse overcast or cross-polarized capture | Usually low baked-lighting score | Keep correction light and preserve the capture |
| Phone photo with broad directional shade | Directional field with broad shadow coverage | Recover shadows, inspect color, and protect real stains |
| Glossy or wet surface with clipped reflections | High highlight and reflectivity risk | Reduce only modestly or recapture with polarization |
| Underexposed scan with crushed blacks | Poor exposure warning | Recapture when hidden detail is needed |
What the detector reports
- Baked-lighting score: a diagnostic combination of low-frequency luminance range, directionality, and likely shadow or highlight coverage—not a physical albedo measurement.
- Shadow and highlight heatmaps: separate evidence views for checking whether the estimate follows captured light instead of real paint, grain, stains, or damage.
- Neutral base-color estimate: a fixed free preview that removes the broad field while retaining local detail and color as far as a single image allows.
- Relighting and seams: neutral-light previews plus a repeated 2 × 2 texture and seam-delta measurement before export.
- Warnings: explicit flags for clipped highlights, crushed shadows, strong reflectivity, and exposure conditions that limit recovery.
Texture delighting and photo-to-albedo FAQ
How do I remove shadows from a texture photo?
Estimate the broad illumination field separately from high-frequency surface detail, lift likely shadow regions, compress likely highlights, and verify the result under new lighting. Material Delighter exposes each estimate so real color variation can be protected with an exclusion mask.
Can one photo produce a true albedo map?
One photo can produce an estimated neutral base color, but not guaranteed physical ground truth. Surface reflectance, shape, illumination, exposure, and specular reflection are ambiguous in a single RGB image. Cross-polarized or multi-light capture remains the stronger source when measured albedo accuracy matters.
What does the baked-lighting score measure?
The score combines low-frequency luminance range, directional consistency, likely shadow coverage, and likely highlight coverage. A high score means the image is more likely to react incorrectly when used directly as base color; it is a diagnostic signal, not a claim about physical albedo accuracy.
Can Material Delighter remove reflections and glossy highlights?
Material Delighter can reduce broad or low-chroma highlights, but clipped reflections do not contain the hidden diffuse color needed for a perfect repair. The reflectivity warning identifies sources where correction should be treated as an estimate or replaced with a better capture.
Does texture delighting damage seamless edges?
Broad lighting correction can create a new edge mismatch. Material Delighter compensates the estimated illumination field across opposite borders, reports seam delta, and shows a 2 x 2 repeat before export. Existing geometric seams or nonmatching source pixels may still need a tiling tool.
Is the baked-lighting detector free?
Yes. The score, shadow and highlight heatmaps, neutral estimate, source warnings, detector report, relighting proof, and seam preview are free. Editable correction controls and full-resolution corrected PNG export require Starter or higher.
Are uploaded texture photos sent to a server?
No. Material Delighter decodes, analyzes, previews, masks, and exports the image in the browser. The selected source does not leave the device.
Method context and primary references
Last reviewed . Single-image delighting is an inverse-rendering estimate, so the interactive tool exposes its evidence and limitations instead of presenting the result as ground truth.