2D engraving vs true 3D relief
Almost every "photo on wood" you have seen is a 2D engraving. The laser sweeps back and forth and marks each pixel darker or lighter - either by burning the surface or by laying down dithered dots - but the wood stays perfectly flat. Hold it up at a low angle and there is no shadow, because there is no shape. It is a picture, not a carving.
A 3D relief is a different animal. Here the laser removes material, cutting deeper in some places and shallower in others, so the finished piece has real topography. Light rakes across it and casts genuine shadows in the recesses. Think of a coin: the portrait is not printed on the metal, it is raised out of it. That physical height is what your fingers feel and what makes a relief read as sculpture rather than a photograph.
The two techniques answer different questions. A dithered photo asks "how dark should this pixel be?" A relief asks "how deep should the laser cut here?" That single change - from darkness to depth - drives everything else in this guide: the source image, the material, the number of passes and the time it takes.
What a depth map actually is
A depth map is just a grayscale image, but you read it differently than a photo. Instead of "how bright is this spot," each pixel means how tall this spot should be. The convention is simple:
- White = highest (the laser barely touches it, or leaves it untouched).
- Black = lowest (the laser cuts the deepest here).
- Every gray in between is a proportional height between those two extremes.
That white-high / black-low convention is the most common, but it is not universal - it depends on how your software interprets the grayscale. In LightBurn's Grayscale mode, for example, you decide whether darker means more power (deeper) or less. The safe habit is to fix a convention, note it, and confirm it on a test tile before you commit hours to a carve. Get it backwards and you engrave a photographic intaglio - every hill becomes a valley.
Because a depth map only carries height, it throws away colour and most fine surface texture. That is by design. You are not trying to reproduce the photo's look; you are trying to describe its shape so the laser can rebuild it in wood or acrylic.
Getting a depth map from a photo
Here is the crux, and where most 3D-from-photo attempts go wrong. There are two ways to turn a flat photo into a height map, and they are worlds apart in quality.
The naive shortcut: luminance as height
The quick trick is to treat the photo's brightness directly as depth - bright pixels become high, dark pixels become low. It takes one grayscale conversion and no clever software. And for the right subject it genuinely works:
- Coins, medallions and seals, where the raised design really is the light part.
- Logos, lettering and stylised signage, where you want a clean two-level or few-level step, not smooth organic form.
- High-key line art on a plain background.
It fails badly on real scenes, because brightness is not distance. A white shirt is not physically closer to you than a dark eye socket; a shadow under the nose is not a hole. Feed a portrait's luminance in as height and you get a warped mask where clothing bulges forward and eyes sink - the uncanny "melted" relief beginners post and then delete.
The real thing: a depth estimate
A proper depth map comes from a model that actually understands the scene - a monocular depth estimator that infers which surfaces are near and which are far, regardless of their colour. Now the nose sits forward, the ears fall back, and the background recedes, exactly as your eye reads it. This is what makes a believable face, pet or figure relief possible from a single ordinary photo.
EngraveBot's Depth Map Studio does both. It can take the quick luminance route when that is genuinely the right call - a coin, a badge, a logo - and it can run a real AI depth estimate when you need honest near-and-far shape from a photo. It shows you a rotatable 3D preview so you can judge the relief before burning, and exports a clean grayscale height map (and a watertight STL if you would rather CNC it).
Best materials for relief
Depth carving punishes bad material choices far more than surface engraving does, because you are removing many layers and any inconsistency compounds with every pass. You want a material that erodes predictably and evenly.
| Material | Why it works for relief | Watch out for |
|---|---|---|
| Basswood / jelutong | Soft, fine, near-uniform grain; carves smooth gradients cleanly | Fuzzy edges - sand lightly and finish after |
| Cast acrylic | Melts to a frosted, glass-smooth depth; superb gradient | Use cast, not extruded; ventilate fumes |
| MDF | Homogeneous, no grain; cheap for tests, deep passes | Heavy smoke/soot; darkens; seal before finishing |
| Hardwood (oak, walnut) | Beautiful finished look | Hard/soft grain bands carve at different rates - uneven relief |
| Metal, stone, glass | Generally not viable | Most lasers cannot ablate real depth; surface-mark only |
The pattern is clear: soft and grainless wins. Basswood is the classic first relief material for a reason, and cast acrylic gives the most striking backlit results. Save hardwoods for when you already know your depth curve, and treat metal and glass as 2D-only - for those, engrave a photo on metal the flat way instead.
How the laser reaches depth
Your laser can only do one thing: deliver energy to a spot. Depth is what you get when you deliver more energy where you want a valley and less where you want a peak. There are two mechanisms, and most relief jobs use a blend.
Power modulation (grayscale, per pixel)
The machine reads the depth map pixel by pixel and varies laser power to match: dark pixels get high power and ablate deeply, white pixels get little or none. It is fast - one pass - and it is how a diode produces shallow relief. The limit is total achievable depth: a single power-modulated pass on wood typically reaches only a fraction of a millimetre to maybe 1-2 mm before you are just scorching, not cutting.
Layered / multi-pass carving
For real depth you stack passes. The laser makes several complete sweeps of the image; each pass removes another thin layer, and you drop the focus (Z height) a little between passes so the beam stays sharp at the new, lower surface. Twenty passes at 0.15 mm each carve roughly 3 mm of relief. This is how CO2 machines and gantry diodes reach the dramatic depth you see in trophy plaques and 3D signs.
In practice you combine them: grayscale power-modulation shapes the fine gradient within each layer, and multiple passes with a Z-step build the gross depth. More power, slower speed and more passes all push deeper - but each also adds heat, char and time, which is exactly why you calibrate on a tile first.
LightBurn Grayscale & material 3D mode
If you run LightBurn, the relevant setting lives in the image layer's Image Mode dropdown. Choose Grayscale (some builds label the workflow "Material 3D" or "3D mode"). Unlike dither modes, Grayscale tells the laser to modulate power per pixel from the image's brightness - which is precisely what a depth map needs.
- Set Image Mode to
Grayscale. Confirm the mapping - which end of the gray range fires more power - and match it to your white-high / black-low map. - Set Max Power to your deepest-cut power and Min Power above your machine's lasing threshold, so faint grays still mark.
- Set the line interval (DPI) for the raster. Relief usually wants a finer interval than a photo - around
0.06-0.10 mm- so adjacent lines blend into a smooth surface instead of visible ridges. - For deep work, enable multiple passes and set a Z-step per pass so the head lowers as it digs.
Every one of these options is walked through, screenshot by screenshot, in LightBurn image settings explained. The numbers here are starting points - never final settings - and the next section is where you turn them into real ones.
Passes, Z-step and line interval
Three numbers control how a relief comes out: how many passes, how far you drop focus each pass, and how tight your raster lines are. Here are sane starting points to calibrate from - not guaranteed-safe values. Your laser, lens and material batch will shift them, so always confirm on a tile.
| Laser | Approach | Passes | Z-step / pass | Realistic depth |
|---|---|---|---|---|
| 5-10 W diode | Grayscale, single slow pass | 1-3 | 0-0.15 mm | ~0.3-1 mm |
| 20 W diode | Grayscale + layered | 3-8 | 0.10-0.20 mm | ~1-2.5 mm |
| 40-60 W CO2 | Layered multi-pass | 8-30 | 0.10-0.30 mm | ~3-8 mm |
Line interval matters as much as passes. Too coarse and you see corduroy ridges across the relief; too fine and you cook the material and lose hours to no visible gain. Start near 0.08 mm on wood and tighten only if ridges show. The Laser Calculator will turn interval, size and speed into an honest time estimate - and relief times are long, so check before you start.
Why high-contrast depth maps carve better
A depth map's contrast is its depth range. A flat, low-contrast gray map - everything hovering around mid-gray - tells the laser to carve everything to nearly the same shallow depth, and the result is a mushy, featureless bump. Stretch the map so the nearest surfaces are near-white and the farthest are near-black, and you hand the laser the full depth budget to work with. Peaks stand tall, valleys cut deep, and the shape actually reads.
This is the same contrast logic as flat photo prep, but the payoff is physical rather than tonal. A few practical moves:
- Stretch the histogram so the map uses the whole 0-255 range before exporting.
- Protect the key feature. On a face, make sure the nose-to-ear distance uses a wide slice of gray, even if it means clipping an unimportant background.
- Do not over-sharpen. Hard edges in a depth map become vertical cliffs the laser cannot cleanly cut - keep transitions smooth for organic subjects.
Depth Map Studio's preview lets you exaggerate or flatten the height range and see the effect in 3D before committing. If a relief comes out weak, boosting depth-map contrast fixes it far more often than adding power does.
The test-tile approach
You cannot eyeball relief settings - the interaction of power, passes, Z-step and material is too tangled. The answer is the depth-carving cousin of the flat test grid: a step tile. Burn a small strip where you deliberately vary the deep-cut settings, then measure and pick.
- Carve a row of squares at increasing passes (say 2, 4, 8, 12) with a fixed Z-step, on an offcut of your real material.
- Measure the actual depth of each with calipers or a depth gauge - this is your machine's true depth-per-pass curve.
- Run one small gradient swatch of your depth map to confirm the gray-to-depth mapping looks right and nothing is inverted.
- Only then commit the full piece.
The Test Grid Generator makes the flat power/speed grid you should already own; for relief, add a passes-and-Z-step tile of your own on the same offcut. Ten minutes of tiles saves a three-hour carve that comes out too shallow to see or so deep it burns through.
Make a depth map from your photo - free, in your browser
Realistic expectations
Honesty keeps you from disappointment. A laser relief is not a CNC-milled sculpture, and knowing what the process does well lets you pick projects that shine:
- Coin- and medallion-style reliefs. The sweet spot - shallow, high-contrast, and forgiving. Logos and seals look fantastic.
- Lithophane-style pieces. Thin cast acrylic or porcelain-look material carved so light passes through the thin (deep-cut) areas - a photo that glows when backlit. Technically a depth job, and one of the most rewarding.
- Signage and lettering. Raised or recessed text and crests, clean and repeatable, especially on MDF and acrylic.
- Portrait and pet reliefs. Achievable with a real depth estimate and soft wood, but the most demanding - expect several tiles before you dial it in.
What a laser relief will not do is give you deep, crisp-walled sculpture in hardwood or any real depth in metal - that is milling territory. Match the ambition to the machine and the material, start shallow, and build up. For flat, framed portraits, a good dithered engraving on wood still beats a mediocre relief every time.
Frequently asked questions
What is the difference between a dithered photo engraving and a 3D relief?
A dithered engraving is 2D - the laser only marks the surface darker or lighter and the material stays flat. A 3D relief physically carves the surface to different depths, so it has real shape you can feel and that casts shadow. Relief needs a depth map and many passes; a dithered photo needs one pass.
Can I just use a photo's brightness as a depth map?
For coins, logos and signage where the bright areas genuinely stick out, yes - luminance-as-height is fine. For faces, pets and normal scenes it fails, because a bright shirt is not closer than a dark eye. Those need a real depth estimate that understands near and far.
What material is best for 3D relief laser engraving?
Soft, even-grained materials carve cleanest: basswood, jelutong, cast (not extruded) acrylic and MDF, because they erode predictably pass after pass. Strong-grained hardwoods carve unevenly, and metal, glass and stone cannot hold real depth on most lasers.
How many passes does a 3D relief take?
It depends on your laser and target depth. A diode doing grayscale power-modulation reaches roughly 0.3-2 mm in one or a few slow passes. For deeper relief you run layered passes - often 5 to 30 - each dropping focus by a Z-step of about 0.1-0.3 mm, so a deep carve can take hours.
Keep going
Ready to try it? Build your height map in the Depth Map Studio, then compare it with a flat approach in how to convert a photo to a laser file. If you decide a 2D engraving suits your subject better, dial in the best dithering settings and pick your surface with the engraving on wood guide.