Stringing, warping, layer shifts, blobs, gaps — every maker hits a wall where the print looks wrong but the cause isn’t obvious. The fastest way to fix it is to diagnose a failed 3D print from a photo, match the visible symptom to a root cause, and apply the exact slicer change that fixes it. This guide walks you through doing that systematically — by eye and with vision AI — so you stop guessing and start printing.
We’ll cover how to photograph a defect properly, the visual signatures of the most common failures, and how to turn a diagnosis into concrete settings in PrusaSlicer or OrcaSlicer. No fluff, no hype — just the workflow we’d use ourselves.
Why a photo is enough to diagnose most failures
The vast majority of 3D-printing defects leave a distinctive visual fingerprint. A trained eye — human or AI — can read surface texture, layer geometry and the location of the problem on the part to narrow down the cause quickly. You rarely need to be standing next to the machine.
That said, a photo is only as good as the information in it. Before you analyse anything, get a usable image.
How to photograph a failed print for diagnosis
- Light it from the side. Raking light at a shallow angle throws shadows across the surface and makes layer-level defects visible. Flat, head-on lighting hides everything.
- Get close and in focus. Fill the frame with the defect. A blurry wide shot tells you nothing.
- Show context. Include one shot of the whole part and one close-up of the worst area, so the defect’s position (corners, overhangs, bridges, one specific layer) is clear.
- Use a plain background. Avoid clutter that confuses both you and any AI tool.
- Note the basics. Material, nozzle temperature, print speed and layer height. The same symptom can have different causes on PLA versus PETG.
Common defects and what they look like
Here’s how to read the most frequent failures from a photo. Match what you see, then jump to the fix.
Stringing and oozing
Looks like: fine hairs or wisps strung between separate features, or fuzzy whiskers on travel moves. Most common on PETG.
Usual causes: retraction too low, temperature too high, or excessive travel without retraction. Start by dropping the nozzle temperature by 5–10 °C and increasing retraction distance, then enable or tune coasting and wiping.
Warping and lifting
Looks like: corners curling up off the bed, or the whole base bowed. Worst with ABS and large flat bottoms.
Usual causes: uneven cooling and poor bed adhesion. Raise bed temperature, add a brim, reduce part-cooling fan for the first layers, and control ambient draughts (an enclosure for ABS/ASA).
Layer shifting
Looks like: the print abruptly offset sideways at a specific height, as if sliced and slid. Everything above that line is displaced.
Usual causes: mechanical — a skipped belt step, loose pulley grub screw, print speed/acceleration too high, or a nozzle catching on a curled feature. This is one to check hardware on, not just slicer settings.
Under-extrusion and gaps
Looks like: thin, patchy walls, visible gaps between perimeters and infill, or missing top layers.
Usual causes: partial nozzle clog, too-low flow rate, temperature too low for the speed, or a slipping extruder. Calibrate flow/extrusion multiplier, raise temperature slightly, and check the PTFE/extruder path.
Elephant’s foot
Looks like: the bottom couple of layers bulge outwards wider than the rest of the part.
Usual causes: first-layer over-squish plus high bed temperature. Reduce first-layer flow, raise Z slightly, or enable elephant-foot compensation in your slicer.
Poor bridging and sagging overhangs
Looks like: drooping strands across gaps, or rough, curling undersides on steep overhangs.
Usual causes: insufficient cooling and too-fast bridge speed. Increase part-cooling fan, slow bridge/overhang speed, and tune bridge flow.
From diagnosis to exact settings — without the guesswork
Reading the symptom is half the job. The other half is translating it into a specific, numeric change in your slicer — and that’s where most generic advice falls down. “Increase retraction” is useless if you don’t know whether 2 mm or 6 mm is right for your hardware and material.
This is exactly what we built AI 3D print failure diagnosis for. Instead of a generic chatbot that hallucinates plausible-sounding numbers, Ask The Nozzle’s Diagnose tool uses vision AI and a curated knowledge base of real-world cases to identify the defect from your photo and return concrete, slicer-specific recommendations — including downloadable .ini patches for PrusaSlicer and OrcaSlicer you can import directly. No retyping a dozen settings by hand.
If you want the deeper rationale on why prints fail in the first place, our companion piece on why 3D prints fail and how AI photo diagnosis fixes it breaks down the cause-and-effect chain in more detail.
A quick manual diagnostic workflow
- Locate the defect. Is it everywhere, at one specific layer, only on overhangs, only at corners, or only on one side? Location narrows the cause fast.
- Check timing. Did it start at the first layer (adhesion/levelling), partway up (mechanical or thermal), or only on certain geometry (cooling/speed)?
- Change one variable at a time. Apply a single fix, reprint a small test, and observe. Changing five settings at once tells you nothing.
- Keep a log. Note material, settings and outcome. Patterns emerge quickly and you’ll diagnose faster next time.
When to use AI vs your own eye
Trust your own eye for the obvious stuff — a curled corner or classic PETG stringing. Reach for a vision-AI tool when the symptom is ambiguous, when the same defect could have three different causes, or when you want the exact numbers for your material and printer rather than a forum’s generic suggestion. The advantage of a purpose-built tool over a generic chatbot is that it’s grounded in real cases and returns importable settings, not confident guesswork.
FAQ
Can you really diagnose a failed 3D print from a photo?
Yes, for most defects. Stringing, warping, layer shifts, under-extrusion, elephant’s foot and overhang sag all have clear visual signatures. A good, well-lit close-up plus your material and key settings is usually enough to pinpoint the cause and the fix.
What’s the best photo to take for diagnosis?
Two shots: one of the whole part for context, and one sharp close-up of the worst area, lit from the side so layer detail and texture show up. Add a quick note of the material, nozzle and bed temperatures, layer height and print speed.
How is AI photo diagnosis better than asking a chatbot?
A generic chatbot can describe failures but tends to invent numbers and can’t see your print. Ask The Nozzle’s Diagnose tool actually analyses your photo, draws on a curated knowledge base of real cases, and outputs concrete, slicer-specific settings — including downloadable .ini patches for PrusaSlicer and OrcaSlicer.
Does it work with both PLA and PETG?
Yes. The same visual symptom can have different causes across materials, which is exactly why the diagnosis is material-aware. Recommendations are tailored to the filament and printer you tell it you’re using.
Bottom line: learn the visual signatures, photograph the defect properly, and change one variable at a time. When you want the exact numbers without the guesswork, upload the photo and let the Diagnose tool turn the symptom into a ready-to-import fix.