3D print troubleshooting guide: AI photo diagnosis and 30 problems
3D print troubleshooting guide: upload a photo of the defect and the AI names it, or search 30 problems across 12 technologies and read the fix steps.
Opens with a free account.
- 9 defect typesnamed by AI from a photo
- 30 problemsin 12 technologies
- 1 creditper photo analysis
- Free account30 AI credits at sign-up
What a 3D print troubleshooting guide is
This 3D print troubleshooting guide tells you what went wrong with the part and what to do. You can upload a photo of the defect and the AI names it, or browse the manual lookup of 30 problems across 12 technologies, from FDM and resin to metal and concrete.
The AI part names one of nine defect types, from layer shift and stringing to a clogged nozzle and a slicer settings error. It picks the closest of ten possible answers; “unrecognised defect” is shown when it picks that one or when its answer cannot be read.
The causes and the steps are not written by the model: they come from the EDi Brain knowledge base.
What it does: AI from a photo and a manual lookup
Names the defect from a photo
You upload a JPEG, PNG or WebP up to 8 MB, or take one with your phone, and optionally describe the material, the printer and the layer where it appears. You get the defect name and a confidence in percent.
Gives causes and steps
For the named defect you get two or three causes and three or four steps from the knowledge base. For stringing, for example: dry the filament, raise the retraction, lower the temperature.
Keeps your latest analyses
The last three analyses stay in your browser, and the “Запази” (Save) button puts one into your Library.
Has a manual lookup
The manual lookup holds 30 problems across 12 technologies: FDM, resin, SLS, metal powder, electron beam, concrete and more. You search by problem or by fix and mark the ones you solved.
When you open it: a defect, an unknown problem, another technology
Open it when a part comes out wrong or you work with an unfamiliar technology.
The part has a visible defect
You take a photo, upload it and get a diagnosis and steps, for example tighten the belts for a layer shift.
The first layer will not stick
The photo or the description leads you to the first-layer entry: clean the bed, set the first layer height, raise the bed temperature.
You work with resin or powder
In the manual lookup you search for pixelation in DLP and LCD or shrinkage in sintering.
You want the same steps for one defect
The steps for one defect type are always the same, but if you ask again with the same photo, the model may name another type.
How the AI names the defect and where the steps come from
The photo answer has two parts: an AI model names the defect, and the causes and steps are ready text from the knowledge base.
The photo is shrunk
In the browser the photo is shrunk to at most 1280 px on the long side and kept as a JPEG. The server takes it up to about 4 MB.
The model names the defect
The server sends the photo, your description and the technology (FDM by default) to an AI image model through OpenRouter. It returns one of ten types, a confidence and two sentences about what it sees.
The steps come from the knowledge base
For the named type the server takes the ready causes and steps from the EDi Brain knowledge base and returns them with the answer.
The credit is taken after the answer
If the service does not answer, the analysis is not charged. Otherwise it costs 1 credit.
| Applies to | Formula | Meaning |
|---|---|---|
| How many analyses the free credits give | 30 credits ÷ 1 credit = 30 analyses | the free credits are one-time and shared by all the AI tools |
| Confidence | the model’s score from 0 to 1 × 100 = percent | the percent is the model’s own estimate, not a measured accuracy |
| Photo size | up to 1280 px on the long side | larger photos are shrunk |
Recognition from a photo and a fix from the knowledge base
The steps are not the model’s idea
The model only names the defect. The causes and the steps are ready and the same for each type; they carry an “EDi Brain” badge, but the source itself is not shown.
There is a manual way too
Searching the manual lookup costs no credits and needs no photo: you type the problem or the fix.
Keeps the last three analyses
The diagnoses stay in your browser; Save sends them to the Library.
The AI photo diagnosis is for FDM printers: nine defect types and steps such as belts, nozzle and filament. For resin, powder or metal use the manual lookup. The model recognises, it does not measure, and the percent is its own estimate, so when in doubt make a test print after every change.
The tool does not see your printer and changes nothing in the slicer: it gives general steps for the defect type. The screen and the answers are in Bulgarian only. Classroom seats have no access to the AI diagnosis.
The manual lookup also has a block “Визуална диагностика (FDM)” (Visual diagnostics). It is not the AI photo diagnosis: by the code the server rejects its photo, so do not rely on it.
Questions about 3D print problems
How to fix bad 3D prints?
Upload a photo of the defect in “AI диагностика по снимка” (AI photo diagnostics) for causes and steps in order, or look it up in “Ръчна справка” (Manual lookup) for steps. Start with the first step and make a test print after each change.
What is wrong with my 3D printer?
The causes in the base include damp filament, a dirty or unlevelled bed, loose belts, a clogged nozzle and an unsuitable profile. A photo or a symptom description shows which one.
How many defects does the AI recognise?
Nine types, from layer shift and stringing to a clogged nozzle and vibrations. If it picks “unrecognised defect” or its answer cannot be read, that is what is shown.
How much does a photo analysis cost?
1 AI credit. A free account gives 30 credits once, about 30 analyses, then you buy packs. Searching the manual lookup costs no credits.
Is the tool in English?
No. The screen and the answers are in Bulgarian only: the app has no language switch.
The guide opens with a free account. Pro opens the rest
Pro costs 15 EUR a month and opens 45 Pro tools. The manual lookup costs no credits. A photo analysis is 1 AI credit: a free account gives 30 once, then you buy packs.
Related tools
Failed print photo diagnosisWhy did my 3D print fail? Upload a photo of the failed print and get a diagnosis. First analysis without an account; with a free account, 3 times a day.
3D print problem diagnosis3D print problem diagnosis: upload a photo of a failed print and an AI assistant suggests the defect, the cause and settings to change. Ultra, 39 EUR a month.
First layer diagnosisFirst layer problems: pick what you see on the bed, your filament and printer, and get causes, fixes and a slicer profile. Needs a free account.
Slicer settings visualizerSlicer settings explained: 90 settings with a drawing, 53 with a slider, the reason and the trap. Search by symptom. Free account. The screen is in Bulgarian.
Find out what is wrong with your print
Opens with a free account.