Porosity Analyzer for Metallography (AM & Coatings)

Measure the porosity of polished cross-sections in seconds: upload micrographs of metal additive manufacturing parts, castings, sintered parts or thermal spray coatings, and get the area porosity with its confidence interval, the pore size distribution, the largest pores, the share of round versus lack-of-fusion pores and a PDF report you can file with the build or batch record.

Try it in one click: leave the image list empty and press Start. The Actor generates five synthetic cross-sections with a known porosity (one of them with mounting resin on the edge), analyses them and shows how close the measurement is to the truth. Demo runs are free.

What you get

Output	Content
Dataset item	porosity % with its 95 % confidence interval, relative accuracy (%RA), overall porosity, material area, pore count and density, pore D50/D90, largest pore, share of round and irregular pores, threshold sensitivity, quality status and checks
REPORT.pdf	main results, pore sizes, pore types, the five largest pores with their position, detection check image, porosity per field, method, quality checks, traceability
pores.csv	one row per pore: type, cut by border, area, equivalent diameter, min/max Feret, roundness, circularity, aspect ratio, position in µm
control-sheet.jpg	every field with the detected pores outlined by type, to check the detection at a glance
summary.json	all statistics in machine-readable form

Who it is for

• Metal additive manufacturing (laser or electron beam powder bed fusion, DED): process window development, build qualification, lot release
• Thermal spray and coating shops measuring coating porosity on cross-sections
• Foundries, powder metallurgy and ceramics labs
• Quality engineers who count pores by hand in ImageJ or rate them against comparison charts
• AI agents that need porosity numbers from micrographs (callable through the Apify MCP server)

How it works

1. The pixel size comes from the microscope metadata (Zeiss or Thermo Fisher SEM TIFF), the scale bar of the image data band, or the value you give (typical for optical micrographs).
2. Uneven illumination is corrected (darker corners are common on optical microscopes), then the image is smoothed lightly.
3. The pore threshold is the lower of the Otsu threshold and the noise limit of the material, so that grain contrast and polishing scratches are not counted as pores when porosity is low. The run reports how much the porosity changes if the threshold moves slightly: a direct measure of how clear-cut the result is.
4. Large dark regions touching the field border (mounting resin, empty space) are excluded from the material instead of being counted as pores.
5. Each field gives a porosity (pore area / material area). The Actor reports the mean, the 95 % confidence interval (Student t × SD / √n) and the relative accuracy %RA, as in quantitative metallography, and how many fields would bring %RA down to 10 %. With fewer than 3 images, each image is split into 16 sub-fields and the report says so.
6. Every pore is measured (area, equivalent diameter, Feret diameters, roundness) and typed from its shape: round (gas or keyhole), irregular (likely lack of fusion) or intermediate. The roundness thresholds are adjustable, because they depend on the alloy and process.
7. Eight automatic checks rate the run green, amber or red: scale consistency, number of fields, relative accuracy, threshold sensitivity, pore/material contrast, pores too small to classify, saturation and share of the field outside the material.

The approach is close to the automatic image analysis method (Method B) of ASTM E2109 for thermal sprayed coatings and to the field statistics of quantitative metallography.

Accuracy

Measured on synthetic cross-sections whose porosity is known to the pixel (texture, polishing scratches, noise and blur included):

True porosity	Measured	Pores counted	Largest pore
0.59 %	0.59 %	29 of 29	137.0 µm (true 138.6)
2.17 %	2.19 %	104 of 104	152.7 µm (true 152.8)
6.16 %	6.21 %	244 of 244	153.7 µm (true 154.0)

With strong uneven illumination (corners 35 % darker), the illumination correction keeps the result at 2.17 % for a true 2.14 %, where a plain threshold would read 1.83 %. The share of porosity coming from irregular pores was measured within 1 point of the truth.

Input example

{
  "imageUrls": [
    "https://example.com/build-12-field1.tif",
    "https://example.com/build-12-field2.tif",
    "https://example.com/build-12-field3.tif",
    "https://example.com/build-12-field4.tif",
    "https://example.com/build-12-field5.tif"
  ],
  "pixelSizeUm": "0.5",
  "sampleName": "BUILD-12-XY"
}

Set "reportLanguage": "fr" for a report in French. Use scaleBarUm instead of pixelSizeUm when the images have a data band with a scale bar, or nothing at all for original Zeiss or Thermo Fisher SEM TIFF files.

Use 5 or more fields spread over the section, all at the same magnification, as-polished and unetched. Images must be direct download links (TIFF, PNG or JPG, up to 60 MB each).

Pricing

You pay per image analysed. There is no charge for images that cannot be downloaded, and demo runs (no image URLs) are free.

Good to know

• Pores must be darker than the material (the usual case in bright-field optical microscopy and in SEM). Tick poresAreBright for the opposite.
• Oxides, inclusions or dark phases as dark as pores are counted as pores. Preparation artefacts (pull-outs, smearing, edge rounding) are not detected automatically: check the control sheet.
• A 2D section measures area porosity, not volume porosity, and cuts pores at random heights. Use CT or the Archimedes method for bulk density.
• Pore typing from shape is indicative.
• This is automated image analysis, not an accredited (ISO/IEC 17025) laboratory measurement.
• Need a report reviewed and signed by a materials imaging scientist? A verified analysis service is available on request (contact through the Actor's issues tab).