Key Takeaways
- High-resolution automated core photography is the raw data source that most downstream AI core-analysis tools depend on.
- Corescan (now Epiroc), Datarock Core and Minalyze all offer automated imaging systems designed to feed AI extraction pipelines.
- This is a “Yes”: production-grade and already deployed across major mining companies.
- Get the photography pipeline right first. Every AI fracture, lithology or mineralogy tool downstream is only as good as the image quality and consistency it runs on.
TL;DR
Use AI on core photography by first standardizing how you capture it: fixed lighting, resolution and tray layout, ideally on automated hardware. Then send those images to computer-vision tools for fracture, lithology and texture logging. Consistent photos make every downstream model more reliable.
What Is Core Photography?
Core photography is the systematic imaging of drill core in its trays, usually wet and dry, before sampling and cutting. It has long served as a permanent record, because the physical core is split, sampled and sometimes lost over time. A good photo archive lets anyone revisit a hole years later.
In an AI-driven core shed, however, these photos are more than documentation. It is the sensor layer everything else builds on. Fracture detection, lithology classification, discing analysis and mineral mapping tools all consume core imagery as their primary input. Therefore, the quality of that photography directly caps how good the AI outputs can be.
How Do I Automate Core Photography With AI?
Most sites move through three steps: pick capture hardware, standardize the images, then run AI models on them.
Choosing core photography hardware
Corescan, acquired by Epiroc in 2022, integrates high-resolution photography with hyperspectral imaging and 3D laser profiling in one pass. It collects on the order of 800,000 spectral samples per metre, so photography is part of a combined capture rather than a standalone step. Minalyze’s Minalyzer CS combines photography with LiDAR-based 3D core-tray modelling and continuous XRF scanning in a single contactless pass. Both overlap with dedicated core scanning systems, which add mineralogy to the images.
Standardizing core photography for AI
Next, lock down the capture settings. Use the same lighting, camera distance, resolution and tray orientation every time. Photograph core wet and dry in the same way. Also record hole ID, depth marks and tray numbers clearly, so software can register each image to depth. As a result, models trained on one campaign still work on the next.
Running AI models on core photos
Finally, feed the images to computer-vision tools. Datarock Core takes standard RGB core photos as its direct input for AI fracture and lithology extraction. It has processed over 30 million metres of core since 2020, proof that this pipeline works at serious scale. The most validated output is fracture spacing logging, which feeds geotechnical design.
What to Expect From an AI Core Photography Workflow
Do not treat photography as an afterthought bolted onto your existing core shed process. Poor lighting, glare, shadows and inconsistent tray layout all become noise in AI outputs. For example, a reflection on wet core can look like a fracture to a model.
Retrofitting also has limits. Old photo archives can be reprocessed with AI, but results vary with their quality. Plan for a validation step where a geologist reviews AI outputs against the core on a sample of holes. Photography also supports quick geological logging, since loggers can work from images as well as trays.
FAQs
Why does core photography matter for AI?
Most AI core-logging tools use photographs as their main input. If images vary in lighting, resolution or layout, the models produce noisy or inconsistent results. Standardized imaging is therefore the foundation for reliable automated logging.
What is the best setup for core photography?
Use fixed, diffuse lighting, a consistent camera position and a set resolution for every tray. Photograph core both wet and dry and include clear depth and hole labels. Automated systems from Corescan, Datarock or Minalyze handle much of this for you.
Can AI read old core photos?
Yes, often. Tools like Datarock can process historical photos, which adds value to existing archives. However, accuracy depends on image quality, so validate outputs on a sample before relying on them.
Is core photography the same as core scanning?
No. Photography captures visible-light images of the core. Core scanning usually adds sensors such as hyperspectral, XRF or laser profiling, though many scanners include photography in the same pass.
Try It With Geocluster
Deciding which imaging hardware and photography workflow fits your deposit type and drill program is worth researching against comparable operations before committing capital. Geocluster can help pull together that comparative research quickly. Read how the Geocluster Research Harness works to see how.
