Rock identification from photos has fundamentally changed what's possible for hobbyist collectors, hikers, geology students, and anyone who picks up an interesting stone and wonders what it is. A decade ago, answering "what rock is this?" meant cross-referencing field guide plates, running streak tests, and often still being uncertain. Today, pointing a phone camera at a specimen and getting a precise identification with Mohs hardness, chemical composition, and formation history takes under ten seconds.
But AI rock identification from photos is only as good as the photo itself. The AI can't see what isn't in the image. Surface texture, color accuracy, and crystal habit — the features the model uses to make an identification — can all be obscured by bad lighting, dirty specimens, and poor focus. Getting these right is the difference between a confident identification and a vague "possibly granite" result.
Step 1: Prepare the Specimen
Before you take a photo, spend thirty seconds preparing the rock. This step matters more than any camera setting.
Clean the surface
Dust, mud, and watermarks change the apparent color and texture of a specimen. Dry mud in particular can make any rock look the same shade of tan. Rinse the specimen if possible, then let it dry — wet rock surfaces are also problematic because the water film creates reflections and alters apparent color. A dry, clean surface is the starting point for any accurate identification from a photo.
Expose a fresh surface when possible
Weathered rock surfaces look different from the mineral's actual properties. The chemical weathering that happens over decades — oxidation, hydration, biological activity — discolors surfaces, breaks down mineral structure, and creates coatings that obscure what's underneath. If the rock can be broken to expose a fresh face, that surface gives the AI significantly more accurate information. A fresh break on granite shows the individual mineral grains (quartz, feldspar, mica) clearly. A weathered exterior on the same rock might just look grey-brown.
If you don't want to break the specimen, look for a natural break or chip that already exposes the interior.
Remove coatings
Some minerals develop surface films that mask their identity. Hematite coatings can make many different rocks look rust-red. Calcite veins can make fragments of other rocks look white and crystalline. Where a coating exists and you can scrape a small area to see what's underneath, that scraped patch is often more informative than the coating surface.
Step 2: Set Up the Right Lighting
Lighting is the single biggest variable in rock identification photo quality. AI models trained on geological images expect certain visual characteristics — and flash photography destroys most of them.
Use diffuse natural light
The best light for rock identification photos is diffuse natural light: outdoors in open shade, or near a window on a cloudy day. This type of light reveals color accurately, shows surface texture through gentle shadows, and doesn't create the harsh specular highlights that obliterate detail.
Direct sunlight creates specular highlights — bright spots where the sun reflects off glassy mineral surfaces. These highlights wash out color information and can make very different minerals look identical. Open shade (under a tree, against a building, or on a cloudy day) eliminates this problem.
Avoid flash photography
Phone flash creates the same problem as direct sunlight but at close range: strong, single-source light that creates hard shadows and specular highlights. Flash photography is the most common cause of misidentification from photos. If you're indoors in poor light, position the specimen near a window rather than using flash.
Use supplemental light carefully
If you need additional light, use a desk lamp at an angle — positioned to the side rather than directly above the specimen. Side-angled light reveals surface texture by creating gentle shadows in depressions and along crystal faces. Overhead light flattens texture.
Step 3: Frame and Focus the Shot
Photo composition for rock identification has one primary goal: give the AI as much relevant information as possible in the frame.
Fill the frame with the specimen
Get close enough that the rock fills most of the photo. AI rock identification models look at surface texture, grain size, mineral inclusions, crystal structure, and color variation — all features that require proximity to capture. A rock photographed from 30cm away with a plain background showing in the frame gives the model less useful data than a close-up where the specimen fills the image.
Most phone cameras have a macro mode or will focus at close range. Get within 15-20cm of the specimen and let autofocus lock onto the surface.
Show the most diagnostic surface
Different rock types have different diagnostic features. For rocks with visible grains (granite, sandstone), show the grainiest, most textured surface. For crystals, show the crystal faces and habit. For foliated metamorphic rocks (schist, slate), show the layering. If you're unsure which surface is most informative, take two photos from different angles.
Keep a neutral background
A white piece of paper, a grey rock, or bare ground works well. Brightly colored backgrounds (grass, fabric) can affect the AI's color interpretation. Neutral is better.
Step 4: Use a Rock Identifier App
With a good photo taken, open your rock identifier app and submit it for analysis. Stone Snap — available free on Google Play — uses Google Gemini AI and returns one of the most complete geological data packages of any rock identifier app: chemical composition, Mohs hardness, crystal system, formation environment, common locations, and industrial uses.
The identification takes under five seconds. Results include a confidence indicator — high confidence identifications of common specimens are generally reliable. Lower confidence results, or identifications of rare minerals, are better treated as starting points for further investigation rather than definitive answers.
Step 5: Interpret the Results
A good rock identifier app returns more than just a name. Understanding what the data means makes the identification genuinely useful.
Mohs hardness
Mohs hardness is a relative scale from 1 (talc, scratches with a fingernail) to 10 (diamond, scratches everything). If the app identifies a specimen as quartz (Mohs 7), you can verify this by checking whether it scratches glass (Mohs 5.5) — quartz should scratch glass, and glass shouldn't scratch quartz. Mohs hardness is one of the most reliable physical verification tests you can run without equipment.
Chemical composition
The chemical formula tells you what elements are present. SiO₂ is quartz. CaCO₃ is calcite or aragonite. FeS₂ is pyrite. This information is useful for understanding why specimens look the way they do — pyrite's metallic lustre comes from the iron sulfide structure, and calcite's softness (Mohs 3) follows from its carbonate chemistry.
Crystal system
Crystal system describes the geometric symmetry of the mineral's internal structure. Cubic (isometric) minerals like halite and garnet form in cube-related shapes. Hexagonal minerals like quartz form six-sided prisms. Monoclinic minerals like gypsum form oblique shapes. The crystal system explains why a mineral's crystals look the way they do when they're well-formed.
Formation environment
Knowing where a rock forms contextualizes where you might find it. Basalt forms from rapidly cooling lava flows — finding it suggests a volcanic history for the area. Sandstone forms in ancient river deltas, beaches, or deserts — finding it suggests a sedimentary environment. Formation environment also tells you what other rock types are likely to be nearby.
Common Mistakes That Lead to Wrong Identifications
- Photographing the weathered exterior: The outside of many rocks looks nothing like the inside. Always try to show a fresh surface.
- Flash photography: Creates misleading highlights and shadows. Use natural light.
- Dirty specimens: Mud and dust change apparent color and texture. Clean first.
- Too far away: If the rock is smaller than 50% of the frame, surface texture isn't captured well. Get closer.
- Photographing concretions or weathering products as the main specimen: The rusty coating on a rock isn't the rock. The white calcite vein isn't the host rock. Make sure you're photographing what you want to identify.
Identify Rocks from Photos Free
Stone Snap uses Google Gemini AI — 5 free identifications on Android.
download Download Stone SnapFrequently Asked Questions
Can I identify rocks from photos?
Yes. AI rock identifier apps can identify rocks, minerals, and crystals from photos with high accuracy for common specimens. Stone Snap uses Google Gemini AI and returns chemical composition, Mohs hardness, crystal system, and formation data from a single photo. Clear, well-lit photos of clean specimens produce the most reliable results.
What app can identify rocks from a picture?
Stone Snap is the most capable rock identifier app for Android. It uses Google Gemini AI to identify rocks, minerals, and crystals from photos and returns detailed geological data. Other options include various "Rock Identifier" apps on Google Play, though data quality varies significantly between them.
How accurate is rock identification from photos?
Accuracy is high for common, well-photographed specimens and decreases for weathered rocks, rare minerals, or poor lighting conditions. For common rocks like granite, basalt, quartz, limestone, and sandstone, AI identification from photos is reliable. For unusual specimens, treat the app's identification as a starting point and verify with physical tests like hardness and streak.
Do I need to break the rock to identify it?
Not always, but photographing a fresh surface improves accuracy significantly. Many specimens can be identified from their exterior if it's clean and unweathered. When the exterior surface has heavy weathering or coatings, exposing a fresh face improves results.