CNC processing images: what the marks on a part actually tell you

• 1
  What CNC processing images actually record
• 2
  Reading toolpaths and setup count from an image
• 3
  Why the same geometry looks different in steel and aluminum
• 4
  What a photo cannot prove
• 5
  Comparing four common surface finishes

Every cut leaves a record. A photo of a machined surface captures three things at once: the finish left by the cutter, the path the cutter took, and the way the part was held while it was cut. That is why CNC processing images are useful before a quote. They show whether a feature was milled in one setup or three, whether a thin wall was supported, and whether the tool reached the bottom of a pocket without chattering.

Finish is the easiest signal to read. Bands of parallel marks mean a continuous toolpath at a steady feed. A patchy, dull surface usually means the tool rubbed instead of cut, often from a low feed rate or a worn edge. On aluminum we normally hold Ra 0.8–1.6 μm on as-machined faces and reach Ra 0.2–0.8 μm where a face seals or slides.

Fixturing is the harder signal, and it matters more. Look at the edges of a part in a photo. Sharp, even edges on a thin rib suggest the part was supported from both sides. Wavy or lifted edges suggest the wall flexed during the cut. That single detail often explains a tolerance problem before anyone opens a CMM report.

Tool marks also carry scale information. The distance between two adjacent marks equals the stepover of the cutter. Wide spacing on a large flat face is normal for roughing. Tight spacing on a small boss tells you a small tool was used, which usually means a longer cycle and a higher part price.

A three-axis part shows marks that all run in one direction per face. The table moves, the tool stays vertical, and every surface is reachable from above. If you see side-wall marks that curve or spiral on a part with deep pockets, the job likely ran on a four-axis or five-axis machine where the part rotated under the cutter.

Setup count is visible in the corners. Each new setup leaves a small step where two machined surfaces meet and do not line up perfectly. A step of a few hundredths of a millimeter on an outside corner is normal. A visible step on a sealing face is a warning. On our 5-axis centers, 16 simultaneous machines let us reach five faces in one setup, which removes most of those steps.

The rotary table size sets a practical limit. Our largest rotary table is Ø400 mm, so a part that needs continuous rotation has to fit inside that envelope. Larger parts can still be machined on our 4,000 mm travel machines, but they may need two or three setups, and the photo will show the seams.

Look at the direction of the marks on a bore. Helical marks mean the tool moved in a spiral. Straight vertical marks inside a bore usually mean a boring bar or a reamer, which is a different process with a different cost and a different roundness result.

Material changes the picture. Aluminum 6061 cuts clean and leaves bright, well-defined marks at high spindle speed. Stainless 316 work-hardens, so a light pass with a dull edge leaves a smeared, dull surface. If you compare two photos of the same bracket in 6061 and 316, the stainless one will look rougher even at the same Ra target.

Titanium TC4 (Ti-6Al-4V) behaves like stainless but worse. Heat stays at the cutting edge, so the marks look burnt or discolored when coolant or feed is wrong. Inconel pushes this further. A clean Inconel surface in a photo tells you the shop used low cutting speed, fresh tooling and a rigid setup. That is a real signal of process control.

Plastics and carbon fibre look almost mark-free. POM and PEEK cut with a sharp tool leave a matte surface, and carbon fibre can fray at the edges if the tool is dull. A photo of a carbon part with clean edges means the shop used a diamond or coated cutter and probably backed the laminate to stop delamination.

Plating and coating hide all of this. Once a part is anodized or electroless nickel plated, the tool marks sit under the coating. Ask for photos before finishing if you need to judge the machining itself.

A photo cannot prove a tolerance. A ±0.005 mm callout is 5 micrometres, roughly one tenth of the width of a human hair. No normal camera resolves that. The image can show that a face looks flat, but flatness and position are measured on a CMM or with gauges, not judged by eye.

Lighting also lies. A raking light across a surface makes small scratches look like deep grooves. A soft, even light hides them. Two photos of the same part under different lights can look like two different parts, which is why we shoot inspection photos with fixed lighting.

Scale is missing unless something in the frame gives it. A part with no ruler, no coin and no known feature could be 20 mm or 200 mm across. When you send reference photos to a shop, include one overall shot with a scale and one close-up of the critical feature.

Finally, a photo is a snapshot of one part. It does not show the spread across a 10,000 part run. Process capability comes from data: in-process monitoring, final inspection and reports you can request. Treat the picture as a pointer to the questions worth asking.