Machine Vision
Camera Bandwidth Calculator
Estimate data rate from resolution, frame rate, and pixel format, then pick a suitable camera interface.
Machine vision workflowStep 6 of 6
Related tools
How it works: Raw bandwidth is width × height × transport bpp × FPS. Mono10/12 use a 16-bit container. Interface recommendations include ~20% headroom for packet overhead and protocol overhead.
Learn about this tool
What it is, when to use it, formulas, and common questions
What is this?
Estimates uncompressed image data rate from resolution, frame rate, and pixel format.
When do engineers use it?
Use when selecting camera interfaces (GigE, USB3, CoaXPress) or sizing network switches and frame grabbers.
Formula
Data rate (MB/s) ≈ (width × height × transport_bpp × FPS) ÷ 8 ÷ 1,000,000
Example
5 MP at 30 FPS Mono8 ≈ 143 MB/s — above standard GigE; you may need dual GigE or USB3.
FAQ
- Is this compressed or uncompressed bandwidth?
- The estimate is uncompressed payload. JPEG/H.264 or camera-side binning can reduce the stream, but inspection often needs lossless or raw images — size the interface for the worst case you will actually run.
- Why isn’t GigE Vision enough for my camera?
- Standard GigE is roughly 125 MB/s theoretical and often ~100 MB/s usable after protocol overhead. 5 MP × 30 FPS Mono8 already exceeds that. Dual GigE, 5GigE, 10GigE, USB3, or CoaXPress are the usual next steps.
- Does color vs mono change the data rate?
- Yes. Bayer/color formats use more bits per pixel (or extra channels) than Mono8. Switching from Mono8 to RGB8 roughly triples payload at the same resolution and FPS. Mono10 and Mono12 are treated as 16-bit on the wire unless your camera uses packed 10/12-bit streaming.
- Should I include protocol overhead?
- Leave 15–30% headroom below the interface’s rated throughput for packet overhead, CPU, and other traffic. If the calculator is close to the limit, pick the next interface up.