Cameras & sensorsTroubleshooting

AcquisitionMode Continuous only: single frames with software trigger

When AcquisitionMode offers only Continuous, the camera has no SingleFrame mode. Confirm the enumeration, then capture single frames with a software trigger.

LumaSpec Vision9 min read Share on LinkedIn

On this page 9 sections
  1. 01What a one-entry AcquisitionMode means
  2. 02Confirm what the camera implements
  3. 03Why some cameras implement only Continuous
  4. 04The fix: one frame per software trigger
  5. 05Worked example: bandwidth and latency of each route
  6. 06When the Continuous-only reading is misleading
  7. 07Bring-up checklist for single-frame capture
  8. 08Choosing a camera when single frames matter
  9. 09Questions

Your code asks the camera for its AcquisitionMode values and the answer comes back with a single entry: Continuous. You expected Continuous, SingleFrame and MultipleFrame, because those are the names the GenICam Standard Features Naming Convention defines for acquisition control. Nothing is broken, and neither is your SDK. The camera's firmware implements one acquisition mode, and the enumeration reports exactly what exists. This troubleshooting guide covers why cameras ship that way, how to confirm the enumeration in about a minute, and the software-trigger pattern that delivers one frame per command on a Continuous-only camera, with the latency and bandwidth numbers that justify the switch.

What a one-entry AcquisitionMode means

An enumeration node in the camera's GenICam XML declares every value the firmware accepts. When AcquisitionMode lists only Continuous, SingleFrame does not exist on that device: it cannot be set, armed, or switched on later by any host-side call. Writing it anyway fails with an enumeration invalid value error, the standard response for a value outside the declared list — which is why reading the symbolics before writing anything pays off (GenICam enumeration invalid value: check symbolics before you set).

The Standard Features Naming Convention standardizes the names and behavior of common features, not which ones a camera must implement. It names three acquisition modes — Continuous, SingleFrame, MultipleFrame — and most cameras default to Continuous. In SingleFrame the camera captures exactly one image and stops the acquisition by itself; MultipleFrame captures a fixed count and stops. A camera that omits both is telling you single images are meant to arrive through the trigger system.

The decision: stop searching for a hidden mode. Continuous acquisition plus software trigger produces single frames on any camera that lists only Continuous, with no stop and restart between images.

Confirm what the camera implements

Three checks, in this order:

  1. Run the Aravis feature dump — arv-tool-0.8 features, with the version number matching your Aravis build — and read the AcquisitionMode section. It prints every implemented value.
  2. From your own code, call arv_camera_get_available_enumeration_entry_values_as_strings on the AcquisitionMode node. That returns the authoritative list, read from the XML the camera itself serves.
  3. Open the node in a generic tree browser — browse and set any GenICam feature with a generic feature tree — to see the value list and the node's current access state side by side.

A one-entry list answers a capability question, not an access question. If the list did contain SingleFrame but the write failed, the usual cause is acquisition still running or the node locked by mode — a different fault with a different fix (GenICam attribute not writable: set the feature mode first). A stale cached copy of the camera XML on the host can also distort enumerations; clearing it forces a fresh read (camera enumeration fails: clear the camera XML cache safely). The symbolics-first habit extends to every enumeration on the device, PixelFormat included (PixelFormat invalid value: list the camera's symbolics first).

Why some cameras implement only Continuous

Single-frame behavior costs the firmware extra acquisition states, and the trigger system already delivers it. The software trigger is also the better route when single-frame capture must be responsive or when consecutive images are needed, because SingleFrame stops acquisition after every image and must be started again for the next one.

GigE Vision and USB3 Vision cameras describe their controls through GenICam, and the standard trigger features are the common implementation across that class — from GigE models such as the MV-CU020-90GC to USB3 models like the LS-CMR-0166.

Area scan camerasHikrobot MV-CU020-90GC 2.5 MP Area Scan Camera, GigE, GMAX4002, ColorMV-CU020-90GC
Sensor format
1/1.7"
Lens mount
C-mount
$309.00View part
Area scan cameras12 MP USB3 area scan camera, IMX226, 1/1.7" rolling shutter, 30 fps, 4000 × 3000 resolution (LS-CMR-0166)LS-CMR-0166
Sensor format
1/1.7"
Interface
USB3
$560.00View part

A self-contained vision sensor is a different case: inspection runs on the device, and single captures may live in the sensor's own configuration tool rather than a GenICam tree. The enumeration check in the previous section tells you in about a minute which kind of device you are holding.

The fix: one frame per software trigger

  1. Stop acquisition if it is running.
  2. Set TriggerMode to On.
  3. Set TriggerSource to Software. If the camera exposes TriggerSelector, leave it on frame start so the trigger re-arms for every frame.
  4. Set ExposureTime and gain now, while the camera is idle.
  5. Call AcquisitionStart. The camera arms and waits — nothing streams, because every frame needs a trigger.
  6. Call TriggerSoftware once. Exactly one frame is exposed, digitized and delivered to the stream.
  7. For the next single image, call TriggerSoftware again. No stop, no restart. Stop acquisition when the session ends.

This is Continuous acquisition that never free-runs. Leave TriggerMode Off and the camera streams at full frame rate the moment it starts — the right mode for live image loops (continuous camera acquisition mode for live image loops), the wrong one for on-demand singles.

Worked example: bandwidth and latency of each route

Take an example camera — 2.4 MP mono, 8-bit pixels, 25 fps maximum, on a 1 Gbit/s GigE link — and put numbers on the ways to get single frames.

frame interval = 1000 ms / frame rate link load = frame size × frame rate per-frame latency = exposure + readout + frame size / usable link rate

Frame size: 2.4 × 10^6 pixels × 8 bit/pixel = 19.2 × 10^6 bit, about 19.2 Mbit (2.4 MB) per frame. Frame interval: 1000 ms ÷ 25 fps = 40 ms. Free-run link load: 19.2 Mbit × 25 = 480 Mbit/s. Triggered load at one frame every 2 s: 19.2 Mbit ÷ 2 s = 9.6 Mbit/s, a 50× reduction (480 ÷ 9.6 = 50).

Latency per requested frame: 8 ms exposure + up to 40 ms readout + 19.2 Mbit ÷ 800 Mbit/s usable payload = about 24 ms of transfer, so about 72 ms from TriggerSoftware to a complete frame in host memory. Use 800 Mbit/s as the usable payload figure for a 1 Gbit/s link carrying standard-size packets.

The stop-and-restart route — acquire, stop, restart for each single — renegotiates the stream channel every cycle, and on a GigE camera that re-arm costs on the order of 100-200 ms. Budget 150 ms per cycle: ten requested frames then carry 10 × 150 ms = 1.5 s of pure overhead, while the armed software trigger adds none. Per-frame latency is similar on both routes; the difference is the dead time between frames.

On a shared switch the load column decides how many cameras fit: two free-running cameras at 480 Mbit/s each nearly fill a 1 Gbit/s uplink, while the triggered pair averages under 20 Mbit/s combined. When free-running cameras do saturate a link, frames drop — that failure mode has its own diagnosis (troubleshooting multi-camera frame drops on GigE).

When the Continuous-only reading is misleading

  • Some wrappers list only currently writable values, so an implemented-but-locked entry can disappear from a filtered view. The raw feature dump from check 1 shows the full set.
  • SingleFrame is listed but the write fails: acquisition is running or the node is locked. Stop first, then re-check the access state (single frame acquisition still busy: why repeated starts need a stop).
  • Frames must align with a part or an encoder: set TriggerSource to the physical input line instead of Software, and the camera still delivers exactly one frame per pulse. If capture halts after the first pulse, the acquisition mode is the first place to look (hardware trigger stops after one frame: check the acquisition mode).

Bring-up checklist for single-frame capture

  • Record the AcquisitionMode symbolics at bring-up, together with the camera's firmware version, before writing acquisition code.
  • Branch the code on those symbolics: SingleFrame where present, Continuous plus software trigger otherwise.
  • Set TriggerMode and TriggerSource while acquisition is stopped, not during.
  • Issue one TriggerSoftware per requested frame and read one buffer per trigger.
  • Time the first ten frames: latency near exposure + frame interval + transfer says the path is healthy; whole multiples of the frame interval say the trigger is firing into an unarmed camera.

Choosing a camera when single frames matter

Across GigE Vision and USB3 Vision area scan cameras the software-trigger pattern is the standard route to single frames, so acquisition-mode support rarely constrains the choice — interface, sensor format and shutter do. The camera finder filters on those axes, and high-resolution GigE models such as the MV-CS200-10GC answer to the same GenICam controls as compact ones. Building the whole station? The vision system configurator starts at the camera step, and the cameras topic page collects camera-side mode and interface questions.

Area scan camerasHikrobot MV-CS200-10GC 20 MP Area Scan Camera, GigE, IMX183, Color, BasicMV-CS200-10GC
Sensor format
1"
Lens mount
C-mount
Interface
GigE
Resolution
16-25 MP
$739.00View partBuild a system with it
CamerasArea scan cameras

GigE, USB3, 10GigE and CoaXPress

1,287 models · from $149.00

Questions

Why does my camera list only Continuous for AcquisitionMode?

The enumeration lists the values the firmware implements, and the Standard Features Naming Convention standardizes names, not mandatory features. Implementing only Continuous is a deliberate design, common because single frames are expected to come from the trigger system. No SDK call, cache clear or shortcut adds SingleFrame to a camera that does not implement it — route single frames through the software trigger instead.

How do I list every AcquisitionMode value a camera supports in Aravis?

Run `arv-tool-0.8 features` (matching your installed Aravis version) and read the AcquisitionMode block, or call `arv_camera_get_available_enumeration_entry_values_as_strings` on the node from your own code. Both read the XML the camera serves, so they cannot disagree with each other. The list tells you what exists; whether the node is writable at this instant is a separate state shown on the node itself.

Does software-trigger capture change exposure or gain behavior?

No. ExposureTime and gain act exactly as they do in free-run; the trigger only decides when a frame starts. Set them while acquisition is stopped, then arm. Because the camera exposes one frame per TriggerSoftware command, frame-to-frame exposure consistency matches SingleFrame mode and beats grabbing the newest frame from a free-running stream.

Can a PLC output trigger one frame per part instead of software?

Yes. Set TriggerSource to the physical input line, wire the PLC pulse to it, and keep AcquisitionMode Continuous: the camera delivers exactly one frame per pulse, just as it does per TriggerSoftware command. If capture stops after the first pulse, check the acquisition mode and the trigger re-arm state before touching the wiring.

What latency should I budget from TriggerSoftware to a delivered frame?

Budget exposure plus readout plus transfer. For the worked example above — 2.4 MP, 8-bit, 25 fps on a 1 Gbit/s link — that is 8 ms + 40 ms + 24 ms, about 72 ms. Add host-side queuing on top. If you measure whole extra frame intervals, the trigger probably fired before the camera was armed: confirm AcquisitionStart completed before the first TriggerSoftware.