You ran the vendor's install script, reseated the ribbon cable twice, and libcamera-hello --list-cameras still prints No cameras available. On Raspberry Pi OS Bookworm that empty list points at one layer first: the device tree overlay that describes the sensor never loaded, so the kernel has no driver node to register and libcamera has nothing to list. The usual culprit is a one-character typo in the dtoverlay line of the boot partition's config.txt — dtoverlay=imx4770 where the sensor name is imx477. This guide walks the failure in the order it happens: read the config line back, fix the character, verify detection, and hand off to the next guide when the failure sits deeper in the stack. It is for engineers bringing an IMX477 CSI module into an embedded camera build.
What "No cameras available" actually tells you
Camera enumeration is a chain, and libcamera sits at its end. At power-up the firmware reads config.txt and applies every dtoverlay entry it can resolve against the .dtbo files on the boot partition. The IMX477 overlay defines the sensor node on the CSI and I2C buses. The kernel probes that node with its in-tree imx477 driver. Only then does libcamera list a camera.
An empty list means the chain broke at or before the driver probe. libcamera is rarely the problem; it is just the first tool that reports the empty state. That is why the search order runs upstream: overlay line first, cable second, kernel pipeline third.
On a healthy line the same command prints the sensor name and its modes. For the IMX477 those modes run from 1332×990 up to the full 4056×3040, and the smallest mode reaches about 120 fps. Seeing mode rows tells you detection already works and the problem is downstream.
Likely causes, most probable first
Work down this table and stop when the camera appears.
| Probable cause | Test | Fix |
|---|---|---|
Typo in the dtoverlay line: imx4770 instead of imx477
|
Read the line back from the boot config | Correct the name, reboot |
Edit never saved, or saved to a config.txt the firmware does not read |
Confirm the path is on the boot partition; read the file back as root | Edit with root, save, reboot |
| Ribbon cable unseated or reversed | Reseat both ends; check latch and contact side | Reseat, retry |
| Overlay applied, sensor probed, pipeline still fails | Kernel log names the sensor, libcamera stays empty | Empty-libcamera guide below |
| Detection works, capture times out |
--list-cameras succeeds, streaming fails |
Streaming-failure guide below |
Fix the overlay typo: read, correct, verify
Back up the file first, then work through the line.
- Copy the file as a backup and open it with root rights:
sudo cp /boot/config.txt /boot/config.txt.bak, thensudo nano /boot/config.txt. Current Bookworm images mount the boot partition at/boot/firmware, so on your image the file may be/boot/firmware/config.txt— edit whichever path exists. - Find the camera line and compare it character by character. The failing line reads
dtoverlay=imx4770. The correct line readsdtoverlay=imx477. The overlay name is the sensor name, not the camera module's product code. - Save, then read the line back before rebooting:
grep dtoverlay /boot/config.txtmust printdtoverlay=imx477— six characters after the equals sign, not seven. One extra digit leaves no matching overlay file, and the firmware applies nothing. - Reboot with
sudo reboot. Overlays apply at power-up only; restarting services does nothing. - Verify twice:
dtoverlay -llists the overlays the firmware actually applied, andlibcamera-hello --list-camerasnow prints the IMX477 with its modes.
When the line is correct and the list is still empty
Move down the chain. Reseat the ribbon at both ends first: connector latches lift, the contact side differs between modules, and a cable fault costs thirty seconds to eliminate.
Next, read the kernel log and search it for the sensor name. The result sorts into three cases, each with its own guide:
- Nothing in the log at all means the overlay still has not applied. Re-check the file path and the saved line, then work through the device tree overlay CSI mapping guide.
- The log shows the
imx477driver binding to the sensor while libcamera stays empty. Detection succeeded and the break sits between sensor frames and the libcamera pipeline; the guide on a kernel-detected sensor with an empty libcamera list picks up at exactly this point. - Detection succeeds and the first capture times out. That is power, clock or lane trouble, not detection; the guide on sensors detected but streaming failing covers the rails and clocks.
The same sensor-first order holds on other carriers: the IMX477 bring-up checks for Jetson Orin Nano run the sequence against a different CSI host. Deeper failures in this chain each have a walk-through on the cameras topic page.
Why the extra character gets in
Three habits produce this typo:
- The box prints a product code and the overlay wants a sensor name, so typing the code adds digits that do not belong.
- Official modules are auto-detected on Bookworm, so a third-party board's
dtoverlayline comes from a vendor script or a hand edit — and hand edits next to script-appended lines are where stray characters land. - More than one
config.txtexists. Booting from SD today and USB tomorrow, or editing the partition from another computer, corrects a copy the firmware never reads.
Prevention is a habit, not a tool: after every installer or edit, grep the line back, keep one commented camera block naming the sensor, and confirm with dtoverlay -l after the reboot.
When CSI is the wrong interface for the build
For a bench prototype the CSI module is fine and the fix above is the whole story. For a production inspection cell, the device tree stack is one more thing to re-validate at every OS image update. Board-level cameras on USB3 or GigE move capture off that stack: they enumerate through their vendor SDK, stream without overlay work, and swap across carrier boards without touching the boot config. The guide on the USB3 Vision acquisition path without V4L2 support compares the library options, and the board camera evaluation checklist covers the rest of the build.
Two parts from the board-level cameras range show the difference. The LS-BCM-0003 pairs a 3.1 MP global-shutter sensor with USB3 and a C-mount, running 120 fps for motion-heavy benches.
- Sensor format
- 1/1.8"
- Lens mount
- C-mount
- Interface
- USB3
- Resolution
- 5 MP
For longer runs on a factory network, the MV-CB016-10GM-M-W is a 1.6 MP mono GigE board camera with an M12 mount and a global shutter.
- Sensor format
- 1/2.9"
- Lens mount
- M12
- Interface
- GigE
- Resolution
- Up to 3 MP
To filter the range by sensor, interface and mount, start at the camera finder, then let the configurator do the matching:
The configurator matches a camera, lens, light, light controller and cables to each other, step by step, and lists only parts that fit what you already picked.
The five-minute checklist
- Run
grep dtoverlayon the boot config after every install script; the name must be the sensor name. - Count the characters:
imx477, neverimx4770. - Edit as root and save; the boot partition rejects a plain-user write.
- Reboot — overlays apply at power-up only.
- Confirm with
dtoverlay -l, thenlibcamera-hello --list-cameras. - Correct line and still empty? Kernel log next, then the three guides above.
- Standardizing cameras for production? Choose USB3 or GigE board-level cameras over CSI.
Questions
Why does a wrong dtoverlay name fail without any error message?
The firmware checks each dtoverlay entry against the .dtbo files in the boot partition's overlays folder and skips names with no match. A skipped overlay boots normally: no console message and no kernel log entry, because the sensor node never existed for the kernel to probe. The empty camera list from libcamera is the only symptom anywhere in the system.
Is the IMX477 supported on Raspberry Pi OS Bookworm 64-bit?
Yes. Bookworm runs the libcamera stack, and the imx477 driver ships inside the kernel package, so no legacy camera stack or extra daemon is needed. Support means the overlay file and the driver both exist on the image — your job is to put the correct dtoverlay=imx477 line in the boot partition's config.txt and reboot.
Where is config.txt on a Bookworm image?
On the FAT boot partition. Recent Bookworm images mount that partition at /boot/firmware, so the file is /boot/firmware/config.txt; older layouts and many guides still use /boot/config.txt. Edit whichever path exists on your image, with root rights, then reboot: the firmware reads the file before the kernel starts, so nothing short of a power-up applies a dtoverlay change.
The overlay is correct and loaded — why does libcamera still list no camera?
The failure has moved past detection. Search the kernel log for the imx477 driver binding to the sensor; if it probed, the sensor is alive and the break sits between sensor frames and the libcamera pipeline. The guide on a kernel-detected sensor with an empty libcamera list covers the pipeline and image-processing checks that come next.
Can the same typo hide a different sensor?
Yes — the failure mode is identical for any CSI module whose overlay name is mistyped: a wrong digit or a wrong sensor name on an imx219 line fails exactly like imx4770 does. The diagnostic never changes: read the dtoverlay line back, compare it to the sensor name, and check the overlays folder for a matching .dtbo before touching hardware.

