Hello !
I’m trying to update our software to support IDS camera (GV-527xCP-C). We already support Flir-Teledyne and Lucid Vision cameras.
We are using CVB 14.00.008. (We also tried CVB 14.01.006) on Nvidia Jetson.
IDS Camera has the latest firmware installed.
We both use C++ API and python bindings depending on purpose (production or calibration).
We installed ids-peak package to get the correct cti files.
After installing version 2.17, the GENICAM_GENTL64_PATH is updated to /usr/lib/ids/cti
Lucid camera are not detected anymore (make sense). IDS are still detected but can’t be opened anymore:Traceback (most recent call last):
File "/home/nvidia/workspace/platform/edge/test_ids_chunk_issue.py", line 18, in <module>
with cvb.DeviceFactory.open(
RuntimeError: Unspecified GenTL error
Reverting the GENICAM_GENTL64_PATH to GENICAM_GENTL64_PATH=/opt/cvb/drivers/genicam, we can open the camera. When Chunk Mode is deactivated, image acquisition works just fine :
Set Node Std::ChunkModeActive to false
-
Composite purpose is Image
-
First item type is cvb.Image
-
Access the image from the composite object
Number of parts: 1, purpose: 0
First part type: <class 'cvb.Image'>
Trying to convert to MultiPartImage
Size of image: 1536.0, 2048.0
When Chunk Mode is activated, it doesn’t work:
Set Node Std::ChunkModeActive and Std::ChunkEnable[Std::ChunkSelector(Timestamp)] to true
-
Composite purpose is custom
-
First item type is NoneType
-
ChunkTimestamp node is not readable
Number of parts: 1, purpose: -1
First part type: <class 'NoneType'>
Trying to convert to MultiPartImage
Size of image: 0.0, 0.0
Traceback (most recent call last):
File "/usr/app/test_ids_chunk_issue.py", line 18, in <module>
with cvb.DeviceFactory.open(
File "/usr/app/test_ids_chunk_issue.py", line 47, in <module>
print(f"Chunk timestamp: {chunk_timestamp_node.value}")
RuntimeError: Node is not readable.
It also works fine with the default CVB GENICAM_GENTL64_PATH for Lucid Cameras:
Number of parts: 1, purpose: 0
First part type: <class 'cvb.Image'>
Trying to convert to MultiPartImage
Size of image: 1536.0, 2048.0
Chunk timestamp: 61804137122696
I think it’s related to the cti file (tbc?)
Based on what I read on this forum and in the documentation, I briefly tried to use the solution by using PFNCBuffer but it didn’t work.
I didn’t investigate more this solution because when I try to access the camera through GenICamBrowser (using theGENICAM_GENTL64_PATH=/usr/lib/ids/cti ), everything works ! Even setting the ChunkMode works, the ChunkTimestamp Node is readable and visible.
So I wanted to know why it’s working fine with GenICamBrowser but not with the python script (or cpp script) ?
Here is the small script I use for test :
import cvb
from typing import Optional, Tuple
if __name__ == "__main__":
exp_serial_num = "4108794564"
exp_device = None
devices = cvb.DeviceFactory.discover_from_root(flags=cvb.DiscoverFlags.IgnoreVins)
for device in devices:
if device.read_property(cvb.DiscoveryProperties.DeviceSerialNumber) == exp_serial_num:
exp_device = device
break
if exp_device is None:
raise ValueError("Device not found")
exp_device.set_parameter("AttachChunk","1")
print(f"token: {exp_device.access_token}")
with cvb.DeviceFactory.open(
provider=exp_device.access_token, acquisition_stack=cvb.AcquisitionStack.GenTL
) as exp_device_ptr:
node_map = exp_device_ptr.node_maps[cvb.NodeMapID.Device]
# Chunk mode active
chunk_mode_active = node_map["Std::ChunkModeActive"]
chunk_mode_active.from_string("1")
#chunk_selector = node_map["Std::ChunkSelector"]
#chunk_selector.from_string("Timestamp")
#chunk_enable = node_map["Std::ChunkEnable"]
#chunk_enable.from_string("1")
# grab
stream: Optional[cvb.CompositeStream] = exp_device_ptr.stream(stream_type=cvb.CompositeStream)
stream.start()
result: Tuple[Optional[cvb.Composite], int, Optional[cvb.NodeMapEnumerator]] = stream.wait_for(
time_span=1000
)
composite, status, _ = result
if status == cvb.WaitStatus.Ok:
print(f"Number of parts: {composite.item_count}, purpose: {composite.purpose}")
print(f"First part type: {type(composite[0])}")
try:
print("Trying to convert to MultiPartImage")
img :cvb.MultiPartImage = cvb.MultiPartImage.from_composite(composite)
print(f"Size of image: {img.size.height}, {img.size.width}")
except Exception as e:
print(f"Error: {e}")
chunk_timestamp_node = node_map["Std::ChunkTimestamp"]
print(f"Chunk timestamp: {chunk_timestamp_node.value}")
stream.stop()