Hi
expert
I'm tring to get 4ch raw image.
I believe serdes and sensor setting are correct.
I confirmed 1ch image is saved using 4ch serdes and sensor setting value.
but I can not see the images of 4ch.
error
Q1. I want to know how to separate each image on CSIRX. maybe virtual channel?
Q2. How to recognize 4ch image to CaptureNode() and save each image to buffer(obj->cap_frames).
Q3. Is attached code correct to save 4ch raw image?
vx_status app_klk_create_graph(AppObj *obj){
/* common */
vx_status status = VX_SUCCESS;
vx_context context = obj->context;
vx_int32 file_size = 0;
tivx_capture_params_t capture_params;
/* sdk */
tivx_raw_image raw_image = NULL;
vx_user_data_object capture_config;
if (vx_true_e == tivxIsTargetEnabled(TIVX_TARGET_CAPTURE1))
{
obj->graph = vxCreateGraph(context);
if (vxGetStatus((vx_reference)obj->graph) != VX_SUCCESS){
printf("[ERR] vxCreateGraph fail\n");
}
/***Initializing params for capture node***/
memset(&sensorParams, 0, sizeof(sensorParams));
sensorParams.sensorInfo.raw_params.width = 1280U;
sensorParams.sensorInfo.raw_params.height = 944U;
sensorParams.sensorInfo.raw_params.num_exposures = 1U;
sensorParams.sensorInfo.raw_params.line_interleaved = vx_false_e;
sensorParams.sensorInfo.raw_params.format[0].pixel_container = TIVX_RAW_IMAGE_16_BIT;
sensorParams.sensorInfo.raw_params.format[0].msb = 11U;
// sensorParams.sensorInfo.raw_params.format[1].pixel_container = TIVX_RAW_IMAGE_16_BIT;
// sensorParams.sensorInfo.raw_params.format[1].msb = 11U;
// sensorParams.sensorInfo.raw_params.format[2].pixel_container = TIVX_RAW_IMAGE_16_BIT;
// sensorParams.sensorInfo.raw_params.format[2].msb = 11U;
sensorParams.sensorInfo.raw_params.meta_height_before = 0U;
sensorParams.sensorInfo.raw_params.meta_height_after = 4U;
raw_image = tivxCreateRawImage(obj->context, &sensorParams.sensorInfo.raw_params);
vxSetReferenceName((vx_reference)raw_image, "RAW_IMAGE");
//for(buf_id=0; buf_id<obj->num_cap_buf; buf_id++) {
obj->cap_frames[0] = vxCreateObjectArray(obj->context, (vx_reference)raw_image, 4);
status = vxGetStatus((vx_reference) obj->cap_frames[0]);
if(status != VX_SUCCESS){
printf("[ERR] vxCreateObjectArray fail\n");
}
//}
tivx_capture_params_init(&capture_params);
capture_params.numInst = 1U; /* Configure both instances */
capture_params.numCh = 4U; /* Single cam. Only 1 channel enabled */
capture_params.timeout = 33;
capture_params.timeoutInitial = 500;
capture_params.instId[0] = 0u;
capture_params.instCfg[0].enableCsiv2p0Support = (uint32_t)vx_true_e;
capture_params.instCfg[0].numDataLanes = 2u;
capture_params.instCfg[0].dataLanesMap[0u] = 1u;
capture_params.instCfg[0].dataLanesMap[1u] = 2u;
capture_params.instCfg[0].laneBandSpeed = TIVX_CAPTURE_LANE_BAND_SPEED_1350_TO_1500_MBPS;
int cnt = 0;
for (cnt = 0u; cnt < TIVX_CAPTURE_MAX_CH; cnt ++)
{
capture_params.chVcNum[cnt] = cnt;
capture_params.chInstMap[cnt] = 0U;
}
capture_config = vxCreateUserDataObject(context, "tivx_capture_params_t",
sizeof(tivx_capture_params_t), &capture_params);
if (vxGetStatus((vx_reference)capture_config) != VX_SUCCESS) {
printf("[ERR] vxCreateUserDataObject fail\n");
return VX_FAILURE;
}
vxSetReferenceName((vx_reference)capture_config, "CAP_CFG");
obj->capture_node = tivxCaptureNode(obj->graph, capture_config,
obj->cap_frames[0]);
if (vxGetStatus((vx_reference)obj->capture_node) != VX_SUCCESS){
printf("[ERR] tivxCaptureNode fail\n");
return VX_FAILURE;
}
vxSetReferenceName((vx_reference)obj->capture_node, "CAP_NODE");
status = vxReleaseUserDataObject(&capture_config);
if(status != VX_SUCCESS){
printf("[ERR] vxReleaseUserDataObject fail\n");
}
status = vxSetNodeTarget(obj->capture_node, VX_TARGET_STRING, TIVX_TARGET_CAPTURE1);//sdk는 TIVX_TARGET_CAPTURE2
if(status != VX_SUCCESS){
printf("[ERR] vxSetNodeTarget fail\n");
}
status = vxVerifyGraph(obj->graph);
if(status != VX_SUCCESS){
printf("[ERR] vxVerifyGraph fail\n");
}
status = tivxExportGraphToDot(obj->graph, ".", "klk_single_cam");
if(status != VX_SUCCESS){
printf("[ERR] tivxExportGraphToDot fail\n");
}
//run graph
status = appStartImageSensor("OX01D10", 1); // setup serdes, sensor
tivxTaskWaitMsecs(5000);
printf(" [INFO] Executing graph ... !!!\n");
vxScheduleGraph(obj->graph);
vxWaitGraph(obj->graph);
printf(" [INFO] Executing graph ... Done !!!\n");
obj->raw = (tivx_raw_image)vxGetObjectArrayItem(obj->cap_frames[0], 0);
file_size = write_output_image_raw("/ti_fs/vision_apps/raw_image0.raw", obj->raw);
printf(" [INFO] raw_0 file size:%d\n", file_size);
obj->raw = (tivx_raw_image)vxGetObjectArrayItem(obj->cap_frames[0], 1);
file_size = write_output_image_raw("/ti_fs/vision_apps/raw_image1.raw", obj->raw);
printf(" [INFO] raw_1 file size:%d\n", file_size);
obj->raw = (tivx_raw_image)vxGetObjectArrayItem(obj->cap_frames[0], 2);
file_size = write_output_image_raw("/ti_fs/vision_apps/raw_image2.raw", obj->raw);
printf(" [INFO] raw_2 file size:%d\n", file_size);
obj->raw = (tivx_raw_image)vxGetObjectArrayItem(obj->cap_frames[0], 3);
file_size = write_output_image_raw("/ti_fs/vision_apps/raw_image3.raw", obj->raw);
printf(" [INFO] raw_3 file size:%d\n", file_size);
// file_size = write_output_image_nv12_8bit("/ti_fs/vision_apps/yuv_image1.yuv", obj->y8_r8_c2);
// printf(" [INFO] yuv file size:%d\n", file_size);
//release
} else {
printf("[ERR] tivxIsTargetEnabled fail \n");
}
return status;
}
I changed a part of code to save 1ch image. only A port image.
line44: obj->cap_frames[0] = vxCreateObjectArray(obj->context, (vx_reference)raw_image, 1);
line51: capture_params.numCh = 1U;
Thank you
