Coordinate system

[IndiGleb]

Required Info
Camera Model	{ R200 / F200 / SR300 / ZR300 / D400 }
Firmware Version	(Open RealSense Viewer --> Click info)
Operating System & Version	{Win (8.1/10) / Linux (Ubuntu 14/16/17) / MacOS
Kernel Version (Linux Only)	(e.g. 4.14.13)
Platform	PC/Raspberry Pi/ NVIDIA Jetson / etc..
SDK Version	{ legacy / 2.. }
Language	{C/C#/labview/nodejs/opencv/pcl/python/unity }
Segment	{Robot/Smartphone/VR/AR/others }

Issue Description

<Describe your issue / question / feature request / etc..>

How i can get coordinate from camera (D455) ? Maybe there is a function that defines this ? As a result, I want to get a matrix containing coordinates in each pixel

[MartyG-RealSense]

Hi @IndiGleb It sounds as though exporting the data to a .csv file may fit your needs. A .csv file is a textual file of all the coordinates that can be imported into a database or spreadsheet. A csv could be exported directly with a program script, or you could record a bag file of the camera streams and then convert the bag to csv format with the RealSense SDK's rs-convert tool. This subject is discussed at #11090

[IndiGleb]

I saw that in the view you can find out the values of coordinates at a point. My task is to get a three-channel matrix containing coordinates (x,y,z)

[IndiGleb]

What is this method ? I search him in SDK, but i can't find that

[MartyG-RealSense]

You could generate a pointcloud with pc.calculate and then extract and print the XYZ values of each coordinate with the instruction points.get_vertices. #4612 (comment) has a Python example of points.get_vertices whilst #5728 is a C++ example.

[IndiGleb]

I can measured coordinate without pointcloud ?

[MartyG-RealSense]

Obtaining the 3D world-space coordinate of a single specific pixel coordinate on an image without using a pointcloud or alignment can be done with the instruction rs2_project_color_pixel_to_depth_pixel and information about this instruction is at #5603 (comment)

[IndiGleb]

where can I find this method ?

[MartyG-RealSense]

Are you using Python or C++ please?

[IndiGleb]

С++. I want to know where the measured coordinate from pointcloud and rs2_project_color_pixel_to_depth_pixel

[MartyG-RealSense]

A C++ example of using rs2_project_color_pixel_to_depth_pixel is at #6239 (comment)

For the pointcloud, using points.get_vertices like in #5728 to obtain the coordinates will likely be the best method to use.

[IndiGleb]

In which cpp source file can i find what is contained in these methods ?

[IndiGleb]

I'm trying to figure out exactly how the coordinates are calculated

[MartyG-RealSense]

These two sections of the SDK's source code may be especially relevant:

rs.cpp

librealsense/src/rs.cpp

Lines 3762 to 3769 in 3371f4d

	void rs2_project_color_pixel_to_depth_pixel(float to_pixel[2],
	const uint16_t* data, float depth_scale,
	float depth_min, float depth_max,
	const struct rs2_intrinsics* depth_intrin,
	const struct rs2_intrinsics* color_intrin,
	const struct rs2_extrinsics* color_to_depth,
	const struct rs2_extrinsics* depth_to_color,
	const float from_pixel[2]) BEGIN_API_CALL

unit-tests-live.cpp

librealsense/unit-tests/unit-tests-live.cpp

Lines 5220 to 5233 in 1b9a40d

	// Search along a projected beam from 0.1m to 10 meter
	rs2_project_color_pixel_to_depth_pixel(to_pixel, reinterpret_cast<const uint16_t*>(depth.get_data()), depth_scale, 0.1f, 10,
	&depth_intrin, &color_intrin,
	&color_extrin_to_depth, &depth_extrin_to_color, from_pixel);
	float dist = static_cast<float>(sqrt(pow((depth_pixel[1] - to_pixel[1]), 2) + pow((depth_pixel[0] - to_pixel[0]), 2)));
	if (dist > 1)
	count++;
	if (dist > 2)
	{
	WARN("Projecting color->depth, distance > 2 pixels. Origin: ["
	<< depth_pixel[0] << "," << depth_pixel[1] <<"], Projected << "
	<< to_pixel[0] << "," << to_pixel[1] << "]");
	}

[IndiGleb]

Post-processing work when camera enable on ethernet ?

[MartyG-RealSense]

My understanding from a RealSense team member's advice at #6376 is that if a network camera running on the rs-server ethernet networking system has a serial number defined in a script then it should behave in that script like a normal non-networked camera. I cannot recall a previous case of post-processing being applied to a network camera though.

The rs-server ethernet system is being removed in the next RealSense SDK version however, so it would not be a suitable networking solution to choose if you plan to update the SDK in future.

The EtherSense Python-based ethernet system should continue to work.

https://github.com/IntelRealSense/librealsense/tree/master/wrappers/python/examples/ethernet_client_server
https://dev.intelrealsense.com/docs/depth-camera-over-ethernet-whitepaper

[IndiGleb]

how i can disable post-processing in code ?

[MartyG-RealSense]

You do not need to disable post-processing filters in code when creating your own program as they are not active by default and have to be deliberately programmed into a script. They are only enabled by default in the RealSense Viewer tool.

[IndiGleb]

How i can check this ?

[MartyG-RealSense]

You do not need to check if the filters are disabled if you have not programmed them into your script as the filters will not be applied. This is because the program does not know how to use filters unless you tell it how to by adding post-processing filter code.

[IndiGleb]

how will ethersense differ from rs-server ?

[MartyG-RealSense]

They share the general principle of being able to send data from a camera attached to a computer (the remote server) to another computer with a display and no camera attached (the host) but otherwise they are completely different in how they work.

[IndiGleb]

How the image will be sent over the network?

[IndiGleb]

BGR will remain jpeg ?

[IndiGleb]

Or maybe H264?

[IndiGleb]

Where can I read more about EtherSense ?

[IndiGleb]

I mean how the image will compress both RGB and depth ?

[MartyG-RealSense]

It is sent by the TCP protocol in EtherSense, whilst rs-server uses the RTSP protocol.

There are not further information resources about EtherSense, unfortunately. The paper linked to earlier is the best guide available.

https://dev.intelrealsense.com/docs/depth-camera-over-ethernet-whitepaper

It appears from the section of the paper linked to below that the streams are uncompressed, but advice is offered on different ways to reduce bandwidth.

https://dev.intelrealsense.com/docs/depth-camera-over-ethernet-whitepaper#network-bandwidth

[MartyG-RealSense]

Hi @IndiGleb Do you require further assistance with this case, please? Thanks!

[MartyG-RealSense]

Case closed due to no further comments received.