[Resolved] Server error

[amsabuncu]

I can't get past the following error:



Is a reboot necessary? Or is it some problem with a TCP/IP socket left dangling?

[amsabuncu]

Server would not start no matter what. Had to do a reboot.

Maybe during the last boot, the device was not connected. This time, the device was connected during system boot.

[Martin]

And it works now? The message you see typically means that the device is connected but it cannot, for some reason, start streaming and tracking. I've notified the developer who wrote that routine to get some more feedback. Please let me know how it goes.

[amsabuncu]

Martin, the server is currently working. However, I do have other problems.

[amsabuncu]

By the way, why is the server a foreground executable? Shouldn't be too hard to make it into a proper Windows service. There is even software that takes any executable and turns it into a service, if all else fails.

Right now, both pieces of your software, the server and the GUI, are unstable and nowhere near mature. At least make the server a service with error logging, debugging, etc. It is really difficult to have two unknowns when dealing with a new technology. Had the server been stable, I could think about writing my own GUI.

[Martin]

Point taken. Requested feature are scheduled but could not be met for the December deadline. We could have done an ordinary Kickstarter-delay and shipped it 6-12 late but I like to keep promises.

Running it as a proper service was left out among some other things in order to meet deadline, we could not verify it properly on all Windows versions in given time. We knew that plain console app would work on all platforms. Logging and generating support requests are coming in the next version.

When the server and hardware is up and running the output from the server console will read:
"The Eye Tribe Tracker stands ready!" This means that the sensor has been properly initialized and that the communication between host and device is working.

The primary source of issues we have seen so far has been related to USB 3.0 drivers, either not installed or wrong version. This is an issue particularly on Windows 7.

Secondary issues stems from placing the device correctly, it should be below the monitor, slightly tilted. When you can see your eyes reflected in the black acrylic plate it's in a good position. The distance is important, from device to eyes ought to be within an arms length.

I would truly appreciate any more details you could provide about your setup and issues your seeing. I'm dedicated to improving the quality and your experience. We will act on your feedback.

[amsabuncu]

I would truly appreciate any more details you could provide about your setup and issues your seeing. I'm dedicated to improving the quality and your experience. We will act on your feedback.

Martin, thanks for your attention. I am using Windows 7 Ultimate x86, ASUS motherboard M4A87TD/USB3, 22 inch monitor. I do not have strange hardware connected. I use my system mostly for text processing.

The problems I am facing with the TET device/software:

1 - I cannot get it to work consistently. When I do get the "green eyes" GUI, I do not know what I did to get it to work. Most of the time the GUI is red. Once in a while it goes green, and as I said, I don't know how.
2 - When the GUI is green, I cannot get it to stay functioning. You have a 3x3 matrix where a node turns red if your gaze is on it. That only works for about 20 seconds, than the gaze functionality hangs. Also, I don't know how this matrix comes up.
3 - I looked at the system performance via Task Manager when the tracker was functioning, and saw that it really overtaxes the entire system. I have 4 cores, and they all seemed to show high utilization.
4 - The GUI is poorly designed. There are options and buttons that should be greyed out that are active all the time. There is no help on what the options/buttons mean.
5 - The forum assistance needs to move away from "have you plugged the unit in" type of approach.

I understand that you want to keep promised delivery dates. But what is delivered is not functioning in a stable manner (at least for me). I had high hopes for this technology but at this point, I can't tell if they will come true.

Hope this helps.

[Martin]

I believe you computer is powerful enough to do the job, my Intel i7 920 is similar in performance and it has no problems keeping up.

The reason you see such high CPU utilization is because the algorithms have difficulties finding and locking on to your eyes. It seems to loose tracking and get stuck in recovering. This can occur for a number of reasons, most often it's related to being too far from the device, having glasses with anti-reflex coating, or too much sunlight in the room. We're working on adding diagnostic features to the software at the moment.

As for the GUI, it is fairly simple, most people won't have to change any settings at all but if you hoover the mouse on the option labels you will see a short description.

You can expect the the robustness to improve as we improve the software. We're working on it every day.

[amsabuncu]

Maybe you can have an option that a user wearing reading glasses can turn on? Could this possibly be helpful to the tracker, knowing that what it should be looking for is a pair of eyes behind glasses?

[amsabuncu]

As for the GUI, it is fairly simple, most people won't have to change any settings at all but if you hoover the mouse on the option labels you will see a short description.

It is pointless to go through the calibration process if the GUI is red. Yet that's exactly what the GUI allows now. If the tracker is not tracking, there's no point in calibration, and calibration buttons should be disabled.

The current GUI does not distinguish between TRACKING and CALIBRATED TRACKING. As a new user, I was under the impression that calibration was the initial and pre-requisite step in achieving tracking. I only learned that this is incorrect, by asking questions on this forum.

Similarly, if the screen size selected is the maximum resolution available on that screen, than there's no point in having the "stretch" setting be active.

The problems I indicate above are the result, I think, of a GUI designed by engineers who are too close to the product. They think that the everything about the product is "obvious" and that anyone should know these. You need to have people off the street test your application for basic useability, first.

And another point: I observed a *huge* discrepency between the quality of the hardware and the packaging, and the functionality of the software. The hardware and the packaging look and feel like they could have been designed by Apple. The software and the useability, on the other hand, are very immature. There seems to be an "impedance mismatch" between the hardware and the software, borrowing from the electronics jargon.