Hey guys, working on some blink rate code in 🕶 invisible , has anyone produced something to plot/determine when in a recording the highest number of blinks occurred?

No way I missed a feature in Pupil Cloud, right?

Yeah, the code is pulling from a raw data export - it's using blinks.csv and events.csv. Most recent developments are in #1047595857610022972 , but I can't for the life of me get the code that either of us made to plot the way I need it to

hey hey, I am trying to install and test piupil core on macos ventura. the problem is that the system sees your hardware as a keyboard and I have no clue how to change this. is there perhaps a plug in or other solution that can help?

when plugged in, a driver installation window pops up with a message similar to 'we're trying to install your keyboard'.

and when I turn the program on, alternating messages pop up (see screenshot)

aahhhh ok! thank you very much for super fast respond

Hey, one more question, thanks to your instructions all the cameras are on, now I have problems with the calibration of the world (I went through the Getting Started section). I believe it is Gazer3D that is causing this problem:

WORLD: Calibration Failed! WORLD: Not sufficient pupil data available. WORLD: Plugin Gazer3D failed to initialize

if you mean pupil capture eye 0 and pupil capture eye 1, then yes. also process of calibration seems to be fine, because im getting the green dots - the error above pops up right after

ok ok! got it! sorry

Hello everyone, I wanted to use Pupil_apriltags to detect Apriltags in real-time for two Pupil invisible video streams at the same time. However, I encountered a Segmentation Fault. I was curious as to what was causing the problem. Is Pupil_apriltags using just a single thread to detect tags in frames?

Hey @papr What does the eye0_lookup.npy used for? In case it is not available in a certain recording, is there any way of restoring it?

It is a cache file that helps Pupil Player seek efficiently. Player generates it based on the timestamp files if it is not present.

Is there any solution to solve the issue? at the moment, i'm using mutex but i would like to know your idea.

Hello @papr

I added surfaces markers to my monitors display and I defined a surface with the resolution of my monitor. How I map those coordinated to a screenshot taken on that monitor from surfaces.pltdata file?

Thanks

That worked! Thanks @papr

Is it possible to stream 'x_scaled' and 'y_scaled' through LSL?

Hi, in my research environment I'm not allowed to use the Pupil Cloud. So I would like to create a pipeline that creates a recording, does the fixation extraction via Pupil code, then does further processing with my own code. As far as I understand if you want to have the Invisible recordings on your computer you should use Pupil Capture instead of the Companion App to do the recordings. But using Pupil Capture I can't automatically start and stop recordings via the realtime API. Any ideas how to resolve this?

Got it, then I'm only able to do recordings via Companion and there is currently no automatable way to get recordings from the companion device to a computer. Would it be feasible to create API methods to export recordings via realtime API? Or maybe event transfer the recording via API?

Hello. I'm working on some research which requires me to read data from the surface tracker plugin API. I have a few questions surrounding the rate at which data comes in and the amount of data. Specifically, I need to check whether or not the user is looking at a surface. In each message, there are 6-10 boolean values that convey this. Is there any way to standardize the number of booleans per message? Furthermore, the rate at which the data comes in seems pretty variable. Is there any way to standardize the rate at which messages are sent? Finally, is there any way to speed up the rate of messages or is the best option using a computer with more performance power?

Okay, also it is only the scene video as far as I know (didnt test it) that is streamed. To make the Export function of the companion app available over the realtime-API is also not possible for the moment right? Then I could track changes in the "Pupil Invisible Export" folder of the comp device

Okay, thanks for your input👍

Hello, could you help me? I'm doing the scan path basically using the pupil labs example but I have a problem that the generated plots always end up centered on the image

Hey, which file do you read the data from?

Also, not sure where I read your questions, but you will need to multiply the normalized surface positions by the image size to get the corresponding Pixel locations

I'm reading from fixations, I'm already multiplying by the size of the image

So is it fixation_on_surfaces or gaze pozitions on surface?