is anyone here?

Hi @nmt First, I saved a binocular video of the process by displaying nine dots in sequence on the screen while my eyes focused on the nine dots and my head held still on a tripod. Then, ellseg is used to calculate the location of the pupil center. Like pupil capture, the pupil was calibrated with five points. I manually selected the pupil information corresponding to each calibration point and used Gazer2D() for calibration. Finally, all pupil positions mapped were displayed on the screen. The result is showed below:

@nmt After I removed (norm_x * norm_y, norm_x_squared, norm_y_squared, norm_x_squared * norm_y_squared) in _polynomial_features, things got better

Hi there! My project involves using the Pupil Core (this old model https://hackaday.com/2013/02/12/build-an-eye-tracking-headset-for-90/) and a portable EEG while the participant runs a task on the computer. I plan to use LSL to synchronize all three data (stimuli, glasses, and EEG). I'm a bit confused regarding this process. Considering that everything is compatible with LSL, I'll have to download the Pupil LSL Relay plugin, LabRecorder (does it works as a hub?), and add some plugin/code in MatLab. Is that correct? Can I have the data separately later, or LabRecorder saves everything as just one file?

One other thing related to this version of the core. I was able to run the software in it, however, the image quality of the eye camera is really weird and it's not even detecting the pupil. The image is really dark (if I change some settings it gets red but still really dark) but I'm not sure if it's an issue with the camera or some configuration in the software. Do you have any recommendations for this?

Hello, I would ask a question about post-hoc gaze calibration. These are my steps: I record my data from a Raspberry pi, then post-hoc the data via a desktop PC. I calibrate the eye tracker on the Raspberry pi and unchecked the pupil detection during the recording. Then, on the desktop side, I select post-hoc pupil detection. My situation is: when I only select Post-Hoc Gaze Calibration, I do not have data on gaze direction. I only get data on gaze direction when I press the button Detect References (see figure bellow). And my question is: is it the right way for having the gaze direction data? Or do I need to do other steps? Thank you in advance in your instruction, have a nice day.

hi team! I am very new to the software, In which file formats should recordings be to be uploaded to pupil player?

can I use recordings that have not been generated with Capture?

Hi everyone, I am currently suing pupil core ET with older adults and I am struggling to record good data because they usually wear thick glasses and it is throwing off the calibration or during the recording session. I was wondering if there is a way around this. Thanks

hello, there is some tutorial to acquire gaze position from pupillab to unity?

Hello! My Lab is planning to record data from two headsets simultaneously and we where trying to figure out if there's a way to run two headsets simultaneously on the same PC or if we will need to use separate PC's?

Hello, I have a question about downloading the timeseries + scene video. For one video, I've tried downloading the timeseries + scene video from different computers, but the download continuously fails. The download works just fine with the other videos. Is there a way around this issue at all?

@nmtWhat should happen when Core is operated in a completely dark room, where the eyes are completely dilated and there is nothing to fix the gaze on> For example, imagine as if operating core at the depth in a murky void?

@user-3578ef, what are you planning to do in such a dark environment?

We've only just done some experiments with a diver in a darkened tub with a monitor but I noticed that when there is some illumination from the monitor reducing the pupil size, its initial tracking might be more robust than when the tub is dark and the pupil initially fully dilated. In fact the initial tracking of pupil diameter is rather bizarre, but there could be plenty of reasons in our setup for that. Only first experiments for now...

Indeed that seems to be the case. But then that might create tracking problems when tasking requires looking at the bright display. It seems it would be better to have the minimum diameter auto-regulated somehow to have a seamless response behavior so that the 3D eyeball tracking can maintain valid pupil diameter estimates.

I also edited my message. But the open question does this large maximum degrade performance in the normal use case where the gaze is turned to the bright screen? The normal pupil sizes are in the range of 2 - 4.5 mm as recommended minmax settings but due to scaling issues related to our eye camera position, these numbers are inherently larger, maybe 2X to begin with and might degrade the Pye3D tracker. I suspect at some point my numbers become unworkable and the pupil diameter dropouts occur.

Neil, notice the large dead zones, and that I do get diameter tracking about midway through the episode and I can see some small reasonable diameter changes during that time. But then it drops again, etc. And I can see from the erratic FPS that tracker is struggling during the dead zones.

Thanks Miguel, I'll check about that with the lab - but I don't think so. And the only time I see this is in "dark" condition and it is consistent across multiple sessions.

It's also worth checking that the eye camera exposure time was appropriate for the dark condition. Exposure time can be set to automatic or manual

Can I read the min max pupil size data from the recording folder to check it?

Hi, I got a question about the data gained by pupil core. From 'pye3d 0.3.0 real-time' method, the left and right pupil's diameters are similar in pixels, while the diameter_3d in the unit of millimeter are quite different, the value of the right eye is even twice that of the left eye. Can anyone explain why?

Hello, I am not getting audio in my core recordings. I see two options in the audio mode in Capture: Sound Only and Silent. I have it set to Sound Only, but my world recordings have no sound. Is there an adjustment or fix for this? Thank you so much!

Hi,

I did a short recording to test whether the calibration markers on a curved screen were detected.

I then ran post hoc calibration - accuracy was 1.7 degrees.

However, the visualisation was a bit odd as you can see in the first screenshot - the two green dots were always connected by the pink line during the calibration process as my eyes moved around the screen. Most of the time it's in a straight line, but sometimes there's a lot of zigzagging like in the 2nd screenshot. Do you know what's going on here please?

Re: the FPS, I don't know why it's so low in the screenshot but in the recording everything is smooth and FPS averages around 23.

Hello, I am setting up an experiment. I need to measure saccades, however the LSL data recorded is only at around 30 FPS. My question is, can I increase the frame rate of the LSL broadcast from pupil capture to 200 Hz? if So how can it be done?
Thanks,

Hello, Does Pupil Core work well for doing screen-based eye tracking research?

For example, using a chin rest?

Thank you. Will that yield more than just AOI, ie scanpath?

It will also give you heatmaps, for scanpaths you will need to follow this tutorial https://github.com/pupil-labs/pupil-tutorials/blob/master/03_visualize_scan_path_on_surface.ipynb

Thanks again! My other question: can you record input like mouse events so they are already synchronized with eye tracking data or do you have to synchronize them in post processing?

or a way to log mouse events by yourself and send the corresponding event annotations

From what I remember sampling frequency could either be 60 Hz or 120 Hz. So, I am not sure what this means. I want to interpolate gaze data to calculate saccades so I need sampling frequency.

If anyone here has experience calculating saccades then please reach out! I am trying t implement someone's code and I have some questions on how to do that in PL. Thanks!

Hi @nmt, many thanks for your reply. It is reassuring to know that the coordinate system can be adapted to the image. I have a new question. On pupil PLAYER the frame number does not match the frame number of the world video exported from pupil PLAYER. Why is this happening?

Hi PupilCore, I am currently analyzing the data recorded from the eye tracker and loading them from the exported csv file. I noticed that the diameter_3d column has some missing values in between some intervals ( i am not sure if the duration of these intervals is constant) . I just wanted to know if this is normal and expected. Secondly, is it possible to reduce the sampling rate of the recorded data? as currently, it is recording 200 data points per second and if we would like to sync it with other devices with different sampling rates, do you have any tips on how best to handle this? Thank you!

Sorry, I just noticed you have answered the sampling rate question above, i will try that.. so just the first question is unclear to me now.

Hey @user-f1866e! The search bar is up in the right corner 🙂

I just find a significant delay in realtime data. topic, payload = Globs.pupil_subscriber.recv_multipart()

it is 2-3 seconds of delay

I tried to monitor normalized pupil direction in pupil group. Any idea???

I doubt its the computer.

Thanks! Delay happens in my tkinter routine.

oops, wrong channel. sorry!

Frame rate

Hi PupilLabs, I tried to use pupil player to do the offline detection, but it only saved the offline_pupil data, while missed blink data and so on, how can I get other data?

Hi PupilLabs community, I have question regarding baseline correction -- So if in case the baseline has not been recorded during the experiment, can we create a baseline before the stimulus was presented and the participants were reading a 'Welcome' text. Thanks

Frame indices

Hello, I want to ask is there anyway we can extract the size of the AOI just to record it? the width and height of each AOI?

Hi everyone! The Pupil Player says “no fixation detected” but it genuinely showed me heat maps and fixation points last time I opened it. What’s wrong? What should I do to detect the fixations again? Thanks in advance!

hi @nmt can we determine the size per AOI? is there any way we can record the size in pixels (height and width)

Hello, I am wondering how I can use the face blurring function that is described in the facemapper enrichment. I cannot find anything about in the documentation, because there it says it will be coming soon.

Hi, we have the pupil eye tracker glasses (pupil invisible) we are trying to connect the glasses to our lsl. According to the instructions we download the pupil capture software (on windows10), but although the glasses were connected to the computer we get the attached errors [and see gray screen]:

we also tried to follow the instructions under the windows section here: https://docs.pupil-labs.com/core/software/pupil-capture/

but nothing work 😦 what can we do ?

Hi team, do you have any other documentation that shows how to generate and print the april tags? I tried to follow this page but the links are not working: https://pupil-labs.com/releases/core/036-marker-tracking/

Hello there

Im johan Lara

Im about to buy the pupil core model for sport shooting purposes

i wanted to know if using the software + pupil core require any specific knowledge in terms of code writing ?

or if i can connect the pupil core to my mac, and start using it as a plug and play system

(im totally new, sorry if its a dumb question !!)

I tried but I have the followed error:

Hi, somehow, the video I recorded through the phone cannot be uploaded to the cloud. do you have any idea why?

Hi, team! I have read the code about calculating accuracy, the final accuracy is calculated as the average angular offset (distance) (in degrees of visual angle, only look for values greater than cos(5deg)) between fixations locations and the corresponding locations of the fixation targets. I have a question about the final accuracy. If the average error of one or two fixation target is 1.8 deg, but the rest fixation targets have small deg error, the final accuracy is also small, such as 0.5deg. Should we consider this calibration a failure?

Hello rcd 5213 mgg 6348 nmt 1123 I m

We bought a pupil labs neon mobile eye tracker. We wanted to see the fixations during the recording of the video but unfortunately when we watch the video there was a message saying no fixations found. Could you help us how to setup the eye tracker to resolve the problem.

Hi @user-480f4c Yes, I have already received the email. Thank you!

Unfortunately, I use DIY headset and 2D calibration code. My teacher said that this result is not good enough. I decided to test the algorithm by making some simulation data as ground truth data.

Recommend, if you can, implementing a custom 2d calibration choreography where the marker moves around and covers all areas of the screen - that should improve region-specific accuracy values.

That should trigger a reupload

Hi, i'm trying to run main.py from this link (https://github.com/pupil-labs/pupil), but after i follow all steps and run i keep getting this error, can anyone help me?

Hi, we pressed "C" after setting up the communication between Unity and pupil capture, but we did not see the calibration mark on the headworn display, and the calibration mark on the computer screen did not change either. can anyone help me?

Hi i have the Pupil Core and Can't seam to get it to work, would anyone be able to help with the "world - [ERROR] calibration_choreography.screen_marker_plugin: Calibration requiers world capture video input." Issue?

The main view with controls shows nothing

hi, we have the pupil labs core and just track instruments on a screen. We noticed a difference between the normal lens and the fish eye lens. Do we have to consider a deviation in the coordinates with the fish eye lens or is this already compensated?

Hi @user-c2d375 .After the calibration is completed, the gaze visualizer is still displayed in the scene. How do I hide the gaze visualizer?

Hi @nmt I need help on how to upload our recording from the local drive to pupil cloud

What we want to see are the fixations in the player so we can see the location where the user looks

@user-d407c1 are using pupil core. We want to visualize the fixations when we play the recording using Pupil player but we can’t visualize it. The link you shared about fixation just shows the terminology about fixations. Is there a way for us to visualise the fixations?

Hi, I have a pupil core device for an eye-tracking experiment. I am wondering if there is any limitation of the eye tracking video recording length, like no more than 40 minutes?

Will do the splitting, thank you very much!

Hi, I have pupil labs for HTC vive. I wonder what camera sensor are you using? They seem very cool for a lot of DIY projects I have in mind. Could you share? because I don't want to take them apart, quite expensive gear 😄

Thank you for your reply @user-c2d375 , but the gaze visualizer I mentioned is in hmd-eye, as shown in the following figure. After the calibration is completed, it still displayed in the scene, which will affect the experience of virtual scenes. I would be grateful if you could help me.

hi I am having an issue with pupil labs head mounted eye tracker V.3.5.1. One of the eye video captures is not working (eye1) it just give grey color. I did try to uninstall and install drivers didn't work. I unplugged the cam and plugged it again is not working. We already purchased a new camera for this eye and it was working until today the issue occurred again. It is giving me error Video_capture.uvc_backend: could not connect to device. No images will be supplied. No camera intrinsics available. Consider selecting a different resolution

i tried different pc not working too

Hi I am trying to read all the topics from the zmq with the code in the helper (https://github.com/pupil-labs/pupil-helpers/blob/master/python/filter_messages.py) to check if the surfaces are well sent over the interface I have uncommented the line to receive all the topics. Some topics are well displayed but I quicly run into an error related to UTF-8 awlays after the topic "frame.eye.X" "UnicodeDecodeError: 'utf-8' codec can't decode byte 0x80 in position 33: invalid start byte" Any idea how to deal with that Thanks for the help

Please post your output over in the software-dev channel 🙂

Sorry I am still unsure what goes where

Unfortunately, I have a defect of eyecamera 0. A replacement part is on its way. Since our experiment is currently at a standstill and we are absolutely pressed for time, I will temporarily use a single camera setup.

Did I understand the Swirski and Dodgson paper mentioned in previous threads correctly that no user calibration is required? Is there anything else I need to consider when setting up PupilCapture for the single camera setup? For the single camera setup, are there any other tips or important papers I should have read somewhere? I understand that contralateral eye movement may not be captured robustly. Thank you. 🙂

https://www.researchgate.net/profile/Lech-Swirski-2/publication/264658852_A_fully-automatic_temporal_approach_to_single_camera_glint-free_3D_eye_model_fitting/links/53ea3dbf0cf28f342f418dfe/A-fully-automatic-temporal-approach-to-single-camera-glint-free-3D-eye-model-fitting.pdf

Hi, I'm trying to write a simple c++ script to open the camera and capture some images using the Core. I can open the camera and get images using the Pupil Labs version of libuvc, however it seems like the IR emitters are not turning on. When I run Pupil Capture , or use guvcview, the camera displays the eye clearly. Through some testing, it appears that with my code the IREDs aren't turning on. I tried looking through the Pupil documentation, but I couldn't find anything about controlling the IREDs. This problem exists on both my VM running Ubuntu 22.04.2 and the current Rasbian release on a Pi 4. Any help is appreciated. This may also pertain to the software-dev channel, so I can move it if needs be. I have some images of the camera outputs in each situation if that would be helpful.

Hi, I am trying to use the gaze position to plot a heatmap through python. I understand that the origin of position should be the center of marked surface, but I am confused about the unit of position. For example, I had a gaze position (1.2, 0.7), what is the unit of this "1.2"? Thank you.

Thanks for the follow up, I have tried using the uvc_set and uvc_get brightness, but I would get segmentation faults or pipe errors respectively. I have been trying to understand what a pipe error is in this context. I know there is the detech_kernal_driver flag, I'm not sure what value that should be set to though.

Hi @user-53a8c4, How would we calculate angular accuracy in the 2D calibration case? I have apriltag markers because of which I have depth info for each world frame as well.

Has anyone tried to replace the world camera with a Go pro?

Hello, we have been experiencing an issue with Pupil Core + Pupil Capture that they won't lock on the pupil of participants, instead, the pupil ROI kept jumping across the eye camera window. Resetting Pupil Capture settings did not help the situation at all. This issue happened only when the participants had light iris colors, such as light blue. Has anyone faced similar problems before? Does the Pupil Capture have some difficulty detecting pupils with certain iris colors?

Hello ,where can I get the NenoNet's infomation?

I'm curious, what are some of the best eye tracking analysis softwares out there for Pupil Labs data that you all are using?

Hi team, we are using live calibration with the Core, however sometimes one eye has pretty good pupil detection while the other has poor detection, and this throws off the gaze estimation. Is it possible to only take the pupil detection for one eye into account when calculating gaze estimation? Thank you!

Hello, I have designed an experiment where I have integrated Pupil Core with Psychopy and recorded monocular data. Now when I was trying to analyze the HDF5 file I saw that there is no zero values for blink in the Pupil column but in the Pupil Labs generated file zeros are present. Is there any reason for this ?

Hi, im trying to implement this code about a 9 point callibration system, but it relies on some plugins I cant find where to download, are any of you familiar with them? They are the following: from calibration_choreography.screen_marker_plugin import ScreenMarkerChoreographyPlugin from calibration_choreography.base_plugin import ChoreographyModen

I found them in this link: https://gist.github.com/papr/339dcb08caef45d3798a68aa4e619269

Hi. can pupil core calibrate for nystagmus?

Hi team, after recording, i noticed that sometimes the offline_data folder is either not created or only has limited token files in it. Will this cause an issue when trying to replay the files in Pupil Player? As I tried to load a recording directory that does not have the offline_data at all and it works but when I loaded a file of about 3GB with some token files loaded in the offline_data folder, Pupil Player is not responding.

@user-a11557 if you place it on the folder, it should be available to you in the calibration menu. Unlike other plugins this, does not need to be enabled on the plugin menu.

hello,i am using pupil labs core with psychopy . i have done everything according to the instructions but psychopy keeps crashing after validation step and it shows me this error

https://discord.com/channels/285728493612957698/973685431495426148/1091120240286502932

my psychopy trial structure is shown below

Hello, could you please assist me with an inquiry? I have recently attempted to use a DIY eye tracker with Microsoft HD-6000 and Logitech C525 cameras. Unfortunately, when attempting to run the pupil-3.5, I encountered issues as it was unable to detect the cameras. Despite trying to run the program with administrator privileges, the problem persisted. I can confirm that both cameras are displayed in the system's camera devices under the camera section, and my computer is running on the Windows operating system. Would you be able to provide me with any suggestions on how I can rectify this situation and enable the pupil-3.5 to detect the cameras? Thank you in advance for your assistance.

Hi I am using Pupil mobile app version 0.37.0 on android 11, google pixel 2 phone. I am trying to find a compatible version of pupil capture. What would you guys recommend ? Also what are the step to connect the app to the pupil capture ?