Hello, we want to fix these cables back. Could you please specify the order of the cables? Thank you.

hey, i'm having an open interview with participants wearing the eye tracker and during answering the questions they will likely move their head around a lot. I would prefer not telling them to keep still or fixate their heads so it's more of a real conversation. Is that a big problem for accuracy or could head movements be corrected someway? Thanks for your help 🙂

thank you! 3d gaze mapping seems like a good option! just one more question: I will later on extract all the frames from the world video, use a face recognition algorithm on it and measure in every frame whether participant is looking at the eyes or not. so far, i used the norm_pos_y / norm_pos_x data to determine gaze position in the frame. I'm guessing i would then need to use the gaze_point_3d columns?

correct, i've got facial features in 2d pixel space. i already got it working using norm_pos_x/y. my question would be whether i need to account for 3d gaze mapping somehow when i do this analysis or can i still use the 2d gaze info? hope it's clear 🙂

perfect, thank you!

Hi all, I'm using Presentation software from Neurobehavioral Systems to control my experiment measuring EEG and eye tracking (we have a Pupil Core). I would like to send messages from the PC with Presentation to the PC with Pupil Capture, in order to remotely rec & stop and/or to send annotations, all synchronized with my EEG triggers. From what I understand, sending these messages requires the ZMQ library, which does not seem to be compatible with Presentation. Do you have any solution for sending triggers from Presentation software? If not, are you planning to have a network protocol compatible with it? (I really need to use Presentation for the temporal accuracy it provides, as I present auditory stimuli to the participants).

Thanks in advance! Félix

@user-0d33c1 another option for synchronisation would be to leverage our Lab Streaming Layer (LSL) relay: https://github.com/labstreaminglayer/App-PupilLabs/tree/master/pupil_capture#pupil-capture-lsl-relay-plugin It would seem 'Presentation' also support LSL: https://www.neurobs.com/pres_docs/html/03_presentation/06_hardware_interfacing/02_lab_streaming_layer.htm

Why cant pupil capture/pl core track blue eyes? So far in our research labs we have known about this issues for quite some time, and chosen to ignore it and just measure dark eyed participants. However, given the current lack of dark eyed people within our group and the Netherlands in general it is starting the become a bit annoying. So far I fail to understand why the algorithm is not capable of detecting the pupil here in this example. Calibration of blue/bright eyed participants results in a loss of 80-90% of the data due to low confidence, and any eye trace data we do record afterwards is mostly noise. I notice that dark eyes in infrared camera's appear a bit brighter then bright eyes, and therefor contrast between the iris and pupil is slightly better for dark eyes. Yet, any pupil detection algorithm should be capable to find the pupil here.

I'll appreciate any help to circumvent or solve this issue from my end (i.e. things that i can do here to improve tracking of bright eyed people), but also i feel there should be some work done on the pupil labs pupil detection algorithm.

Note that the photo added below is taken in light circumstances, yet for both bright eye pupil tracking in the light as well as dark this problem exists.

Thanks @nmt ! I'll look at these 2 options. I was also thinking of a backup plan which would be to run a "server.py" on the eye-tracker PC, able to receive the messages from Presentation (because using standard TCP protocol I'm able to receive messages from Presentation in a python server), then forward this message to Pupil Capture, using ZMQ this time. This would be an intermediate interface. The main risk I see is reliability, notably in terms of lags.

In general, do you know if the delay with which remote messages (e.g., "R" for record) reaches Pupil Capture (via ethernet) should be consistent? Or is it random? I'm trying to find a way to quantify that as well.

Generally, I would not recommend estimating these kind of delays. I would rather recommend subscribing to the events published by Capture's Network API. The recording.started notification contains the start time of the recording.

@user-0d33c1 For general information on time syncs see https://github.com/pupil-labs/pupil-helpers/blob/master/python/simple_realtime_time_sync.py

Thanks a lot for all your replies, it's much appreciated! I'll look at all these options.

A last question: if I run a 1-hour experiment with 40 trials, do you think it's reasonably doable to record the Pupil Core continuously 1 hour while sending annotations of the events, or is it necessary to "rec/stop" for each trial? I imagine because of the main video data that may be heavy, the risk of recording 1 hour is to crash the PC, if it's not particularly powerful?

Okay thanks very much! 🙂

Hi - I've sent an email to the info pupil labs account, but we’re currently in-field conducting some eye-tracking today and we’ve had an issue with one of the interviews. I can see in our Cloud area that the file is 9:56 long (it is still currently uploading), but in-field we're only able to see the first 40 seconds-ish of the recording on the phone. Are you able to help us with this and identify what has happened? – We can’t see anything from the device settings etc on our side and we have another session very soon where we don't want the same issue to occur.

File name - 2022-02-02_10:00:22

ID - 9f03e566-d635-4932-a92d-e28d2f5ac377

hello everyone! I desperately need to understand what's wrong. my set up includes two pc's: The first pc has only pupil labs installed (on window) the second pc contains the TASK on MATLAB/PSYCHTOOLBOX with ZMQ library (with zmq-matlab and zmq-matlabmsgpack). the two PCs communicate via local network: via MATLAB (second PC on UBUNTU) the start/end recording and time synchronisation scripts work perfectly. Is there a way to send notifications (/triggers) for each TASK stimulus in psychtoolbox via MATLAB to the first pc (windows with pupil labs)? am i missing something? do i need to install something else? Thank you very much in advance for your help!

If yes, you only need to call the corresponding functions from the example scripts within your psychotoolbox experiment.

yes, if you are asking me if pupil capture starts, yes! if i start the scripts from MATLAB (ubuntu, second pc) pupil capture (first pc with win.) starts recording etc..

yes!

the script crash when I try to send a annotation

the script crash when I try to send a annotation

ok, being several different annotations, do i need to create them a priori on pupil capture? or can i create them by specifying on the matlab script? and thanks a lot for your support!

thanks a lot!

Hi @nmt, thanks for your reply. I believe the 2d detector settings are already at default settings as we use the 3d detector for gaze mapping. We record pupil data with auto aperture setting, so that pupil capture finds it own optimal light setting, then we fix the 3D eye model, and then continue with a the calibration and recording. After the weekend I'll do a test recording with my own (blue) eyes, and share it (as we typically do not store the recordings due to the high noise and I therefor have nothing lying around atm).

Hello, as far as I know that we are able to remotely start/stop Pupil Recorder through zqm (e.g. socket.send_string("R")) and setting the Pupil time (e.g. socket.send_string("T 0.0")).

I am wondering are there any comments that is able to set the recording path and the recording session name?

Please note that we no longer recommend using the T command. See this documentation and example on how we recommend to do time sync https://github.com/pupil-labs/pupil-helpers/blob/master/python/simple_realtime_time_sync.py

Hi, does pupil core support Monocular vision?

Can I use my only one eye at a time?

Thank you so much. I will check it

Hi guys. I am curious how can I read the gaze coordinates over time. The only files I am seeing are gaze.pldata and gaze_timestamps.npy . The latter only shows the time vector. Are the coordinates stored in gaze.pldata? If yes, how can I open them in windows?

Hi @papr . We bought an Intel Realsense depth camera, and we would like to integrate its depth information into the post-hoc calibration routine (i.e. instead of letting the calibration routine to move both the calibration target and the eye orientation, we want to fix the calibration target and let the calibration routine only change the eye orientation). Could you please point out what I should do in addition to installing the depth camera plugin in the pupil capture? Are there any existing code that implements this feature?

hi everyone! I've tried to modify in every way the script "send_annotation.m", but what I get are only the reception with empty timestamp and label! can someone help me to understand how to configure the script for this to happen?

I summarize briefly my set up: I have two PCs connected via LAN, the first with only pupil labs (win) and the second with ubuntu operating system, in which there is the task created with MATLAB/PSYCHTOOLBOX. during the task there will be about 300 trials, I need to send triggers for each stimulus within the individual trials. The two PCs communicate perfectly for synchronization, start/end recording via zmq/matlab. I am stuck with triggers/annotations, because without them it is impossible to analyze 300 trials. Also, I bypassed the use of markers, they are not useful to send annotations for stimuli that I already have with psychtoolbox in the second pc, right? thank you very much in advance!

these are some of the many ways

Yeah, unfortunately the https://de.mathworks.com/help/matlab/ref/containers.map.html docs are not perfectly clear on how to create such an object. Please try

send_notification(
  socket,
  containers.Map(
    {'topic', 'label', 'timestamp', 'duration'},
    {
      'annotation.my_custom_annotation',  % must start with `annotation.`
      'my custom label',
      42.0,  % use time synced value here
      0.0,  % duration
    }
  )
)

thank you so much, really, I'll try right now! thank you so much for your time!

Does the "Binocular Add-on" for VIVE have SDK support for linux. I see the pupil-labs github repos and from what I can tell the company generally supports linux. Specifically I am asking if the Binocular Add-on is supported by pupil-labs SDK's on linux (before I buy one).

Hi, Pupil core can be use with camara web or unique function with device of Pupil

Hi, I was wondering if it was possible to project the coordinates from each of the eye cameras into the coordinate system or the image of the world camera?

So would this gaze estimation function be applicable to recordings and purely using the pupil data? Also is the source code for the gaze estimation function available?

Hi @papr (Pupil Labs) . We bought an Intel Realsense depth camera, and we would like to integrate its depth information into the post-hoc calibration routine (i.e. instead of letting the calibration routine to move both the calibration target and the eye orientation, we want to fix the calibration target and let the calibration routine only change the eye orientation). Could you please point out what I should do in addition to installing the depth camera plugin in the pupil capture? Are there any existing code that implements this feature?

Thank you so much!

Hi~ I have a question about the code in pye3d/observation.py in the image. Why is the direction of the self.gaze_3d_pair "circle_3d_pair[i].center + circle_3d_pair[i].normal" (the red line), not just "circle_3d_pair[i].normal" ? Because I think the normal vector of circle_3d_pair is the same as the gaze vector/direction of 3d_gaze_pair. (the yellow lines). Thank you!

Hey @user-f9f006, thanks for your question. You are right, the directions of the two line segments in gaze_3d_pair are given by the two unprojected normals. Note, however, the line segments have a specific location in actual 3D eye camera coordinates, as they have a start and end point. This is important for calculating their projections into 2D image space (see definition of self.gaze_2d in your code snippet), as well as for defining what in the code is referred to as Dierkes lines (see a few lines later in observations.py). The latter being linear constraints for eye ball position in 3D eye camera coordinates. More details about the involved algorithm for eyeball center estimation can be also found here: https://www.researchgate.net/publication/333490770_A_fast_approach_to_refraction-aware_eye-model_fitting_and_gaze_prediction

Hi! Last week, I noticed that the right eye camera stopped working, even though the world view and the left eye camera both work fine. There were four error messages on the application, which ran in Linux Ubuntu’s system: EYE1: Could not connect to device! No images will be supplied EYE1: No camera intrinsics available for camera (disconnected) at resolution [192,192] EYE1: Loading dummy intrinsics, which might decrease accuracy EYE1: Consider selecting a different resolution, or running the Camera Intrinsics Estimation

@user-f9f006 Thanks for following up and bringing up this issue. I do get your point now. We will look into this further.

@marc is it possible to vary enhance the image quality via pupil player?

the image quality of invisible is worse than core for the same illumination.

Hi, is it possible to use pupil software with cheaper cameras, with a lower frame rate, than the ones suggested? E.g not 200hz but 30hz. What are the implications? Thank you

Hi @nmt , I've recorded some example data for both blue and dark eyes. Our problems with the bright irises remain even after fiddling with the settings (the 2d was set to default). We took the pupil labs outside our labs, and that seemed to improve conditions with recording in the light (Which may be due to lighting in the office being continuous LED, while lighting in our labs is fluorescent tube lighting and thus flickering). However in the dark pupil tracking for the blue eyes is still poor compared to dark irises. I've sent you that data via email as well. Kind regards, Jesse

Hello, we are having problems opening our data in Pupil Player on Windows 10. The data were collected on a Mac, so can this be a compatibility issue? The folder can be opened and processed on the same Mac it is created. This is what happens each time we try to open it in another PC. Thanks in advance!

Hello all, yesterday during an experiment, the software had a hard time detecting my subject's pupil. During the day it was not a problem. (Same person)

It was not due to the orientation of the eye cameras. When I restarted the software, an exception/Error was thrown. "Eye0/Eye1 could not set Backlight compensation" and something regarding ZBeacon.

The solution then was to restart the software with factory settings and set the eye cameras to 192x192p and manual exposure, instead of 400x400p and auto mode.

I don't have the logs unfortunately. I still tried to reproduce the error. At the current startup I get the following warning in the logs.

2022-02-11 09:45:56,454 - eye0 - [WARNING] uvc: Could not set Value. 'Backlight Compensation'.

What can I do to make the "Backlight Compensation" work properly.

System: MacOs Big Sur 11.6.2 (Apple silicon) with the newest Desktop Software

Kind regards,

Cobe.

Hi Cobe. The camera exposes this UVC control but does not actually allow to set it. This warning does not affect the eye tracking and can be ignored. For tracking at night or in the dark, I would suggest setting the exposure manually at the time of recording to optimise the contrast between the pupil and iris

Hi @user-55ae67 👋. Would you mind sharing the names of a few files in the recording folder, e.g. "world.mp4". I want to check if any of them have been renamed.

Thanks for sending that! Please follow these steps: 1. IMPORTANT: Make a backup of the recording 2. Rename any files that have "._" at the beginning of their name. For example: ._world.mp4 should be world.mp4 3. Try loading the recording in Pupil Player on your PC

Unfortunately, I'm unfamiliar with MacOS. @papr will be able to help out when he returns from vacation next week 🙂

Hey, yeah in the end It came down to the python and Ubuntu version I used. In my case the only working combination was system python (3.6) on Ubuntu 18.04. Weirdly python 3.6 on Ubuntu 20.04 won’t work for me. Btw to test it more efficiently I used a docker container.

Hey all, I am trying to create a LSL using the pupil labs LSL relay python file but got an unexpected error "No module named 'plugin' Can anyone tell me any way around to fix this problem? or guide me to properly install Pupil labs LSL

Hi @nmt . I checked the default calibration routine used by the Pupil Player here (https://github.com/pupil-labs/pupil/blob/master/pupil_src/shared_modules/gaze_mapping/gazer_3d/calibrate_3d.py#L48). This calibration routine is "guessing" where the calibration targets are to minimize the reprojection error (i.e. "optimizing" the depth of the calibration targets). I noticed that the default calibration routine allows me to set the calibration target as fixed by setting the field fix_gaze_targets (https://github.com/pupil-labs/pupil/blob/74865d3ed5b6d2b462533abf09a9d485df4f551e/pupil_src/shared_modules/gaze_mapping/gazer_3d/bundle_adjustment.py#L23). Also, I found that there is a plugin for Intel Realsense camera (https://gist.github.com/romanroibu/c10634d150996b3c96be4cf90dd6fe29), so I bought an Intel Realsense camera to obtain ground-truth depth information of the world scene. However, that plugin was designed for Pupil Capture, and I need to do post-hoc gaze calibration because of our specific setup. Besides, I need to remotely stream the video to the Pupil Capture, so I built a customized video backend for Intel Realsense camera. I would like to change the post-hoc gaze calibration routine of the Pupil Player such that it can extract the depth information of the calibration points from the recorded world frame and perform gaze calibration with fix_gaze_targets set as True. Could you please where in the code should I look at to achieve this?

Hi, I think the world.mov file is causing the issue. Please rename it to something that does not start with world, delete the world_lookup.npy file, and try reopening the recording.

The ._-prefixed files can be ignored. These are generated by macOS if files are stored on external storage devices and contain meta information. Pupil Capture explicitly ignores those.

Thank you so much.

Hi, I've been trying to make an IR detecting camera by following: https://docs.pupil-labs.com/core/diy/#replace-ir-blocking-filter-on-the-eye-camera

I'm not using the recommended webcam, I've tried this with NexiGo N60 and AUSDOM AF640. Both have led to blurry images. I've tried using gloves while handling the film but that didn't really help. I'd really appreciate any help/advice you could give me on this - been stuck on it for a while now

Hi, I forward the question of one of our lab members:

"After annotating the data in Pupil Capture, there is a difference between the fixation numbers in the files I export. The annotations.csv file contains 121 more fixations than the fixations.csv file. What could be the reason for this difference?

Also, fixation durations show as 0 in the annotation.csv file. Is it possible to do post fixation duration detection in Pupil Player to see the fixation durations after the annotation?"

Thank you!

Hi,

Annotations created in Pupil Player are timestamped with the time of the current frame at which the annotation was created. Their duration is always set to zero. To model duration, you can annoate start and end events instead. Also, you can create multiple annotations per frame. There is no need for these to be equal to fixations.

Here are some steps you can take to improve pupil detection in dark environments: 1. Position the eye camera to minimise 'bright pupil' and/or glare 2. Manually change the exposure settings to optimise the contrast between the pupil and the iris: https://drive.google.com/file/d/1SPwxL8iGRPJe8BFDBfzWWtvzA8UdqM6E/view?usp=sharing 3. Set the ROI to only include the eye region, excluding the dark corners of the image. Note that it is important not to set it too small (watch to the end): https://drive.google.com/file/d/1NRozA9i0SDMe_uQdjC2jIr000iPjqqVH/view?usp=sharing 4. Modify the 2D detector settings: https://docs.pupil-labs.com/core/software/pupil-capture/#pupil-detector-2d-settings 5. Adjust gain, brightness, and contrast: In the eye window, click Video Source > Image Post Processing

Does the 2D detection affects the 3d pipeline since i am using the 3D model ?

Hi @nmt , I've been trying to make an IR detecting camera by following: https://docs.pupil-labs.com/core/diy/#replace-ir-blocking-filter-on-the-eye-camera

I'm not using the recommended webcam, I've tried this with NexiGo N60 and AUSDOM AF640. Both have led to blurry images. I've tried using gloves while handling the film but that didn't really help. I'd really appreciate any help/advice you could give me on this - been stuck on it for a while now

@&288503824266690561 is there information for the pupil core eyecamera manufacturer/ or a spec sheet for the camera and drivers that come with the base pupil model?

Hello, I am able to established the LSL using Pupils Labs LSL plugin. Now the problem is I wanted to send triggers to mark the different part of the gaze data.

Suppose, I have System1 which is running the Eye tracker and system 2 which connected using the LSL. Now from system2 I wanted to send triggers that will mark very specific part of the record time stamps. Can anyone tell me how I can established the connection?

Could you please share the full error trace back?

Hi! People from my lab tell me that if they record on a tablet using Pupil Capture and turn off the screen during recording (to safe battery), when they turn on the screen again, everything looks fine on first glance, but the recording actually stopped. (The video just shows a frozen screen.) Is this intended behavior? Can we do something about it?

Hi @nmt ! Good point! I'll have a look at the power settings (or ask the colleagues to do so 😉

Hello together, I recently installed the pupil core app and I added the camera due a USB 3 port. But the world camera does not work. Could you help me how i can start it. I get the following error message:

Restoring original ctypes.util.find_library... Original ctypes.util.find_library restored. attempt to release unclaimed interface 0 world - [INFO] video_capture.uvc_backend: 20:6 matches Pupil Cam1 ID2 but is already in use or blocked. world - [ERROR] video_capture.uvc_backend: Could not connect to device! No images will be supplied. world - [WARNING] camera_models: No camera intrinsics available for camera (disconnected) at resolution [1280, 720]! world - [WARNING] camera_models: Loading dummy intrinsics, which might decrease accuracy! world - [WARNING] camera_models: Consider selecting a different resolution, or running the Camera Instrinsics Estimation! world - [INFO] video_capture.uvc_backend: Found device. Pupil Cam1 ID2. attempt to release unclaimed interface 0

Thank you in advance

Hi @nmt ! It turned out that the colleagues tried to turn off the screen of the tablet using the power button. But actually, they sent the tablet to sleep mode.... (But what is quite unfortunate is the fact that the videos all seem to have the correct length. Only from the moment when the tablet goes to sleep mode, they just show a still image. Therefore, this was only noticed after finishing the study.... Maybe it would be possible to include a warning in future versions? Something like "Warning: Video recording has been paused for some time and might be incomplete". But I'm not sure how difficult it is to detect that the cameras have been "sleeping".)

@papr do you mind taking a look at the above question linked here? I am trying to figure out system specifications and am not seeing the information i need in the documentation. If you can provide the answer or point me in the right direction help is much appreciated. https://discord.com/channels/285728493612957698/285728493612957698/943219733539467265

Hi @nmt ! My current guess what happened is like that: It seems that older versions of Win10 had the option that the power button just turns off the screen. But newer versions have lost this setting and the power button sends the tablet into "partial" sleep mode. Now if Pupil Capture manages to keep running, but Windows shuts down communication with the cameras to safe battery, the software only "sees" the last frame in the buffer and happily keeps on capturing still images.

hey guys, i'm running an online experiment on pavlovia and would like to use eye tracking on that. is there any way to send triggers to pupil capture to separate trials and blocks?

thanks for replying! could you point me to some tutorial to learn how?

thank you!

Hi @nmt , Thanks for the suggestion. We might try this. I think it might also work to lock screen (as long as the tablet is configured to not go into sleep mode).

May I ask, are you using Pupil Core, already? And if not, how did you find this channel?

i have several tasks i use it for, one is run on psychopy/pavlovia, the other one is very different. I've already been asking a lot of questions on this channel, hope it's fine 🙂

no, but that's just great timing for me then 🙂

Hi which version of python is recommended to run pupil on? (for ubuntu 18.04(

i have done symlink on virtual environment running python3.9 on ubuntu 18.04, im unsure how to resolve this error, my cv2 version is 3.2.0

Hey nice to meet you! I am look for a way to get Tobii eyetracker 5 to work in Arch based system when I encountered pupil-eye-tracking. However it was unclear to me if it would work with Tobii?

Thank you :) I find eye tracking and VR quite intriguing and am currently using Tobii on a windows based system. However I would like to migrate to arch based linux since I use that basically in any other scenario. Which caused me to find your interesting project in my search for compatible drivers

I know that your version can be found in the AUR (Arch User Repository) which is where I encoutered it. My only concern is the compatibility with Tobii at this point :)

I was actually not aware of that. Could you share a link to it? here you go https://aur.archlinux.org/packages/pupil-eye-tracking-bin

By default, Pupil Core software is not compatible with Tobii hardware. And I am also not sure if their license agreements would even allow the usage of third-party software. also I didn't consider that aspect of it at all, would make sense if that's the case. I guess I'll stick to windows for now

now that I look at it you're right. However I have yet to encounter a package on AUR that didin't work

I only have the Tobii hardware as of now so I wouldn't know how to test it. However I can try compile it

Their Python API should be available on Arch if I am not mistaken https://pypi.org/project/tobii-research/

the latest post which I saw (November 2021) from Tobii development suggest that they don't plan to provide linux support as of yet, however they are on the lookout to see if linux gains more popularity

This is what I get when I run it on Arch

In that case the link I provided should be sufficient for Arch users :)

@papr /home/jetbot/.local/lib/python3.6/site-packages/cysignals/signals.cpython-36m-aarch64-linux-gnu.so is the error im getting as soon as i plug my world camera in (before crashing) do you know what's the issue?

Cysignals is only the module catching the segmentation fault. The error message reveals the actual cause.

Hi, this one looks like a race condition when calling uvc_stream_stop(#3) https://github.com/pupil-labs/libuvc/blob/master/src/stream.c#L1423 Not sure what is causing it

Hello I am trying to install pupil april tag module in my PC to use the Apriltags .Though I have Visual studio 2022 installed still I am getting this error . Kindly anyone let me know a proper way to install Apriltags

Hi folks, has anyone tried running capture on a 8gb raspberry pi 4 with 64-bit raspberry pi OS and the pi noir camera? If not, any thoughts / guesses on feesibility? Thanks!

Hi, I am trying to record with the glasses core using a mobile with the app Pupil Mobile, but I can't get the video, someone can help me?? Thank you

That is okay. The thing is that I can see more documents, but not the mp4 about the eyes and the other camera

Yes, and it put that they are active

And the video preview is displayed correctly, too? (tap the corresponding entry to open the preview)

Yes, the preview is okay

I see only 1 photo from each camera when I record

Okay, now it is working, thank you!

Yes

Can anyone let me know what is the format of the timestamp of the exported file of Pupil Player?

Hi, Player exports seconds in Pupil Time. See https://docs.pupil-labs.com/core/terminology/#timing for more info on Pupil Time.

Hello - we have suddenly encountered issues with the main camera getting blurrier - but the resolution/frame rate hasn't changed, and pupil cams still look okay. Resetting to default settings doesn't improve things. We have mostly been using a laptop version of pupil capture, but the change happened after testing with a mobile version. I can't really think why that would make a difference though... do you have any suggestions for anything else we could try? Many thanks!

Hi, the scene camera lens can be rotated to adjust the manual focus. It is likely that you rotated the lens by accident causing it to be out of focus. See here https://docs.pupil-labs.com/core/hardware/#focus-world-camera for details

It should not take a lot of force. Also, it is unlikely that you need to turn a lot. Small movements should be sufficient.

Hi @papr ! I'm having a bit of trouble with pyplr. The message "error:multiple repeat" keeps appearing with every single pupil player recording I try to load using this guide https://pyplr.github.io/cvd_pupillometry/04d_analysis.html#Export-with-Pupil-Player. Maybe I just haven't read enough (I discovered pyplr yesterday). In any case, thank you very much for your time!

Hi @user-f62f53 👋. This is a NumPy binary file. Trying to open it in a text editor will just get you gibberish. Instead, I'd recommend loading your recording in Pupil Player, our desktop software: https://docs.pupil-labs.com/core/software/pupil-player/#pupil-player. From there you can export your data into human readable files. Check out the documentation for instructions: https://docs.pupil-labs.com/core/software/pupil-player/#export

abs

Hi there, sorry in advance if this is the wrong channel to post this questions, but it seems to be the correct one from my understanding.

I am using Pupil Labs in Unity with HTC Vive headset with eyetracking where I am currently trying to access the ‘confidence’ variables. Specifically, I am trying to do so I can blink with one eye instead of having to blink with both eyes before it registers it. Any suggestions? 🙂

I can only access to the overall confidence value, but I'd like to get it for each eye seperatly 😅

Yes. We are accessing the confidence float from Gazedata.cs with the floatfromdictionary helper function. Is there a key for both confidence values or is there another solution?

Please see the software-dev if you encounter any new software-development related questions 🙂

Thank you, papr!

Hi is there a way to adjust the lsl plugin code so that it is able to send surface tracking gaze data as well? I noticed the gaze and pupil data are obtained from a “gaze” structure. Can we access the gaze positions on the surface through this “gaze” structure as well? If yes what’s the name of the child structure storing this information? Thank you!

Thank you very much! I hope he does! The program crashes at that point

Hey @papr ,

I wonder to know the model of the IR eye camera if it is possible. In the doc related with hardware I found only brief about world cam. It would be nice to know techical features of eye cam for my a manuscript.

Thanks a lot for your help till now.

Hi Is there an electronic version of the product certificate of pupil core? Thank you

Hi, can we use Pupil Core with Unity? I can see a Unity project for HMD. I am wondering if it will work with Pupil Core eye tracker?

Hi @user-4f103b We maintain a project called HMD eyes that contains the building blocks to integrate Pupil VR with Unity3d. You can use this in combination with Pupil Core open-source software. Check out the repository here: https://github.com/pupil-labs/hmd-eyes#hmd-eyes

This is wonderful, thank you. it will open many opportunities to develop some great applications.

ok thank you great help.

Hello, we're having an issue with Pupil Core where pupil detection doesn't work too well for people with long eyelashes. A quick search in this channel reveals that particular care needs to be taken when adjusting camera positions in order to avoid long eyelashes covering the pupil.

Interestingly, we're also observing that these cases coincide with autoexposure yielding severely overexposed eye videos. Is it possible that this is linked, i.e. is it recommended to set exposure manually when dealing with long eyelashes?

Hi @user-ab5f54 👋. For people with long/straight eyelashes, I have found that positioning the cameras more underneath the eyes, so they effectively look up into the eyes, can help. With the appropriate adjustments it's usually possible to achieve good pupil detection. We also recommend avoiding the use of heavy makeup/mascara. Camera exposure and eyelashes aren't connected. If you think the images are overexposed, try toggling the autoexposure off and on again, or set it manually.

Hello, what is the required IOS version to download the pupil core software? Right now I am working with Mac OS X 10.11.6. If thiis is too old of an OS, is there still an option to use Pupil Core apps with older systems?

Thank you!