Hello, my name is Adrián and I am working on a project with the Pupil glasses. I am developping a program using the Pupil Capture API to get samples and recordings from the eyes. I discovered that in order to use the functions of the Pupil API it is necessary to do a calibration with the Pupil Capture App. I would like to know if it would be possible to save that calibration in order to use the glasses several times without the need of calibrating each time.

Hello Miguel, thank you for your fast response. I understand you, but, it is necessary in order to use the API, to have installed and opened the Pupil Capture App?

Ok Miguel, thank you very much.

Hello, I'm designing a plugin for Pupil Capture to capture footage from five cameras. pyuvc seems to be able to help me capture video streams from unofficial cameras. However, after installing the camera's USB driver as libusbk, I can only use IPC to retrieve video from the camera on the Pupil core, and I cannot retrieve the unofficial camera video streams that are already using the libusbk driver via pyuvc. On the other hand, I can call the unofficial cameras normally in Capture. Does this indicate a problem with my backend? I've provided the connection method between the camera and the computer, as well as the libusbk device name shown in Device Manager. Many thanks!

Hello, I'm designing a system that uses Pupil Capture together with other sensors, and I’d like to know if it’s feasible to build a portable setup using the Pupil Core glasses.

Do the Pupil applications (Capture or Service) have a resource consumption low enough to make such a portable design practical, or is it too demanding?

Is there any documentation or benchmarks available about the resource usage of the apps and the glasses?

Thank you!

Hi there, I am building a Pupil Core DIY version, and would like to know is there a channel or forum for people sharing information and their experience on similar projects Please let me know. Many Thanks!!!

Hi @user-025238 👋 ! You can share your experiences here at 👁 core or if you want to flex your set-up /experiment you can also do in the 📸 show-and-tell channel.

Hi ,we are working on a project with the pupil core but we are stuck at calibration with screen marker is there anybody who can tell us any tips to go at the next stage which is data collection. Thank you beforehand!

Thank you very much ,we'll sent it right away.

Thank you very much ,we'll sent it right

Hi @user-f43a29, my name is Jason Chen. I am experiencing an issue with the right-hand side camera of the Pupil Core, which is unable to accurately detect the pupil. Please refer to the screen capture attached. Both the left and right cameras have the same settings, and despite my manual adjustments (indicated by the red circle with a point) to ensure the pupil is within the green circle, the confidence level still displays as 0.

I have tried resetting the Pupil Capture software and reconnecting the eye tracker, but I'm still encountering the same issue. I'm not sure if the problem lies with incorrect settings in the software or if there is an issue with the right-hand side camera sensor. Could you please help me identify the possible problems I may be missing and guide me on how to reset everything once again? Thank you for your assistance.

Some participants in previous experiments reported a slight electric shock sensation near their right ear. Could this be caused by an issue with the cable at the connection port?

The right connection port

Hello@user-f43a29，in the driving scene built in Unity, delineating dynamic areas of interest for pedestrians crossing the road, I previously attempted but still couldn't make the areas of interest follow the pedestrians. Do I need to attach markers around the computer monitor or on the pedestrians themselves? If markers need to be attached to the pedestrians, one issue is that when the vehicle is far from the pedestrians, the pedestrians appear very small, and the QR codes attached to them cannot be recognized. Can PupilPlayer alone achieve the function of analyzing dynamic areas of interest?

Hello! I'm curious whether the two eye tracking cameras in Pupil Core are the same model? If so, how do you achieve different camera label recognition in pPupil Capture & Service? I'm currently trying to run Pupil Capture with two identical eye-tracking cameras, but eye0 and eye1 often compete for the same camera's access. Many thanks!

Hi, it seems like the world view camera cannot be rotated (i.e., it is stuck), and thus, I cannot make it focused as needed. This may probably be due to a mechanical restriction issue. What is the recommended guideline for this kind of scenario?

Hi @user-d9be4a , while Pupil Player can track AprilTags as they move around, it will not be able to track them if they are too small. In other words, if you cannot see the AprilTags, then the algorithm will generally not be able to see them either.

Pupil Player was not explicitly designed with dynamic areas of interest in mind.

A solution in your case could be to put the AprilTags up on the corners of the screen. Then, you would have gaze in screen-mapped coordinates. You can combine that with a segmentation/tracking computer vision algorithm that could track the pedestrians and tell you when gaze is on the pedestrian.

If your system can provide you with the exact projected pixel coordinates of the pedestrian on the simulator screen, then you would not need a segmentation/tracking network and the process is easier.

Hi @user-b02f36 , even if two cameras are the same model, they should present themselves as two separate devices to the computer. Are these the custom cameras from your post here (https://discord.com/channels/285728493612957698/285728493612957698/1435544217689788546)?

Hi @user-f65018 , is it possible that the world camera lens is already fully tightened (i.e., "fully focused") in that direction?

I am looking to integrate PsychoPy with Core glasses and all the links for information are invalid or no longer found. The error pops up that the routine is looking for the neon glasses but I have no way of selecting core instead of neon. Are they no longer compatible?

Hi @user-bbffe9 , PsychoPy has been been re-designing their documentation.

The Pupil Core is still compatible with the latest PsychoPy and latest Pupil Labs plugin. Do you see a choice called just "Pupil Labs" in the Experiment Settings > Eyetracking tab?

Hi @user-4c21e5, thanks for your kind response. I have reset the Pupil Capture software. One critical problem is that when I look at the screen, the right-hand side sensor can not detect the pupil and displays a red point in the correction view. Please take a look at the video and see more details. The same actions on the left-hand side sensor can successfully detect and catch the pupil. Please help figure out the problem.

Thanks once again for your help.

Thanks for sending that. It looks like one of the 2d detector settings has been tuned incorrectly. Please restart the Capture software with 'default settings'. The button you need is in the main settings of the Capture window (cog icon)

Can you please make a test recording and share the full recording directory? Not just a screen capture. Thanks!

I see. Are you trying to use 4 Pupil Core eye cameras at once within Pupil Capture?

[email removed] (Pupil Labs), we attempted to "

I see. Please note that Pupil Capture was not designed with this use case in mind.

You will probably need to change the Pupil Capture source code to enable that. I would start with this section of the code for the detection logic.

I cannot say for sure, as I do not know your requirements, but you could also try making a Pupil Capture plugin.

Perhaps adding a new video backend, as was done for RealSense cameras, could be enough for your purposes.

Hi @user-1a14a8. I'll need the full recording to provide concrete feedback. Not just a screen capture. Full details below to avoid further ambiguity 😅 1. Ensure Pupil Capture is closed 2. Search on your machine, pupil_capture_settings. Once you've found that, rename the directory to pupil_capture_settings_backup 3. Open Pupil Capture and start a recording 4. Set up the headset as usual, adjusting the eye cameras and making sure the pupils are centred in the eye camera field of view 5. Slowly move the eyes around (e.g. by rolling the eyes) or look around 6. Stop the recording 7. Make a zipped copy of the entire recording folder, e.g. 001.zip and share that

Hi @user-b02f36! In principle, it's possible. But it's uncharted territory so you may just have to just try it!

Thanks for sharing the recording. For a majority of the recording, you are getting excellent pupil detection. As seen in the eye image overlay. If it helps, the debug view you're currently using to examine pupil detection is probably not so helpful. Instead, you should look at the default eye images - a robust red ellipse overlaying the pupil and a blue circle surrounding the modelled eyeball are the goals.

For the instances when eye0 confidence drops, I think this is because the image is over-exposed. You'll probably want to adjust the exposure level. - Change the eye camera exposure settings to optimise the contrast between the pupil and the other regions of the eye - It can also be helpful to set a Region of Interest 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)!

I hope this helps!

Hi @user-b02f36 , you are welcome. Although I have not tried, I think it should work. It is certainly worth a try!

Hi @user-bbffe9 , when you select "Pupil Labs", as you have done there, it searches for Pupil Core. Can you show what you mean by "it is looking for settings in the Neon"? Could you share the full output of the PsychoPy console when you encounter the issue? Thanks.

Hi @user-f43a29 its 2025.1.1. Even though I had downloaded the plugin, the trace back error was saying the module was not found. From reading more it sounds like the folder should be in the \AppData\Roaming\psychopy3\package subfolder so I copied the plugin package from github to that folder. I am now able to select Pupil Labs Core (iohub) from the experiment setttings. The module error no longer pops up but now the routine just starts but does nothing. no script print out. I have to force quit the program.

Which version exactly? 2025.1.1 or 2025.2.1 (Beta)?

Hi all, is anyone know how to proces "3d_eye_states.csv" to display like in the clould ?

Hi, @user-af95e6! Could you please elaborate more on the specific elements you're trying to display and the intended structure? This will help me give you more meaningful feedback.

Hi @user-af95e6! Thanks for following up. Your question, 'How you can calculate these value?', could refer to a few different stages in the data workflow. To ensure I give you the correct, helpful answer, could you please clarify which of the following you are asking about? 1. Data Availability: Are you reporting that these values are not present within the .csv or data file you downloaded? 2. Calculation Methodology: Are you asking about algorithmic details (formulas, definitions, and assumptions) we use to derive these values from the raw data? 3. Plotting/Display: Do you need help plotting/visualising the existing data in your analysis software (e.g., Python, MATLAB, Excel)?

Once you clarify which stage you need assistance with, we can send over the definitive details!

Hello there, I am reaching out to you regarding the Recalculate gaze distributions function and mandatory calibration step. Indeed, I am using the surface tracker plugin, but there is no Recalculate gaze distributions button in its parameters in Pupil Capture or Pupil Player ver. 3.5.1. In addition, the screen calibration marker in the bottom right corner of a TV monitor (or a laptop screen) can only be detected by turning the head on the right. Participants of my study sat at around 2 meters in front of the TV monitor. Could you please help me fix these issues? Looking forward to your enlightenment.

Hi @user-4d0d35 , in newer versions of Pupil Player, there is no Recalculate gaze distributions function for the heatmaps produced by the Surface Tracking plugin.

To better understand your second question, are you open to sharing the recording data with [email removed] Then, we can take a closer look.

Hi everyone, my lab is having issues with one of our eye-trackers. The world view is functioning fine; however, the eye view constantly disconnects and reconnects. Eventually, the eye view stays disconnected and does not come back. Sometimes when this happens, the eye panel closes, and we cannot reopen the panel (the "detect eye 0" button stays blue). We can swap to the eye 1 panel, but eventually that freezes and closes too. Stopping the recording after this happens tends to make us lose one or both views in the saved files. Starting a few days ago, sometimes the disconnection also results in the computer blue-screening, which also loses views.

We have another eye-tracker which works perfectly fine, no issues at all.

We are using Windows 11. Any help is much appreciated!

This is what appears in the .exe: "eye0 - [WARNING] video_capture.uvc_backend: Camera disconnected. Reconnecting..."

*The world camera also feels hot to touch?

We have tried with no success: - Rebooting the computer; - Unplugging and plugging the eye-tracker back in; - Closing and re-opening the eye panel; - Trying different computers; - Reinstalling Pupil Capture; - Uninstalling all drivers; - Resetting Pupil Capture to default settings; - Using different USB ports and cables; - Holding the eye-tracker and cable at different angles and heights. - Running Pupil Capture as administrator.

Hi @user-e1b2da 👋, Sorry to hear your eye cameras seem to be disconnecting. Thank you for sharing the debugging steps you've already tried. That helps a lot.

Please try the following: 1. Close Pupil Capture. 2. Disconnect the eye cameras. Each camera is attached with a JST connector. Disconnect these and check that all the pins are straight. 3. Reconnect the cameras. 4. Rename the "pupil_capture_settings" directory on your computer to "pupil_capture_settings_backup". 5. Restart Pupil Capture and connect the eye tracker. Does this make any difference? Let me know how it goes.

Hi @user-4c21e5 , could you help me with this ? Thanks

Yes in familiar with python. So there is no way i can plot or analyze the pupil diameter or eyelid apenture data ? I can only see them in the cloud ?

Thanks for testing. We would then need to receive the system back at our office for inspection/repair. Could you send an email to info@pupil-labs.com mentioning our conversation here? Someone from the operations team will look after you. It would be great if your could include your original order ID.

Hi everyone,

I am working on a research project and trying to receive real-time gaze data from Pupil Core using Python on Windows 11.

My Goal: I want to get gaze coordinates (x, y) in real-time in my Python script to provide feedback to the user.

The Problem: No matter what method I try (ZeroMQ or LSL), I cannot receive any data from Pupil Capture. My Python script connects successfully but hangs at the data receiving step (e.g., await socket.recv_multipart()).

What I found: I suspect the issue is on the Pupil Capture side. When I check the Plugin Manager in Pupil Capture (v3.5.1), I cannot find the standard plugins like: Gaze Mapper (2D) Frame Publisher LSL Outlet

Since Gaze Mapper seems to be missing, I assume no gaze data is being generated/broadcasted, which explains why my Python script receives nothing.

Steps Taken: Installed Pupil Capture v3.5.1 (Windows x64) from GitHub releases. Tried reinstalling it multiple times (both .exe and .zip versions). Deleted pupil_capture_settings folder in my User directory to reset settings, but the plugins still don't appear.

My Question: Is there a specific way to enable Gaze Mapper or LSL Outlet in Pupil Capture v3.5.1? Or should I be using a different version for real-time API access with Pupil Core?

Any advice would be greatly appreciated. Thank you!

I am unable to use a generic webcam alongside Pupil Capture, but Pupil Capture is blocking the webcam capture. Is there any way to configure this/exclude this webcam from Pupil Capture?

The source code clarifies quite some things, thank you. It seems that manually selecting a camera should solve the issue

Hello! I’m not sure if this is the right channel to ask this, but I’m trying to build a DIY version of the Pupil Core. I’ve been trying to get the glasses frame from the i.materialise store, but it looks like I can’t purchase it anymore. Is the option no longer available?

If so, would it be possible to buy the STEP/STL file for the frame, or is there any alternative you would recommend for building a compatible setup?

Thank you in advance for any guidance!

Hi @user-da7106 , the i.materialise webshop closed and is transitioning to a new framework.

We plan to have the DIY Pupil Core files up on Thangs within 2 weeks.

Hello,

Quick question, is Screen Marker Calibration Choreography only useful for scenarios where a user is looking at a screen or can it be generalized to more complex scenes? In more dynamic/complex scenes where we want to mostly look like gaze data, and not so much AOI. Is it better to use Single Marker Calibration Choreography or Natural Feature?

[email removed] 👋

Hi everyone,

I’m using Pupil Core to obtain gaze coordinates while capturing eye images with another eye camera.

In Pupil Core, the infrared illumination is always on for eye tracking. Is there any way to turn this on/off?

I’d prefer not to physically block the light or cut any wires — I’d like to handle this purely in software. However, I couldn’t find anything in the code that controls the IR.

Hi everyone,