Hi! I wonder if it is possible to get the pixel size of eye camera in pupil core. I have got the intrinsic matrix of eye camera and I want to get the pixel size to calculate the real focal length. Many thanks

Hi everyone, one of my student upgraded her pc, she has a mac book (MacBook air 13, chip M1. 2020. 8 GB macOS Sequoia version 15.0.1), and capture doesn't work anymore. That is, the message appears that the cameras are detected but they are actually disconnected, even though among the usb ports appear to be connected--how can this be fixed?

Hi @user-a4aa71 ! Could she try the following troubleshooting steps? If those don’t resolve the issue, could you please open a 🛟 troubleshooting ticket with additional details, such as the version of Pupil Capture being used?

Hi I apologize if this question has already been asked, but I couldn't find it when I searched.

When collecting data with Pupil Core, I take care to detect the pupil, make the necessary adjustments, ask the subject not to make sudden movements with their head, and perform a correct calibration, with good levels of confidence. When analyzing the data in the pupil player, I noticed that, in some cases, the participant's fixations were not detected (unfortunately, I didn't notice this at the time of recording). Is this normal for some participants? I noticed that the number of fixations was very low and it is possible to see that fixations were not detected during the video (pupil player). Could you explain what might have happened or something I can do? Thank you very much!

Hello, Please in my 13 minute study (see attached screenshot), I had 2 calibration points and 2 validation points (So calibration then validation, then calibration, then validation at the end). The part of my data I really care about is between the 2nd calibration and validation (at the end of data) points. If I leave Pupil and Gaze data set to "from recording", how does the exported csv manage these calibrations? Does it simply use the most recent calibration to calculate gaze data until when a new calibration arises and then it uses that one moving forward, or does it somehow mesh and use both calibrations for the entire data span, or does it calculate gaze some other way?

Additionally, I am working with a surface. Please how would this surface gaze data interact with the 2 calibration sessions?

Finally, is there an ideal configuration for extracting post-hoc fixations optimally? or should I just stick to the default settings for the fixation detector plugin in pupil player?

Also, I noticed that although it is two calibrations and two validations in my data, for some reason, it posits that there are 3 calibrations (which is not the case) (please ignore the "Pre-trial calibration", as I manually created that). Please can someone help me understand why this would happen?

Can I use ESP32 camera model for capture eye? @user-d407c1

Is it possible to track the eye movements using a video? By inputing it

Is there provided guideline?

https://discord.com/channels/285728493612957698/285728493612957698/747343335135379498 in this message it says "For other cameras, you could theretically write your own video backend, but that might be very time consuming." is there anybody who done it ??

any help please?

@user-f43a29 @nmt Please could someone have a look at my messages above https://discord.com/channels/285728493612957698/285728493612957698/1304594934053212162

Hi Is there a way to remotely tell pupil capture to add a surface in the surface plugin, as with sending a request to calibrate through pupil remote?

We are using the pupil labs core for real time eyetracking to use an interface. We use the surface tracker plugin to get the gaze within the surface which is our interface. But as we are running it through docker, pupil capture starts anew, so without any extra plugins initiated. So i was just curious as to see if there was a way to initalize a plugin and add surfaces without having to setup the program

so by requesting it through pupil remote, so the setup would be a part of starting up our program

Firstly, the capture world settings are stored in the pupil_capture_settings directory on desktop machines. Is there a way to make this persistent in your setup?

yeah we should be able to copy an existing settings file, and place it into the program files on startup

With the right settings, the surface tracker plugin should load automatically.

Also note that defined surfaces are saved in a file called surface_definitions in the pupil_capture_settings directory.

If you restart Capture and the Surface Tracker plugin loads, your surface definitions from previous sessions will be loaded.

I would give that a shot and see if you can make it work

we've found the files, we'll see if we can make that work, thank you!

Hello all, a question about how the "diameter_3d" is calculated using diameters from eye 1 and eye 0: if one eye has an outlier reading or missing data, how is it handled in the calculation?

Hello and thanks again for your previous help! As I mentioned, we are planning to conduct research sessions in dyads (parent and child) and we were wondering if you could recommend an optimal screen size for these studies. Given that we’ll have both adults and kids interacting, any guidance on screen dimensions that would work well for both would be really appreciated. Thanks so much in advance! Best, Asia

hi Team, i would like to know if there is a way to compute the heading and pitch of the gaze without the head pose information? I have exported the gaze data but there was no head pose information being recorded. Thanks

Hi @user-98789c , we do not reject said data. We provide you with all estimates, as computed by the underlying pye3d model.

Since outlier definition can be dependent on research context, we leave that decision to the user, rather than reduce the data you get. This way, you get as much flexibility and data as possible.

Could you clarify what you mean by "missing data" in this context?

Hi Rob, yes I was using Pupil Core and i did not have any other devices that recorded the orientation of the wearer's head and I could not use the Head Pose Tracker as I did not apply any markers as they were not easily detectable in a dark environment.

Okay, hopefully i can get other tips as we have the gaze data which recorded the world camera as well.

I'm sure this question gets asked all the time, but is the core compatible with contact lenses? I know the neon is. After searching on your website and on this discord chat, I couldn't find the answer for the core specifically. Thanks!

Hi @user-0f7e53 , I now think that I may have misunderstood your original question. I had the impression that you wanted gaze in a world-based coordinate system. For example, a system where, if you look towards magnetic North, gaze azimuth would be 0 degrees, whereas when looking South, it would be -180 degrees.

However, this latest message suggests rather that you want to know the angular deviation from neutral gaze, regardless of head pose? Do you want, in a sense, data related to the rotational orientation of the eyes within the socket?

Hi @user-0f7e53 , so, the gaze ray provided by Pupil Core originates at the center of the world camera, which is offset and can be essentially arbitrarily oriented. The gaze ray is represented in the world camera's 3D space, so it indicates the direction from that camera to the point that a wearer is looking at. This is what gaze_point_3d_x/y/z intend to specify

The transformation, as done in that tutorial, is fine for determining gaze velocity, as an offset and arbitrary rotation do not change such velocities. However, that transformation alone does not provide a "gaze heading" such that an azimuth and elevation of 0 always corresponds to looking straight ahead (i.e., it does not account for the arbitrary rotation of the world camera)

Rather, you can try this:

• If you had a measure of neutral gaze (i.e., "looking straight ahead"), then you can mark that as elevation = azimuth = 0
• Then, you apply the azimuth & elevation transformation, as provided in the tutorial, and correct for the offset from said neutral point

Also, please note that the tutorial uses "traditional" spherical coordinates, where:

• 90° azimuth (i.e., theta) and 90° elevation (i.e., psi) corresponds to "forward" in the world camera coordinate system (not the same as "gaze straight ahead")
• 0° elevation corresponds to "upwards" in this same system

An easy validation is to make a recording where you first look left, then up, then right, then down, and you can better understand the relationship to the spherical coordinates.

Also, did you apply the negative_z_values correction?

With respect to the calculations in your earlier post, gaze_point_3d_x/y/z and eye_center0/1_3d_x/y/z are not compatible in that way, so they cannot be directly combined like that to derive an azimuth and elevation for gaze

Please let me know if this cleared things up

I just got a secondhand pair of Pupil Core glasses and am planning to use them for basketball performance analysis, focusing mainly on free throws. I’d also like to explore their use for more dynamic movements. Can these glasses can operate without being continuously connected to a laptop, or if there are alternative setups for more freedom of movement?

I’m planning to do coding in Python on a MacBook Pro with an M1 chip. Could you recommend any useful websites for API references? Also, could you explain any precautions I should take when using an M1-powered computer?

I am preparing for development by following the instructions on https://github.com/pupil-labs/pupil/blob/master/README.md#installing-dependencies-and-code. However, I am having trouble downloading the Pupil Labs libraries listed in the requirements.txt file on a MacBook with an M1 chip. The error "error: subprocess-exited-with-error" occurs.

Hi @user-f43a29 , thank you for your response. I haven't decided on the specific program yet, but eventually, I plan to conduct psychological experiments using Python and PsychoPy. For now, I aim to perform operations such as calibration and recording using programs created in Python.

At this stage, I have created a virtual environment with Python version 3.11, as recommended, and executed the following commands in the terminal:

git clone https://github.com/pupil-labs/pupil.git cd pupil git checkout develop

However, when I ran:

python -m pip install -r requirements.txt

I encountered errors while downloading the following libraries:

// Pupil-Labs ndsi==1.4. pupil-apriltags==1.0. pupil-detectors>=2.0.2rc2 pupil-labs-uvc pye3d>=0.3.2 pyglui>=1.31.1b1

It seems that these libraries are causing issues during installation.

Since I'm using an M1 MacBook, I was able to launch Pupil Capture programmatically by running the following code:

import subprocess

//Command written as a list command = ["sudo", "/Applications/Pupil Capture.app/Contents/MacOS/pupil_capture"] // Execute the command subprocess.run(command)

⚠️ // is a comment line in code.

Hi Folks - haven't fired up the core in a while. I was thinking about an in-class demo, but found that Capture can't find my cams on Sonoma. I don't see Capture listed in "settings / privacy&security/cameras".

Is it simply no longer functional on mac?

Hi Rob. Thanks for your response. What I meant was: Since there are two sets of calibration and validation, the first occurring at the beginning of the experiment and the second occurring mid way through, I wanted to know if the first calibration is used for the first half of the recording and then from the point where the second calibration occurs, does the first calibration get discarded and the second calibration is then used for all recording that occurs after it?

Can someone please clarify if this camera will works for eye capture feature?

eye0_hardcoded_translation = 20, 15, -20 # ？ eye1_hardcoded_translation = -40, 15, -20 # ？ ref_depth_hardcoded = 500 This is a fixed value for the eye position in the code. If I want to modify the relative position relationship, for example, the world camera moves up some, how to do it. I can't currently see exactly what these numbers mean in the world coordinate system because they don't particularly agree with what I'm measuring.

The gaze data from the eye camera is slifhtly misaligned. Is it possible to shift it slightly on the x-axis and y-axis by offline calibration after recording? Is it possible to shift by manual correction?

Hi everyone, I am a human factors engineering student. I'm currently designing an experiment with a python programming software that requires subjects to perform some mouse clicks on a computer. I also want to capture data from their eyes and be able to synchronize it with my mouse data. (For example, the eye data will start to be captured only after the subject tries the start button on the screen). I have both invisible and core devices. I don't know which one is easier to program ( since my English and programming skills below average). It would be nice if there is a documentation or project about it! Thanks guys! I really need your help!🥹 👍 👀

Hi @user-74c615 👋 ! You can programmatically control Pupil Capture and send event annotations using the Network API. While the method for detecting and capturing mouse events is up to you, once captured, you can send event annotations through the Network API to mark these events. See below some resources that can be of interest for you:

If you're planning to implement a gaze-contingent paradigm, the Alpha Lab guide on building gaze-contingent assistive applications might be helpful. Additionally, Pupil Core offers integration with PsychoPy through the PsychoPy plugin for Pupil Core, which can be useful if you're developing your experiment to control stimulus presentation.

Finally one more note, while it's possible to start recording at a specific point, we recommend starting the recording before performing the calibration. Check out our Best Practices for more recommendations on data collection.

Hello I am trying to figure out the DIY case for starting a high school project. The link to the shapeways store is broken. Are there any other cameras that are tried and tested for the DIY kit? Thanks

Hi, I am running the Pupillabs Core source(https://github.com/pupil-labs/pupil) in a Python 3.7 environment, but I get the following error:

Traceback (most recent call last): File "/home/~~~~/PycharmProjects/pupil_with_deepvog/pupil_src/main.py", line 39, in <module> from version_utils import get_version File "<fstring>", line 1 (get_tag_commit()=) ^ SyntaxError: invalid syntax

How can I resolve this? I want to run it in a Python 3.7 environment.

This happens when you download the source code from github's website, rather than using the git clone command to grab it. Just so you know though, we provide pre-made builds of the Core software, and most users will not need to run from source.

Hello everyone. I tried the Network API today and it worked, thanks to your clear documentation! 🥰 But I still have 2 questions. 🥹 Can I use the pupil remote command to change the name of a subfolder? I only see the use of 'R rec_name' to change the name of the main folder. But I would like to have a lot of recordings in one main folder and be able to name the subfolders myself, instead of using names like 000 ,001. The second question is, if I use my computer screen for calibration, does the bottom left corner of my screen correspond to (0, 0) in the world camera? What I'm trying to accomplish is that I want my eye position to be normalized based on screen size. Or do I have to define my screen as a surface? I found this out because I noticed that the coordinates inside 'gaze_postions.csv' (e.g. gaze_point_3d_x) would have negative values. I dont know what dose negative values mean🧐 Thanks again for your work and help!🙌 🫂

Hi @user-74c615! 1. If memory serves, that first command should generate a custom name for the recording folder. In a sense, each recording is made in a new folder. 2. If you want screen-mapped gaze, you'll need to use the Surface Tracker Plugin in Pupil Capture. You'll need to set up a surface that outlines your screen. The mapped coordinates are described in the documentation. Briefly: x_norm and y_norm are coordinates between 0 and 1, where (0,0) is the bottom left corner of the surface and (1,1) is the top right corner.

Bonus: You can use the real-time API to get real-time surface-mapped gaze. This example script shows you how!

Hi all, is there a data sample available somewhere? https://pupil-labs.com/blog/demo-workspace-walkthrough-part1 leads to a 404.

In particular, I was looking for a monocular recording, ideally at 50hz with a non-moving head (e.g., chin rest)... a few minutes would do. This would be enormously helpful!