Thank you, Papr, I will explore if it is doable and workable for me.

I understand that you do not support sound recording long ago (since 2.0), but do you have any recommendation on which system to use if I intend to use PupilCapture before 2.0 with sound recording?

In that case, we will need a pupillometry-only version that does not require calibration. I will be able to work on that within the next 2-3 weeks.

Hi all, apologies for the repeat question but is there any example of how to display screen markers on the client side (i'm not sure how to implement screen marker choreography on the dual_display_choreography_client)? Thank you

This is the default screen marker choreography https://github.com/pupil-labs/pupil/blob/master/pupil_src/shared_modules/calibration_choreography/screen_marker_plugin.py The original screen marker calib. will not know where in the scene camera coordinate system the markers will be displayed. Therefore, it needs to draw something that can be recognized in the scene video.

But keep in mind, with the dual-hmd calib you do not calibrate in the scene camera coordinate system but in those of the mirrored screens. Your client needs to send the target location in mirrored screen coordinates. As a result, you can use anything as a gaze target in your case! e.g. a cross.

hi, we have a pupil invisible model, and we want to check the pupil detection, but when we connect the eye tracker into computer, the computer didn't detected the eye tracker and there is nothing on the pupil capture when i opened it

do you know how can i fix it? thank you so much

Hi, guys, why use the program filter_gaze_on_surface.py to get the gaze position based on the surface coordinate system, each loop has more than one gaze position data point, can I only get one point per loop?

Hello, I have a question for you about wearing Pupil Core. They are expected to be worn by children and are between 8 and 14 years old. Is pupil recognition possible even for children?

Hi @user-6338f3 👋. Yes, our pupil detection algorithms work well for children in that age range and below (see my own research: https://pupil-labs.com/blog/research-digest/developmental-coordination/)

Hello, I'm trying to record EEG using BrainVision Recorder and pupil size data using pupil capture and it seems to me that LSL is the only option.

If I have BrainVision and Pupil Capture running on the same PC, is there an easy way to sync both data sources? e.g. sending a trigger

Thanks for the quick reply! Is it possible to do it the other way around? i.e. send pupil size data to the Brainvision recorder? My EEG data has a higher sampling rate than what is possible with pupil capture and I don't want to downsample it to the pupil capture sampling rate.

edited: alternatively maybe I could just send a brainvision trigger to pupilcapture via lsl?

The coordinates are the Pixel locations within the video, assuming that you are playing the video fullscreen

Right! I forgot about that. In this case, you can easily convert the pixel locations like this:

norm_pos_x = pixel_x / total_width
norm_pos_y = 1.0 - (pixel_y / total_height)

Hello, I want record Saccade Amplitude & Saccade Duration by Invisible or core, but I can not find these data in export file... @nmt @user-9429ba

Hi guys, quick question here: is there any page with information on the exact labels and meta info of the lsl streams obtained from pupil labs? I'm wondering about the coordinate system being used for the xyz gazepoint data, when i train a network to predict pupil position based on the normalized xzy per eye (6 inputs mapped onto az-el output), i get something very similar to the mathematical conversion from the gazepoint_xyz data except that it's flipped along the vertical axis. (so left is right and right is left).

Hi, see https://github.com/labstreaminglayer/App-PupilLabs/tree/master/pupil_capture#data-format and https://docs.pupil-labs.com/core/terminology/#coordinate-system for reference. The xdf file contains labels for each channel.

Hi @user-64c4d3! We don't offer a turnkey solution for classifying saccades. Although, with the new fixation detection algorithm available in Pupil Cloud for Invisible recordings, the inter-fixation intervals (gaps) can implicitly be regarded as saccades. See documentation here: https://docs.pupil-labs.com/invisible/explainers/data-streams/#fixations

Otherwise there would be two options: 1. Manual annotation of saccades that occur between fixations. 2. Implement a saccade filter using the raw data exposed

Does anyone have any idea how to plot norm_pos_x and norm_pos_y using Matlab?

Thank you for responding @papr This is the Gaze data that I got after extracting the. Xdf file i was wondering if you could tell me how to find the norm_pos_x and norm_pos_y in this in order for me to plot. Thank you kindly.

Hi Papr but I don't have the following channels you mentioned above

Yes, it contains 22 channel labels

Excuse me, could I get some help with setting up the pupil core glasses? I am getting that the automatic driver installation is not working and to consider manually installing the drivers.

Have you familiar with any visualization method to visualize the eye tracker data? Thank you.

okay, thank you!

also, is the installation the same for windows 11?

thank you again

My project is all about testing pupils' core eye tracker to make it work with LSL and the recorder data to matlab in order to extract the data, final part is the visualization, i have to visualize the gaze and fixation data.

Thank you for the information. I was able to get the plot, now I just wanted it to overlap with the recording that I recorded from pupil capture. so I can compare the movements.

simply a way to overlay the plot on top of the video.

Quick question: in the gaze_positions.csv export, what are the units of the gaze_timestamps?

hi, we have a pupil invisible model, and we want to check the pupil detection, but when we connect the eye tracker into computer, the computer didn't detected the eye tracker and there is nothing on the pupil capture when i opened it do you know how can i fix it? thank you so much

hello Can you please helo me fix this issue

Hello, I am consistently getting a pretty high angular accuracy number ( 2.379 degrees) during calibration. I have attached my calibration targets that the participant looks at through the mirror as well as the reference locations ive implemented into the dual display choreography client. Do you think its my visual target configuration? Should it span more of the display?

In Device Manager (System > Device Manager)View > Show Hidden DevicesExpand libUSBK Usb Devices, Cameras, and Imaging Devicescategories.For each Pupil Cam device (even hidden devices) click Uninstall and check the box agreeing to Delete the driver software for this device and press OKUnplug Pupil headset (if plugged in) and plug back in.Right click on pupil_capture.exe > Run as administrator. This should install drivers automatically.

I tried this but didnt work

https://docs.pupil-labs.com/core/software/pupil-capture/#windows

I recommend increasing the spread of the dots such that they fill the subject's field of view. Since you are using the 2d gaze estimation method, you might want to add more points, e.g. a 3x3 grid. I would also add an inner "center" dot for each gaze target to ease fixating the target for your subjects.

Hello, when i open my pupil player i dont see the fixation detector plugin. Any idea why ?

Could you please tell me what is the visual ange range for the pupil core, e.g.. the visual angle between the furthest left and the furthest right (or any two opposing directions) that the eye tracker can reliably measure pupil position? I apologies if this is reported online, I counldn't find it.

Is the pupil capture software not compatible with Mac OS Monterey? I’ve plugged in the headset and it doesn’t seem to register in the software. We have another laptop on Catalina that it works fine on though

Monterey requires admin privileges to access the cameras :-/ see the release notes where you downloaded the app for details :)

Sorry, I am reffering to only the range for 2d tracking by the eye camera and not gaze estimate. To explain, I am using circle_3d_normal_[x,y,z] to calulate the angle between the position of the eye at time point 1 and the position of the eye at time point 2. I would like to know the max limit that this angle could possibley be (i.e the largest measurable angle).

That makes sense and is kinda what I had been assuming but was wondering if there is a max angle that can be assumed under optimal conditons/camera position, or if it's truly a cases by cases bases?

Thanks Paplo. That alone is really helpful. Just knowing (or rather confirming) that the few values I have over 80deg are liklely problematic is useful. Intresting that you mention pupil size because I have found that can make a hugh difference for my data. I have had excellent tracking for some people while the light is on but as soon as I switch the light off (which have to do for my experiment) and pupil size increases the tracking becomes terrible. This happens a lot for people with contact lens.

I've actually found that the main cause is that when the pupil gets bigger the edges of pupil become much paler and harder to distinguish from the iris (even after playing with the min and max pupil intensity and pupil size max), hence the poorer tracking. This mostly only happens for people with contacts so I think it is to do with how the light reflects off/is transmitted through the contact lens as you get further from the center of the lens. So I think you are probably right about the lighting.

Hello, can I know the drawing numerical value of the exact product among the Pupil Core components? When I change the length of the component and try to recreate it, I need the exact blueprint value.

This is the section that I was referring to before:

Due to new technical limitations, Pupil Capture and Pupil Service need to be started with administrator privileges to get access to the video camera feeds. To do that, copy the applications into your /Applications folder and run the corresponding command from the terminal:

Pupil Capture: `sudo /Applications/Pupil\ Capture.app/Contents/MacOS/pupil_capture`

https://github.com/pupil-labs/pupil/releases/latest

hello! I have a quick confirmation about the pupil gaze coordinates. The coordinates is in reference to the size of the world data frame right? So if we have gaze coordinates that are in positive x and positive y values then, those gaze coordinates correspond to a pixel in the image?

Hello,

I am recording pupil diameter in a cognitive task that involves learning and decision-making, so ambient light and target distance are fixed and only pupil's psychosensory response is important for me. As stated by Sebastiaan Mathot in his very new preprint (https://www.biorxiv.org/content/10.1101/2022.02.23.481628v1, page 11, "trials should ideally be slow-paced"), such responses ideally need 2 to 3 seconds to show up on pupil diameter.

I'm concerned that this would make my experiment so long.

Based on your experience, do you think this 2 to 3 seconds is really necessary?

Hi again, aparently we still have the issue... IT asked me the following: "Please contact the vendor and have them send us the Anti-Virus exclusion list for their software so we can verify we are up to date." Could you provide me this info? Thanks!

Hi everyone, I am trying to load videos in pupil player but I couldnt. It was working fine until recently but now when i drag the folder to player the video doesnt work. It works fine on pupil cloud. Please help. Thank you.

Hi, I'm currently getting this issue when building Pupil Capture from source, is there any way to fix this?

Does anyone know if it is possible to have the world camera be used in 2 places at once, similar to this https://github.com/pupil-labs/pyuvc/blob/master/WINDOWS_USER.md but can be on in both pupil capture and as a camera device?

@papr Hi Papr, got i small question for you again. When I compute the azimuth and elevation (using the method you previously described: "r = sqrt(x^2 + y^2 + z^2) el = acosd(y/r) az = atan2d(z,x)" ) from the xyz coordinates i get rather large offsets roughly 80-100 degrees in both directions. Do you know where these offsets come from? (so points where the participants is looking "straight ahead" are then at (90,90ish) rather than (0,0)).

if you pass [0, 0, 1] through this equation, you should find out what the "straight-ahead" el/az should be. If this value is not desirable for your use case, you can simply rotate the coordinate system by subtracting fixed values from el/az

~~@papr Everything was working well for me, but things suddenly started returning an error. I haven't changed any of the code.~~

Error calling git: "Command '['git', 'describe', '--tags', '--long']' returned non-zero exit status 128." 
 output: "b"fatal: unsafe repository ('/home/ubuntu/Eye/pupil' is owned by someone else)\nTo add an exception for this directory, call:\n\n\tgit config --global --add safe.directory /home/ubuntu/Eye/pupil\n""
Traceback (most recent call last):
  File "main.py", line 36, in <module>
    app_version = get_version()
  File "/home/ubuntu/Eye/pupil/pupil_src/shared_modules/version_utils.py", line 81, in get_version
    version_string = pupil_version_string()
  File "/home/ubuntu/Eye/pupil/pupil_src/shared_modules/version_utils.py", line 57, in pupil_version_string
    raise ValueError("Version Error")
ValueError: Version Error

~~I checked earlier recommendations. I downloaded from github (git clone) and built from source~~ Nevermind. Running the recommended command fixed it. Leaving this up though because I think I'm not sure how this happened, and I'm interested in knowing if this can be prevented

When your recording in capture or watching a recording in player what data is used to make the dot indicator?

Yes, but I guess I should have asked which gaze_positional value? Gaze_point_3d? gaze_normal? How do I get the values that make up that dot? Sorry should have been more specific.

Hi, I am recording my data with the notification system in a csv, i would like to have the position of the center of the pupil (to compute eye gaze saccade etc... What data should I use from the pupil notification ? I see that there is "circle_3d center_x/y/z" data, I believe I should use that, yet I am wondering : is the value is pixels ? mm ?

Also as I have 2 pupil core device (different generation) I figured out that the resolution of the eye camera are not the same, will this impact how I should handle the center position ?

Thanks for your work and the quick help that you provide

You can see a description of pupil and gaze data made available here: https://docs.pupil-labs.com/core/software/pupil-player/#pupil-positions-csv circle_3d_center* is the centre of the pupil in 3D camera space (mm). For saccades, I'd recommend working with theta and phi (spherical coordinates) or their cartesian representation circle_3d_normal* Note that these values are enabled by the pye3d eye model, so make sure you fit the model well: https://docs.pupil-labs.com/core/best-practices/#pye3d-model. After that, resolution shouldn't affect how you handle the data

Hai, anyone can help me on how to resolve this problem? Left camera is not connected.

I have checked the small connector and now it’s connected to the software. However there’s a lag when recording.

(if the model 3d is not fitted I m having trouble to calibrate and I am receiving weird pupil size (23mm), I am for now forced to use the 2d diameter which is in pixel (not very good for my experiment)

(and as you can see the 2d model is working pretty well)

I am wonderring if is is not the camera position ?

@papr Hello, Do I have the option of calibrating the pupil core for a visual impairment subject like age-related macular degenerations? Due to the Scotoma, they are not able to fixate on the calibration points.

I am using the camera of the pupil core mounted on the Hololens with a prototype, I m not sure how I can tune the focal lenght ? and where can I see the log related to instrinsics ?

In that case, the intrinsics should be fine. You should find that information in the capture.log file

here you can see how it is mounted, I am recording and sending the image through network by the way

yeah of course I will send you a recording

What kind of intrinsics are we talking about ? What kind of parameter should I tune in it ?

Should I send you the code so you check ?

Yes so I should look forward in the backend code to send an appropriate matrix for this resolution ?

Cause I guess it is sending the wrong matrix right now

Okay and where can I found/compute the right matrix for the right camera ?

Oh my bad you send me in the link

Thank you so much for your quick answer, you guys are helping me so much 😁

That solves the issue, thank you SO much

Hey last question, I noticed that with these cameras (I have 2 pupil core device) I have sometimes a weird thing eye on the video (not only related to video streaming because I had the issue with USB connection), do you have an idea of what could cause that ?(see the video) :

https://www.youtube.com/watch?v=ApCg0XbLofY

50% of the time I will say, usually unplugging and replug the USB is fixing the issue but I have to do this 50% of the time 😅

Unrelated to above, but just a question about pupil measurements. For Eye 0 and Eye 1 we are getting different pupil measurements by about 1-2mm between the eyes for every frame. Is this a form of parallax error/ could it be due to the pupil cameras being at different distances from each eye? Would this affect the pupil diameter measurements? For example, if Eye 0's camera is 3 cm from one eye, and then Eye 1's camera if 6 cm from the eye, is this expected to produce different measurements between the two pupillary diameters, and would it be expected that Eye 1 reads a smaller diameter as a result of this because the camera is farther away? Thanks!!

Apologies I have just read the following: "Pupil size in pixels is dependent on the eye-camera to pupil distance and is not corrected for perspective. Pye3d, on the other hand, accounts for differences in eye-camera to pupil distances and corrects for perspective. It thus more accurately reflects pupil size and will be preferable for most users."

Hey Pupil Labs! Are there any updates on this or is it still not the plan to support the Vive Pro 2 in the future?

Hello. I am not sure if I got an answer to this question I asked a few days ago. Do I have the option of calibrating the pupil core for a visual impairment subject like age-related macular degenerations? Due to the Scotoma, they are not able to fixate on the calibration points.

Hi there, I have a question regarding saccades please. I am using Pupil Core to record eye movements and am interested in the saccade count (number of saccades) and saccade duration. I have read that I can consider that the time during two fixations is a saccade - but what about the blinks ? How can I take those into account / out of the equation to calculate my saccades ? Thanks in advance. Joanna

Hi all, I am writing to ask if anyone has had experience 3d printing their own headsets. This researcher has uploaded .STL files https://www.lukaszkedziora.com/en/2019/02/22/eye-tracker-elements-ready-for-workshops/ , but I am not sure what filament to use, and the settings, such as the % of infil, etc. Also, I would like to know if there are other homebrewed solutions available for people without the budget to purchase the real thing. Thank you in advance!

Hello, is it possible to remove the green dot that displays Pupilcore data information on the screen? This is the checked part in the photo.

Hi, yes. The dot is visualized by the default plugins Vis Circle and Vis Polyline. Turn them off to remove the visualization.

Hello, sorry to bother you with another question - when I export data from pupil player, I do not seem to have a "gaze.csv" file, I just have a "gaze.positions" file (as well as pupil positions, fixations and blink). The Raw data exporter and gaze positions exporter are activated. Is there something I am missing ? Thanks in advance for your help.

Oh, ok, thank you ! I am missing the "id fixation" and "id blink" columns in the gaze.position file, that's why I thought it was weird.. Is that normal ?

Thanks very much for your answer. Okay, so I need to code that manually then I imagine !

I will get coding on R then ! Thank you. Do you know if any programs are already available on github for example ?

Thanks again, it's very helpful.

Hi! I'm writing to you regarding advice on building my set-up. I am not sure that the way I created the connection is the most functional, so I expose as briefly as possible my set-up. My experiment is written in Matlab (in Ubuntu) is connected with pupil capture(in Win) with TCP protocol via ZMQ. Everything works perfectly, except for the timestamp that sometimes is accurate to the last millisecond and sometimes is not (LAN and in Win I have disabled any external process like anti-virus and other) [I attach a picture of the function, Matlab is integrated with Python]. The function has applied the variable mean-offset because I have previously created a function that sends request and receives response for 20 times between the two PCs to create an average of the delay -Could the UDP protocol be better to avoid delays? as I'm willing to lose packets due to the large number of trials/conditions. - Using Labstreaming recorder in Matlab could be a better method for synchronization? I have already installed the plugins on pupil capture, but I am not sure that using Matlab and having only eye tracking data is the right choice. In case it is, having already the plug in pupil capture, in the communication between Matlab/Lab recorder and pupil capture, Matlab would still need the ZMQ library to receive the annotation and time stamp respectively from the stimuli in Matlab and from pupil time? Sorry for the unclear questions. Thanks for your attention!

can we reform the pupil core to be wireless? e.g. any bluetooth accessories?

Hello colleagues! Our team has a question about testing new mobile app designs. How is image scrolling synchronized with the observation process? How are heat maps built? Are there any ready-made solutions from pupil team to handle such tasks? Thank you🤪👏

Hi 👋

There are no fully turn-key solutions that we can offer you for purely screen-based work. That being said, we do have marker based tracking algos built into Pupil Core software that would enable you to track the mobile device screen and get gaze relative to the screen. This would get you gaze relative to the screen and aggregate on gaze of multiple participants on screen (but not aggregate onto dynamic content).

If you want to aggregate gaze of multiple respondents/subjects/participants relative to dynamic content on screen, then you would need to build your own tooling to capture screen content and then aggregating onto that screen content.

Are you looking to do only eye tracking on mobile or do you have other use cases in mind? I ask because there might be alternative solutions to wearable eye tracking that might be more purpose-built to your research/application area.

@wrp Thanks for your solutions! For instance, we want to make humans do some navigation task to look for a specific object in a room. It will be not convenient for a person to hold a laptop to do the free navigation. Do you have any solutions for this scenario?

😀 great, so the tablet can be only with windows/ubuntu but not iOS, right?

btw, do you have any recommendations on tablets?

Thanks for the reply! We are interested in tracking the view on dynamic content that is user-driven.

Can you tell me if there is a semi-fabricated solution or a proponent of a screen capture solution using lsl, maybe? Any and all help would be greatly appreciated as we make the decision to purchase two core devices.

Howdy howdy all

Just want to make sure this product (Pupil Core) will work for what I want, and hopefully play nice with the others. I am going to do an academic study where I am going to introduce (what I believe) is a new stressor. Participants will come in, do a survey on a computer, do one of the manipulations (control, experimental condition 1, experimental condition 2). I want to do a manipulation check on the manipulations (prior research already shows ET/HR/GSR response), but I need to do it because one of them is new and I need to test it. Then the individuals will do another survey. I think some of the questions might actually trigger a stress response, so would like to keep check on that. I am looking for an eye tracking (ET) hardware that can do pupillometry (I think the Pupil Core can do that: https://pupil-labs.com/products/core/); additionally, at the same time (and timed together so I can match up ET/ pupil response to text and images on the screen), I would like to collect: galvanic skin response (GSR), photoplethysmography (PPG) to capture heart rate (HR) and oxygen saturation (SpO2)(this equipment: https://verisense.net/verisense-pulse). I'm assuming I will need some sort of software for this so I was wondering what would work, for example Shimmer software (https://shimmersensing.com/wearable-sensor-products/software/)?

I’m a novice, so I might not even be asking the right questions. With this work?

Hi folks. I have a real hard time with getting good confidence values for the pupil core eye-tracker. Is there any advice on how to set up the glasses in a way, that they capture the pupil even in difficult angles? I worked it out in a way, that I can achieve confidence values > 0.6 on my self. However, when participants use the glasses the values the set-up is way harder, takes a lot of time and values are not great at all. I even used the Youtube videos as a guide, but I cant seem to figure it out. Maybe I am missing one essential element when setting up the eye-tracker? Would appreciate any help!

oh and is there any way to convert the timestamps within the csv files in to relative time? So I can work with minutes and/or seconds instead of the absolute time values

Hi, would it be possible for you to share a hard case, i.e. a case where you think that you have positioned the eye cameras well but the confidence is not as high as expected? This would allow us to give more concrete feedback.

Hello, Could I learn FOV and infrared filter's nanometer of the eye camera? Have a good day!

So could you mention something about nanometer amount of infrared filter?

It's 850nm

Hi there. Can anybody think of a reason why blinks would not be detected on pupil player even though the eye0 video shows blinks (or rather... half-blinks, the participant half closes their eye each time) ? Fixations and pupil dilation are being detected and data is of good confidence. Thanks in advance for your help !

Hi, thanks for your answer. So, if I understand correctly, as the participant is not "traditionaly" blinking, pupil labs can't detect it ?

Not sure one can say it like that. I would rather say: Because the subject does not fully close their eyes, the confidence does not drop as much as the software would expect. By adjusting the thresholds, you should be able to compensate for that.

What @user-7b683e is referring to is that the pupil detector will always try to fit an ellipse instead of a polygon. If one had a polygon one might be able to estimate by how much the pupil is occluded by the eye lid. But out blink detector does not work like this.

Hi guys, here in our labs we would like to 'unmatch' the data if still possible. However, we can no longer find documentation on the pupil matching algorithm. Do you still have that somewhere online? Gr. Jesse

Well currently when we look at the data it isn't exactly clear to us for each idx from which frame it is being computed (and if this is consistent for each channel?)? Ideally we would like the eye tracker to give just data points for every sample it gets for the left and the right eye at the set sampling rate. But it seems you get twice the sampling rate in data points and there is some logic algorithm decidiing on which frame / img is being used to compute the parameters from right (usually alternating, but sometimes repeating and duplicating, or filling in nans)? What i would like in the end is 200 samples per second per eye to compute azimuth and elevation from the gaze_norm0/1_x/y/z with no duplications and a matching confidence per eye.

Ahhh! That's very handy. I believe base data is not available in LSL, however i think we can then use some xcorr to synch the offline basedata with our lsl data and then add another channel with the frame idx, right?

well i havent used anything other than lsl data yet, but i believe if we drag and drop the recordings in pupil player, we can then extract the data to some deliminated format and import that in our programming environment again, right? If we then find a channel that is both in lsl and the extract file; Synch the two together, We then can superimpose the index of the frame as a 23rd channel or so on our lsl data.

Yes, the native recording (the one you drop onto Pupil Player) has some overlap with the LSL recorded data. So your approach would work but would leave you with the too-high-sampling-rate that you are trying to fix.

Hi all, just had a (perhaps silly) couple of questions regarding connectivity between Pupil Capture and Pupil Core: is active usb connection required at all times during recording, or is there local memory on Core device which you upload via usb having finished recording? If the former is true, is there a wireless means of data streaming? Thanks.

Hi, without third-party software, Pupil Capture requires an active USB connection, yes. There are projects like https://github.com/Lifestohack/pupil-video-backend/ that can run on a raspberry PI connected to the headset and stream the video to a computer running Pupil Capture. But that has some limitations. If you want to be mobile I recommend having a look at our Invisible product: https://pupil-labs.com/products/invisible/

Hi again, sorry I now have a further two questions: what is the performance of the Invisible in terms of accuracy and precision and is the Pupil Capture software free with the Pupil Core or is it a supplementary purchase?

See https://arxiv.org/abs/2009.00508 re Invisible performance. Pupil Capture is free and open source. Please be aware that Pupil Invisible runs with a different software, the Pupil Invisible Companion app, which is also free.

What steps do I need to follow to do a post-hoc calibration in Pupil Player v.3.5.1?

Okay, but I don't have the option Post-hoc Calibration, only Gaze Data from Recording. How do I get to that option?

Yes that is right. So it is not possible for the Pupil Invisible?

@user-2bd5c9 what where you trying to achieve with the post-hoc calibration?

If you are trying to correct for a constant offset, you should have a look at this Pupil Player plugin https://gist.github.com/papr/d3ec18dd40899353bb52b506e3cfb433

I want to shift the calibration to the correct form, so it matches with the calibration we've done beforehand

Oh yes, correct for the offset is the purpose. Where should I put in this plugin? Do I have to download python?

Does anyone know how to combine the data of Psychopy and Pupil lab? I made it to the point where pupil lab is recording from the experiment in psychopy, but I still get the data with the gaze position separately from the timing of the stimuli.

there is an option in psychopy where you can define the recording start point in one of the routines

Hello)

Can I ask you for photos and dimensions of the shipping box for the core kit?

Amazing! do I need to put it with code? Im not sure what the process requires

Im sorry Im confused, the picture shows how to activate the pupil lab in the experiment, is the data tab should be there as well?

OK, and for the pupil lab exports - pupil_gazepositions, what is the time unit at pupil_timestamp ?

Hi, I recorded data using pupil invisible now, I am trying to analyze the data. I am not sure how should I start with data analysis, I came across a video explain to use pupil player in order to create the heatmap but when I am trying to create a surface it shows that no marker is found and I am not sure how define the markers.

I would appreciate if some one can help me

Hello everyone

I noticed that if the eye resolution is (192 x 193), the highest frame rate will be 200Hz. This way, it won't be easy to fine-tune pupil detection in the algorithm mode. The eye mode, which is the blue circle will be blue but not centrally placed (I guess not an issue) since the Pupil was well detected (red circle and dot). While an eye resolution of 400 x 400 will give max frame rate of 120Hz and easy to find tune pupil detection In algorithms mode.

I just want to be sure I am safe using either resolution. Please what do you think should influence my choice? We are currently using powerful computers.

Another question

2D and 3D gaze mapping pipelines

We have been using 3D pipeline and got good results if we don't have to worry about accuracy being more than 1.5 degrees.

Since we are using a chin rest, do you think using the 2D pipeline will still give the best results? No slippage compensation but low gaze accuracy

Hello,

Frequency amount can be set according to a work to be done. Some experiment requires high hz. If you don't need to 200 Hz, probably 400x400 can give best results. In addition to this, maybe you can use some filter to pass jumping data such as median, savitzky-golay or kolmogorov-zurbenko.

I can suggest for pipelines, 2D should give more accuracy by using a chin rest. Maybe you can change some settings such that distance between the headset and your surface.

Hi！We are gonna use the projecter screen to display stimuli to participants and collect their eye movement. Can we use the apriltag markers as well on the projecter screen? And do we need to ask participants not to wear mascara or eyeshadow? Thanks!

Hi there. I have a question : when I upload certain files in Pupil Player, this message pops up and Player crashes. Is that data not usable ? Thanks in advance for your help !

Thanks very much ! I am available on discord or via email [email removed] for any information !

Thanks! @papr 😀

Thanks very much, I will try that !

Hey @papr , I have tried the new plugin and Pupil Player crashed when I wanted to export the data (not before, which is progress) : this is what has appeared, I don't know if it's linked to the same issue ?

@papr

So what I have noticed is : with the original blink detector, only certain data files make Pupil Labs crash. However, with this new fixed blink detector, it makes every file crash

I have now switched back to the classic blink plugin but it is still crashing : "Offline_Blink_Detection" object has no "attribute timeline. I would really appreciate a rapid response for this issue please, as I have all my data to process very rapidly (PhD student with a deadline haha). Thanks in advance

Hello! I'm new here and new to Pupil Labs. We are using the Pupil Labs VR/AR Binocular Add On (but in a non-VR application). We have a stereoscopic calibration sequence. It seems to work fine, but once it's complete, it shows all the gaze data shifted to the upper left corner of the Pupil Capture screen. Almost like there's a fixed offset between the world center and what the software thinks is the center. Suggestions?

Hello! I need to ask our programmer to be sure, but I don't believe we are. We have a custom calibration sequence in Unity.

Thanks so much, it's fixed !

Sorry for spamming the forum ... @papr , you had shared a tutorial for fixation and blink detection (https://nbviewer.org/github/pupil-labs/pupil-tutorials/blob/apply-blinks-fixations/10_apply_fixation_and_blink_detection_to_gaze.ipynb) but the link now shows 404 Not Found. Has the tutorial been deleted ?

I use macOS but the eye cameras doesn't work. Where can I switch the camera?

world window has the same problem

Hi, I have a question regarding the accuracy of the Pupil core. Below is a screen in which the participant is looking at the screen. Red dot is the gaze point which is detected by the Pupil core. Suppose the the accuracy is 0.06⁰, does that mean the actual gaze point will be in the circule (dashed line) with a radius of 0.06⁰? I appreciate if you can help to clarify the question.

Hi! We're starting a project that requires doing pupillometry (we want to measure pupil diameter changes across conditions) with patients that cannot comply with instructions so calibration is not possible. We're not doing gaze tracking, just pupillometry, would this be doable under these conditions with the core?