The accuracy of gaze is bad in dark environments. Is there any tips on improvising the requirements? Note: I have calibrated it properly though.

The scene camera does not record at perfectly even intervals. Also, the head pose can only be estimated if the 3d model can be located with a scene frame, i.e. the markers need to be visible. During quick head movements, it is possible that marker detection fails due to motion blur. For these frames, no head pose can be estimated and the frame will be skipped.

Yeah sometimes it flickers. Any fixes?

Generally, yes, but in case of the Pupil Core headset, incrementing phi for one eye camera does not result in an decrement for the other, because the right eye is upside down.

@papr Hi, I am integrating the pupil into a dive mask and was looking to swap the recommended world view camera to an underwater fishing camera (either the spydro or the gofish cam) I wasn't sure who I could talk to to see if these would be compatible. But you look like the man to reach out to. Any help?

@papr the eye cameras will wire into a laptop above water and the world camera will feed to the same laptop.

For the camera to be recognized by the software it needs to fulfil specific details. See this message https://discord.com/channels/285728493612957698/285728493612957698/747343335135379498

@papr Hi, I tried to install the pyglui as mentioned in ubuntu 18.04. But I encountered the problem. The error information is displayed in the txt. Can you help me?

@paprYeah, I installed the Pupil Capture, and tried to run the application directly. But I get the another error.

I installed the Pupil Capture 1.0 before, but I have uninstalled this application. I searched the solutions, which seems to be related to the early packages. The version I tried to run is 3.3.

Sorry, the screenshot is from running 3.3.

@papr I have tried to uninstall all the pupil* application, and use dpkg to install the deb files. But I get the same error.

I installed this application in the Windows 10, and it works. But I found the 3D eye model seems to be abnormal. The 'No. of supporting pupil observations' always equals to zero. The mesh model doesn't show the 3D pupil circle.

Thanks for your reply. I first found the debug widnow has such problems in Windows 10. So I tried to run the application in Linux. You know, I found the software always has the different performance in different OS, which I consulted you several months ago. But I get these errors in Linux as mentioned before.

Does this debug window bug also include the abnormal 3D eye model?

I means the corneal mesh model doesn't overlay with the real region.

The negative pupil diameter is indeed unexpected given how well fit the model looks. Are you referring to that?

Reset view

Please be aware that the displayed pupil is not corrected for refraction visually which is why it looks different than in the recorded eye image.

I use the Pupil Lab device. But I don't the hardware version.

Can you get it?

Can I calculate the IR camera intrinsics and set it into the software?

Thank you. You know, it is difficult for the IR camera to capture the image in compute screen. So I need to print the chessboard. How can I update the camera intrinsic parameters into the application? I don't get the idea from the webpage.

I have just tried to reproduce the issue with a freshly setup Ubuntu 16.04 VM. I am not able to reproduce the issue. Looking at the responses here https://answers.opencv.org/question/18461/importerror-lib64libavcodecso54-symbol-avpriv_update_lls-version-libavutil_52-not-defined-in-file-libavutilso52-with-link-time-reference/ it looks like this issue appears due conflicts with other ffmpeg installations on your system. 😕

Hi, I have a question: How can trigger be sent into pupil capture? Would it be possible to write down trigger (as annotiations) coming from e.g. Neurobs Presentation or also Matlab Psychtoolbox?

Just built an open source tracker per website. Installed latest Pupil Labs desktop software on Windows 10. Getting these errors when trying to run Pupil Capture. For some reason the eye and world cameras are not being detected. Windows Camera can see/play them. Any thoughts/hints to resolve this would be most appreciated. Thanks. Mike Brown, Boston MA

One would think that by release 3.x the Windows 10 install would be nailed down. No wonder Tobii is #1 in market share.

If you are referring to the DIY eye tracker, please follow steps 1-7 from this link https://github.com/pupil-labs/pyuvc/blob/master/WINDOWS_USER.md

Hi everyone, sorry for the triviality of the question but this is my first time using pupil core. Eye 0 (right eye) connects without problems, Eye 1 (left eye) camera is disconnected. Do you know how I can solve this? If I choose manual camera selection I find only the camera of the Eye 0.

In this case, you need gaze data because the pupil data's phi/theta values are relative to their corresponding eye cameras. And since the eye cameras have different orientation in space, their horizontal/vertical axis are not comparable. Gaze uses the scene camera coordinate system.

If you do not see an "unknown" entry in the list, it means that a second eye camera is not connected. Please be aware that this is an expected behavior should you be using a monocular headset (one camera arm only on the right; vs one on each side).

eye centers are the location of the fitted eye models relative to the origin of the scene camera in mm

ok, then it means that the drivers were not correctly installed for this specific camera. Could you let me know the name of the successfully connected eye camera?

In order to calculate between-both-eyes-comparable horizontal/vertical eye movement, yes.

and if i use the surface tracker, then can the system still differentiate between the eyes?

Pupil Capture runs the driver installation on launch. Have you tried restarting the application yet to give the installation an other try?

ok, please try following these steps https://docs.pupil-labs.com/core/software/pupil-capture/#troubleshooting

They are normalized to have length 1. See this for the description of the coordinate system https://docs.opencv.org/2.4/modules/calib3d/doc/camera_calibration_and_3d_reconstruction.html

hi, can i know the recording rate of the eye tracker?

thank you!

followup question is this correct for all eye-pupil models?

I'm just not sure if these specs match my model. How do I know which model I'm looking at?

there are multiple pupil core models correct?

i understand thank you

Hi, does anyone have a CE certificate for the pupil core? Would you mind sending me a copy?

I see. Thanks!

Hi This happens when I use surface tracking. What is the reason?

OK thanks

I have an error "KeyError: 'gaze on surfaces'" when I try to run the filter_gaze_on_surface.py code in a PsychoPy experiment. Does this mean that the surfaces are not being properly sent to PsychoPy, and thus not recognized?

They are coordinates of where your gaze from that eye is, normalized within your surface bounds. I'm assuming this is for surface tracking?

Oh the 3d gaze is relative to the world camera

no, im trying to visualize the gaze from each eye. i know that their values are based on the world camera system but i want to make sure that the vectors they create are through the eye centers (basically that the vector of the gaze is originated from the eye and if this plot is correct)

Ohhhh that's definitely more of a papr question lol

thanks!

It works when I run the pupil script alone, the problem is when I integrate it into my PsychoPy experiment then..

If I was to have the Pupil Core connected to a desktop computer, what would be the longest length of a USB extension where I would still have a reliable connection? I'm conducting a study measuring visual fixations during gait (walking), and need as much flexibility/slack from a USB extension as possible. For instance, would a 25' extension still yield consistent data collection?

Hi, my lab just got a pupil eyetracker so I’m very new. I was wondering if there are guidelines to set the different thresholds for example blink onset/offset ?

Hey, im looking to upgrade the Pupil Core DIY to a wide angle camera that is compatible with the software. https://www.amazon.com/Arducam-Computer-Fisheye-Microphone-Windows/dp/B07ZS75KZR Would this be okay? (it is UVC compatible)

Hi, can I somehow set it up that I will see on a recording the gaze with only one red dot ? .. since yesterday it's just nowhere

now I have only two green circles..

where do I see if the world camera is in focus

ahj Im lost a bit

is it possible to send you some pictures?

@nmt so I thought I send you some pics so you can better understand what I mean

instead of a green circle or two green circle.. I need

such a red circle

Hmmm I gave it a try

let's say now I have two red dots without that line

but how can I combine them to only one dot so I would get

only this one circle

what else do I have to set up

hmm so.. I just had a one participant now and measured a bit in our study.. Now suddenly it looks ok

Hi, we are trying to use HoloLens 2 with the pupil labs hmd add on kit. Has anyone here designed 3d prints for mounting it on HoloLens 2?

Hi I encountered some problems, that is, when viewing the surface tracking data, I have some questions about the data in the red area in the figure. What time does world_timestamp refer to? Why do fixation_id and duration have multiple identical numbers? How do I choose from them, thank you.

Does online or offline pupil detection tend to be more accurate?

@nmt So how to choose these same values, are all valid or just choose one from the same value, for example, the surface I defined, such as multiple same fixation durations, then how should I use these data, thank you

Good morning! Is there any possibility to turn off 2D detection and only use 3D detection during recording? Glad for any help! 🙂

So would you still recommend "rolling with eyes" even if in this case only 2D detection would be possible? Id like to prevent some fails in Post-Hoc processing since I want to use 3D data.

How I can 'fit the [3D?] model' in PupilPlayer? And yes, that sounds good. But how can I turn off pye3d while running PupilCapture and having checked that the model is fine?

In Player, you just use the post-hoc pupil detection. This will rerun both detection algorithms on the recorded eye videos.

how can I turn off pye3d while running PupilCapture You either need a custom plugin or you send the command via the network api.

and having checked that the model is fine? To clarify, you would check first and then disable it, assuming the post-hoc detection will reproduce a similar eye model

Hi, I want to get pupil data for both eyes, at the moment I am using:

subscriber.subscribe('pupil.0.3d') subscriber.subscribe('pupil.1.3d')

and a while True block to capture the data, but in some iterations I am not getting both, is there a better way?

sure

@papr I'm still a little confused. For example, I defined a Book surface, and now I need the fixation duration data of this surface. How can I extract these data with the same ID, choose all or only one? Thank you

Hello, if you have multiple different video recordings with the same apriltag defined surfaces, how can the surface definitions be shared between the recordings? So far I have tried to copy the surface_definitions-v01 file between recordings, but it doesn't work that way. Thanks in advance!

Hi! I'm working with pupil capture video to feed it into a YOLOv5 model to detect objects within the frame as a subject is walking through space (example frame output in file named 'yolo_pupil_ouput.png'). To feed the post processed images in, I'll be using ffmpeg & mp4fpsmod to generate a new .mp4 file from extracted jpeg's with preserved relative timestamps [since the pupil capture video has a variable frame rate]. I was able to convert the raw pupil capture video to jpegs and back again into a video with the same relative timestamps, but when when I try to launch Pupil Player with the remade mp4, I receive a 'Found no maching pts! Something is wrong!' error (frameiterator_error.png). I've also attached the ffprobe output for each file (the original & remade files), can anyone comment on why I might be receiving this error?

update - I realized I didn't copy the metadata from the original file, attached I have the ffprobe output on the remade file after adding in the metadata from the original file. I still have the same error when trying to use this new file

Sharing the surface_definitions* files should work. 🤔 Please copy them while the recording is not opened in Player and only open the recordings afterward.

Does pupil core record pupil size?

Hello, I had a problem with the right camera of the Pupil Labs Core, so I changed the left camera to the right side. Since most people have the right eye as the dominant one. When testing the equipment, the model is quite lost, and the large red circle in the photo appears. What should I set to make it work that way?

The large red circle is part of the algorithm view. If you change the mode in the general settings it will go away

The eye model is independent of the calibration. Try looking into different directions to fit the model, like on the left. (Note: This screenshot is taken with our latest version. The colors might differ from your version.)

But first, you need to make sure that the 2d pupil detection is stable. From your screenshot, it looks like your eye lashes might be detected as false positives. Please use the ROI mode to adjust the region of interest to exclude your eye lashes.

Hi - I'm having trouble downloading the pupil core software on windows 10. I'm trying to run from the command prompt but get an error saying

'The installation package could not be opened'. Contact the application vendor to verify that this is a valid windows installer package'. Sorry I sent too early but any help would be great. Thanks!

When I download it, it has an .msi extension. Does this still need to be decompressed?

Chrome

Okay I now seem to be able to run the installer. Is it correct that the publisher should be unknown? Application: pupil_v3.4-0-g7019446_windows_x64.msi Publisher:
Unknown publisher

All sorted and installed! Thanks for your help.

I didn't think .msi extension needed to uncompress so it was just a case of unraring

Hello, I’m trying to figure out the time interval between fixations (by using the fixation plug in). I downloaded the data from the fixation plug in and may someone walk me through what does the “start_timestamp” and “start_frame_index” means and how would it be same or different from to the timeframe on the pupil player?

Hi after installing the latest release, I am getting a "refraction corrected 3d pupil detector not available" error. What could be the cause of this?

From bundle, macOS mojave

Hi, can i display Heat map for several respondents ? How can I do this?

Hello, We are using Pupil Labs on a computer that multiple researchers will log into. I installed Pupil lab under my user ID (I am an Admin) but the Pupil lab software does not show up under other user ID's, I have had to reinstall under each person user ID. Is there a way to have it install once and be used under each user ID?

hi! Can I get additional guidance on the gaze_timestamps value from exports? I grabbed the pts values for each frame using ffmpeg & mps4fpsmod independently (both methods as a sanity check) and I'm not getting equivalent values when looking at the gaze_timestamp for the beginning frame if using 0 based indexing pts from the world.mp4 using ffmpeg/mps4fpsmod at frame #3 is 916.966666666667 if using 1 based indexing pts from the world.mp4 using ffmpeg/mps4fpsmod at frame #3 is 957.977777777778 where gaze_timestamp at first instance where world_index = 3, pts = 1.747019805

Additionally, is there any advice that you could give for reconstructing the world.mp4 video from ffmpeg extracted jpeg's not aligning with the expectations that pupil player has about packet contents/pts for that same video? I extracted those frames using ffmpeg with vsync=0, setting pts time with mps4fpsmod with the original pts values, and setting the video metadata with the original file.

In all, I need to check whether gaze (with diameter=0) collided with any bounded box (grabbed from extracted frames being run through yolo) AND get the timestamps for the gaze data. Can anyone advise?

Hi, unfortunately, I do not know if this is possible. I will have to look into that next week. Can you confirm that you are using Windows?

Hello the support team. I have a pupil core and one of the eye's connection cable was broken. how can I fix it?

If the connection cable for the eye camera is broken, do I need to mail it back to the company to repair it?

hi, any chance in using a realsense camera as environment cam? Or is the pupil hardware required for the software to run?

thxn

i can't use it with capture 3.4, anyone knows the proper release version?

Hi,I have a question about pupil mobile app. I want use mobile app for researchers- in pupil capture I have "pupil detection color scheme" -then I can see if the pupil is correctly detected, but in mobile app there is no "pupil detection color scheme". If i want real results, what should I do to get my pupils correctly detected?

Thanks for your help

okey so, I understand, that "pupil intensity range" is only in pupil capture and in pupil mobile app this parameter isn't taken into account? There are only post-hoc pupil deteciton, Yes?

Okey, thank you very much

hello everyone, I hope you are doing great. I have a pupilab simple question. i want to extract the pupil size for my eyes recording. anyone knows how or have a link to a walkthrough. thanks you in advance

Please be aware that the quality of this data is highly dependent on the 2d and 3d pupil detection quality.

Hi, I had to reinstall pupillaps and the annotation list is removed. I already have annotated a few videos. Can I find the annotation list in the files?

yes I have reenabled the annotation plugin. The annotations are visible when you click trough the video (they pop up) but the annotation list is removed.

https://media.discordapp.net/attachments/849077710671708182/865758903827693608/20210716_205611.jpg

I tried out the new release. Please how is it ?

There were times I saw very very faint and disappearing orange lines around the blue circles... I don't know if is by eyes fooling me.

https://media.discordapp.net/attachments/849077710671708182/865764333287899136/20210716_211759.jpg

yes the corresponding hot keys

@user-7daa32 Please see the release notes ("Dynamic pye3d model confidence") in this regard https://github.com/pupil-labs/pupil/releases/tag/v3.4

If chinrest is being used, do you think 2d pipeline is the best to use ? I have issue creating surfaces when the precision is low. I encountered a situation where participant looking at A but the gaze will be in B. Creating Surface A would have to extend to B

We are manually taking data by watching the video on play because we need the transition time between AOIs. Which though we can't get the transition time between the last gaze in last fixation in one AOI and first gaze in the first fixation of the other AOI. We are currently thinking of using the AOI fixations counts

"Freezing the eye model is only recommended in controlled environments as this prevents the model from adapting to slippage"

What's the purpose of freezing ? I thought freezing is for the purpose of creating a surface. I am actually concerned about the gaze being precise enough to hit the AOI. We might to train the participants before carrying out the experiment.

Freezing the eye model is independent of freezing the scene video stream for the purpose of setting up a surface definition. It is something completely different.

You don't have to worry about it if you use the 2d pipeline. It only applies to the 3d eye model data (red and blue circles in your screenshots)

hi, would anyone know what exactly is in a .pldata file? I tried looking over the documentation but it's still unclear to me

https://docs.pupil-labs.com/developer/core/recording-format/#audio-file'

I also tried opening a .pldata file using the code linked at the bottom of the page (https://github.com/pupil-labs/pupil/blob/315188dcfba9bef02a5b1d9a3770929d7510ae2f/pupil_src/shared_modules/file_methods.py#L57-L87) but I kept recieving this error

thanks!

How would one go about rewrapping edited .pldata files?

Hi there. Do I understand correctly that the DIY unit only has one camera? (I.e. pupil detection for one eye)

Hi, i purchased the upgrade to pupil core DIY - fish-eye lens (DSL235D-650-F3.2) but after replacing the lens, the camera appears to be completely out of focus? I have tried adjusting the absolute focus, it does not seem to help.

Correct, the DIY frame is monocular.

Hi, I am running post hoc pupil detection for the first time, and, was wondering if I have to re run the 'Raw Data Exporter' to get the updated raw data files after post hoc detection

Also what are those token files in the "offline_data" folder,what is the purpose, how do I open them. Thanks

@papr Thanks

Hi guys, do you Knowles if it is possible to record pupil data without recording world camera ? I'm just insterested in pupil size variations and world movies are huge and not useful for me. Thanks !

Technically, this is possible. I just do not know if there is a user interface option for it. I made note for myself to check that. If it is not yet possible, I will add an option to our next release. Until then, you can delete the world* files afterward. Player will adapt automatically.

Hi, thanks for the help so far. We are not sure where we're going wrong with rewrapping the .pldata file, but when we try to reopen the folder containing the edited .pldata file, it doesn't run on pupil player. We're not sure if we're using the suggested class correctly, but we tried following the example rprovided and while it didn't produce any errors, it doesn't run

We're undergrad interns and have been asked to write a program that will allow users to crop the video shown on pupil player. Our approach was to crop each type of file individually and edit out timestamp data that was unimportant. So, we were attempting to open, edit, and resave the .pldata file without the timestamp data that was outside of the cropped video range. I think we're going wrong when it comes to rewriting the .pldata file after editing

Ah, so temporal crop, instead of spatial crop? Are you aware that you can temporally crop the export?

We were told you can manually crop videos, but we were asked to automate it, that's all we really know. And yes, we were able to edit the timestamps using that class, but when we reopened the .pldata file using the load function you mentioned earlier it threw an error

The former is much more difficult and error prone than the latter. The latter is actually very easy.

So sorry, we're not sure what you mean by the export

Our supervisor asked us to not use the exports; what is the best approach to trim the actual recording?

ok, in this case you are on the right path. Even though we do not give any guarantees regarding the intermediate format (in other words, it might change in the future; currently there are no such plans though).

Reading the pldata files and excluding samples outside of the trim range should work, too. You might want to adjust the info.player.json file as well in regards to the start_time and duration fields.

I suggest using https://github.com/pupil-labs/pyav for trimming the videos. Here you need to rewrite the timestamp files, too. Delete any _lookup.npy files that might exist already.

Feel free to share the error message in 💻 software-dev. I might be able to give further hints.

Btw, if this works well, this tool might be useful to others as well! Should you consider open-sourcing your solution, please add a link to the project over at https://github.com/pupil-labs/pupil-community

Hi! I am new to using pupil labs and was wondering how to relate the timestamps given in the csv file to clock time. This is what my time stamp data looks like, and I am not sure how to interpret it. Would really appreciate some help!

Hi! The clock starting point is arbitrary. On Windows, the clock start can be negative sometimes. (Unfortunately, I have not been able to find out the reason for this). These timestamp unit is seconds.

There is. Check out https://github.com/pupil-labs/pupil-tutorials/blob/master/08_post_hoc_time_sync.ipynb

Hi,

I have a question regarding the pupil_postions file generated after post-hoc processing.

1) I can see that pupil time stamps have changed. For example in the file generated from the recording the first pupil time stamp was '711705.765601999' but after post-hoc the first pupil time stamp changed to '711705.831237'. Shouldn't both be the same?

2)The 3d diameter generated by the recording was something like - '1.9...' or '2.4...' and it looked normal and in the expected range. However, after post-hoc I have certain rows, even with confidence level above 0.6, with 3d eye diameter as large as '174.75' or '55.7879' which look absurd. I am not sure if I did something wrong while doing the post-hoc or forgot to do something, or the data got corrupt.

Thanks for your help.

@paprHi, I have a question about the solution of device slippage. In the Pupil Core 3.4, the short term model of eyeball is built for adapting to slippage. However, as far as I know, in order to compute the right 3D gaze in the scene image, the rotation and translation matrix obtained in the calibration should also be updated in real time according the newest eyeball model. But I cannot find the code segment from the newest version of Pupil Core. Can you help me?

@papr Thanks for your reply. Which parameters have been calibrated in the calibration? I learned the code of Pupil 1.0 before. I think the matrices describe the relation between the sphere center <-> scene camera.

These are the only two parameters estimated for the binocular gaze mapper: https://github.com/pupil-labs/pupil/blob/master/pupil_src/shared_modules/gaze_mapping/gazer_3d/gazer_headset.py#L226-L227

@paprI think the person-specific kappa angle has been computed in the calibration. Do the two parameters include the information about the kappa angle?

the person-specific kappa angle has been computed in the calibration That is correct Do the two parameters include the information about the kappa angle? They do not explicitly, which is why the slippage compensation is not perfect and requires recalibration after some amount of time.

Kappa is encoded implicitly within the transformation matrices.

Did you mean that the transformation matrices are the eye_camera_to_world_matrix0 and eye_camera_to_world_matrix1?

So the eye_camera_to_world_matrix0 not only describes the relation between eye camera <-> scene camera, but also includes the kappa angle. Why does the get_eye_cam_pos_in_world state that the eye_cam_pose_in_world is the eye camera pose in world coordinates?

@papr Hi, I mainly don't know, how to solve the slippage problem only by updating the eyeball model in real time?

Thank you for the clarification. I will try to think of a way to answer your questions more clearly than I have above.

Hi! I just converted the time stamps to date/time but was wondering what the timezone is

I'm really finding it difficult to learn Psychopy. It seems most of what's done there involved programing and coding.

What's the best way to approach it?

Hi, I'm using Pupil Core for a project, and I need to access to angular velocity of the headset as well as angular velocity of eye of a participant. I think that I can reach to eye gyro data. However I don't sure that how can I collect angular velocity of the headset via Network API. What can I do?

In addition, I have a copy of Pupil Capture v1.7.

Hello! I am using pupil core and I am getting readings on Diameter_3D for pupil size that are negative values. Is this an issue that we can correct on our end or is this an issue with the tracker?

Other thing I have not understood is the data structure incame from Network API. In it there is a field named as "base_data". Sometimes I receive base_data field in a different base_data field, and each one have a norm_pos pair that is different than other identical keys. What is different among norm_pos data in gaze_on_srf and base_data fields?

Pupil Capture processes data by applying a series of transformations. These can include changes of coordinate system (while keeping the unit).

Each transformation generates a new datum, which references to its "ancestor" in base_data

This is the rough pipeline: 1. Capture eye image (eye camera coord system) 2. Pupil detection (eye camera coord system) 3. Gaze estimation (scene camera coord system) 4. Surface mapping (surface coord system)

Read more about the coordinate systems here: https://docs.pupil-labs.com/core/terminology/#coordinate-system

So in your case, norm_pos in gaze_on_srf is in surface coordinates, while the norm_pos of its base_data is in scene camera coordinates

@user-7b683e

{
  "gaze_on_srf": {
    "norm_pos":"bla",  # gaze in surface coordinates
    "base_data":{
      "norm_pos":"bla",  # gaze in scene coordinates
      "base_data":{
        "norm_pos":"bla"  # pupil data in eye camera coordinates
      }
    }
  }
}

Hi guys, I've trouble with the sampling rate of the pupil camera when I record data using Pupil Capture. I fixed the sampling rate of the pupil camera (directly in pupil capture) to 120Hz and after data collection I noticed that the mean sampling rate for my recording was around 247Hz. If I fix the sampling rate to 200Hz it does not change anything. Moreover, the difference (in second) between the time points (t i + 1 - t i) are absolutely not constant, it oscillate between ~0.006 and ~0.014. Is it normal ?

Hi papr, No, I am looking at the diameter data in the pupil_position. csv file.

Yes, I already separate left and right eyes and 2d/3 data but it does not solve the problem, I still have varying difference between time points, while these difference should be constant if the sampling rate was correct (right ?)

0.007961 0.008041 0.007913 0.007999 0.008021 0.007962 0.008381 0.012261 0.003437 0.007916 for example, these are the 10 first differences between my time points

So ti+1 - ti

If you need evenly spaced samples, I suggest linear interpolation. In fact, I recommend looking at Kret, M. E., & Sjak-Shie, E. E. (2019). Preprocessing pupil size data: Guidelines and code. Behavior Research Methods, 51(3), 1336–1342. https://doi.org/10.3758/s13428-018-1075-y

Ok ! Thanks for the answer an the reference, I will have a look ! 🙂 I have another issue, this time using LSL. I am recording synchronised EEG and pupillometer data using a plateform that use LSL to synchronise the devices. When I look at the data I have even bigger differences between time points (until 12 seconds sometimes !). And it is not rare that I have negative difference that suggest that the [ti +1] point is before the [ti] point. Do you know what it means ?

In this case the samping rate is closer than 250Hz than 200Hz. Even of the sampling rate is varying, is it normal to have such a big difference ?

1.1

It is the 2.1 sorry

2.1 forwards gaze, i.e. this is subject to the mixed streams mentioned above.

And I'm using Pupil Capture v3.1.16

Larger time gaps indicate a transmission issue.

Ok, I understand. Thanks for your help papr 🙂

Hi,I have a question about raw data-can i read the time to first fixation and the time of looking at a surface?

hello, is the model number of the eye camera used in pupil core available for disclosure? I wonder where do you source cameras in such small form-factor while having really high refresh rate

Hi, I am trying to use the surface tracker, I have 3 screens that are in the same plane but I have an additional screen that the participant will hold. It says in the doc that all surfaces have to be in the same plane. Is there any way to have surfaces defined in more than one 2d plane ? If this guideline is not followed will the results be worthless ?

oh yeah sorry my bad, I misread that and panicked for a second. Thanks for the prompt reply.

I have downloaded Capture v3.6 and received data via Network API. However, I couldn't see circle_3d->normal field you mentioned. In gaze_on_surfaces, there is only one base_data field. For example, "base_data":["gaze.3d.0", "12647.33"] was written on the field.

So, how can I receive circle_3d data on Capture v3.6 such as v1.7.

Opsh, yes 3.4.0. I'm sorry this day was so long

What can I do to be able to receiving angle of eye and head both? Is there any version these two data can be received?

1. You have said me that you can use head_pose sub key to receive headset rotation info. For this reason, I would like to get head_pose data.

2. I don't need to real-time degree data, but these must be sync.

Okey, I learnt that I can use recording file for this case. However, I want to ask my question more clearly again.

I want to receive two data - eye degree and head degree to be done on x-axis. By using these data, I will calculate velocity.

My questions are; 1) Can I get degree data (of eye and head) on Network API? 2) If I can do this, with which version of Pupil Capture?

You need to subscribe to pupil and head_pose and process these two streams.

I would like to thank you. In base of versions, data structure was changed drastically. I have got data I want to now.

Hello everyone! As I am analyzing the pupil core data for my dissertation, I have a question related to fixation. Do we have a term defining the time elapsed between the first and the next fixation? I am attempting to call it "Fixation-changing time". Please let me know if anyone knows. I will continue to look it up on the documents from pupil labs and published papers. If I found anything, I will follow up here. Thanks!

Kia ora from New Zealand. I'm currently tasked with extracting saccades data from pupil labs tracking data. I've attempted to follow through the preprocessing work by @user-2be752 (https://github.com/teresa-canasbajo/bdd-driveratt). I'm a bit out of my depth and was hoping I could get a few pointers. I managed to get the 'make' file to complete, however when trying to run the python script I'm getting lots of errors. I was hoping someone might have some fairly detailed instructions that I could follow. Any help would be hugely appreciated. - Mat

Hello. Thank you for your kind supports. I am trying to use data of fixation.csv to estimate subjects' visual angles. However, sometimes variance of values of gaze_point_3d_x,y,z is extremely high. What can be expected as the cause of this error?

hello, what is the best cell phone to use with the pupil core

?

I'm using one but with a lot of problems when we download the file to generate the results

Hey papr,

After you have said to me about getting info on pupil and head pose, I have looked documentation of these fields. However I haven't found some details I would like to see on the doc.

Firstly, in pupil datum structure, I want to eye rotation amount as degree. But its unit has been declared as mm - I hope it is millimeters.

Second stuff, I haven't saw a guide on head_pose sub key. Like eye rotation, I want to get rotation amount of headset in degree unit.

I was hoping that you could mention some things in below; 1) Can I reach degree values to calculate angular velocities? If it is yes, which field can I receive? Apparently, circle_3d->normal isn't suit value due to its unit - so I need to degree. 2) If it is not possible, how can I calculate degree amounts by using existing fields in currently datum structure - so Capture 3.4.0. 3) Or is there any different method you can mention to me to calculate velocity amount of eye and headset in deg/sec unit.

Sorry, I don't think that the doc in this subject is enough for my purposes. By this reason, I have conveyed you many question.

Thanks.