Events

Hi guys,

I am having some issues with the 'reference image mapper' function. Every time I try to create an enrichment, it doesn't 'run' properly. In the past, prior to the change in interface I was able to get this function to work however now it seems to have been altered? Just for further information: I have a reference image which is clear + a scanning video of me holding the glasses while going over the scene for 40 seconds from multiple angles. As I said, this was enough in the past now it keeps failing. I've tried multiple different pictures + scanning videos but still it either doesn't 'run' or 'runs' and then has an error.

Reference Image Mapper testing

Hi everyone, I didn't think we could record blinking with Pupil Invisible but when looking at the data downloaded from the pupil cloud I found a blink.csv. Is blinking now recorded by Pupil invisible?

Hey, I have a recording on my phone with a duration of 0 sec. When I click on it, the recording actually plays for 6.22 min. Is there a way for me to properly extract this data from the phone and analyze it in Pupil player?

Hey, second question. I want to use the Pupil player but I miss some important files. I export the data from the phone but I miss the timestamps and video's. How can I fix this?

Recording duration inconsistent

Hi, I was wondering, is there any way to download enrichments using the cloud API v2? I am able to download recordings, listing projects etc, but no luck for enrichments. Andit would be supercool to create enrichments with the API. Is there any hidden feature not documented here (https://api.cloud.pupil-labs.com/v2/), or any plans to allow that in future versions? Or am I missing something completely?

Hi @user-d4e38a ! I reply to your message https://discord.com/channels/285728493612957698/285728493612957698/1115829174603628616 here, as it seems like you are using Pupil Invisible.

This sounds an issue with the cable but to be sure, would you mind contacting us at info@pupil-labs.com so we can follow up with further debugging steps?

Hey guys, what kind of text environment / scan environment disparities have you been able to get away with for reference image mapping? Let's say I want to scan the inside of a subway car cabin; the banner ads across different subway cabins will differ, but the core geometry of the inside of the cabin will remain the same. The specific features I'm trying to measure are consistent across cabins (directions, security information, etc), but I'm concerned that the disparities in the environments will not be accepted by Pupil Cloud when I try to upload them.

Hello, I am a new user of pupil invisible. I wonder if there is a way to directly extract gaze positions from raw data, without using pupil cloud or pupil player?

Thanks, Dom! I have exported the data from the smartphone to my PC. I get a lot of files like gaze_ps1.raw or gaze_right_ps1.time. Is there a way to extract these type of data to the same human-readable format like gaze_positions.csv?

Thanks a lot guys, that was really helpful!

Hi, I'm not sure if the question has been asked before (it probably ha, but I couldn't find it). But I've made a recording by pupil invisible (and hopefully soon Neon 🙂 for which I would like to read out gaze data and gyroscope data using Python without pulling it through pupil player since it is so time consuming to export it (at least the writing of the movie). Is there any way to do so?

Thanks a lot. But this is on Neon isn't it? Does it work the same for invisible? In my recording I can't seem to find the json and csv files (only after pupilplayer)

Ideally I would not like to use the cloud for privacy reasons (patients) until my organisation has made a decision

I mainly see npy and pldata

When you open a recording in Pupil Player, it edits some of the files in the recording folder. npy nad pldata files are all generated by Pupil Player.

If you download a recording off of the phone, you shoud see files like e.g. gaze ps1.raw which you could read using the Python code.

The json and CSV files are only available in the download from Pupil Cloud.

During recording through pupil invisible glasses we found error and recording was saved later, I have uploaded the recording file on google cloud but I am unable to open/can't play the recording . How can I solve this problem? Please help me

Sorry for the mistake actually it is Pupil Cloud not google cloud. Ok recording ID is 7e172289-33c1-4510-9576-69b690a956c1 and aa4bbea1-2b18-42c9-b330-1207d4bff64d .

Hi @user-e40297! The .raw files are our intermediate binary format. They're not really meant to be consumed directly, but you should be able to parse them with the code snippet my colleague, @user-cdcab0, shared earlier

Hi Neil, Thanks a lot. Is there a reason I shouldn't use it?

I would be perfectly happy using the csv fies. Only exporting and creating the movie requires a lot of time

Hi @ Marc, do you intend to make more than 1 AOI visible at the same time in Pupil Cloud (as it is the case in Pupil Player)? I think this is the most important thing and a great disadvantage of PupilCloud, because when I set one AOI (in the middle) and I want to define two more to the left and the right it is impossible to know where the boundaries of the first AOI are. In this matter Pupil Player is much better, but no Fixations are being calculated and one can research only each subject's gaze separately (I can't put all participants at once).

Hey folks, I can't seem to generate any heatmaps using the reference image mapper - I am quite sure I have followed the recommended steps to a tee. Help!

Processing videos in the cloud takes an unusually long time, several hours even for very short data. What can I do?

Hey @user-df1f44 👋 ! Could you please clarify if the Reference Image Mapper enrichment has been completed successfully? Can you see the gaze mapping from the recording onto your reference image?

Thanks for the clarification @user-df1f44! Usually, it takes a while until the enrichment is completed. It's also possible that the enrichment has already been completed, but the browser hasn't reflected that. In this case, try performing a hard refresh (command + shift + R).

However, if the enrichment cannot be completed, this might indicate that the mapping failed. I'd recommend checking our scanning best practices https://docs.pupil-labs.com/enrichments/reference-image-mapper/#scanning-best-practices

Hi @user-7c714e! Thanks for the feedback! We have gotten similar requests from other users too and are in fact currently looking into this! I don't have a precise ETA yet, but we'll have an update to address this soon.

Video Data is still processing, can you help?

I am also having this issue. How long should processing normally take?

With the new Pupil Cloud formats do you have to carry out again the enrichments you have already applied to uploaded video or are they instantly transferred into the new format?

@user-162faa @user-cca81b We have found and resolved the cause of the issue with recording processing. Computation for processing should now finally actually start and the accumulated queue of recordings should be finished within the next 1-2 hours. So check back soon!

Hi, me and my resaerch team was doing a test with Pupil Invisible and once we did the recordings, it was giving an error and did not allow to create a project. Do you know what can be the problem?

Hi @user-6127f0 and @user-cca81b! Indeed seems like the issue from yesterday is not entirely resolved. Apologies again for the inconvenience. We are again actively working on it!

Thank you!

@user-6127f0 @user-cca81b We have localised the issue that led to the error during recording processing and fixed it. We have triggered all affected recordings to be processed again. Its quite a few recordings, so it will take a couple hours until all is done, but then they should finally all be available and future recordings should go through with no more issues!

hi there, I am new here and have a quick question. When we watch our recording on the cloud, the gaze is clearly visible, yet when we download the recording, the gaze disappears. Could you please help me out with how to download, keeping gaze in? it is very crucial for us to be able to do this. thanks

Hi there, are there any suggestions for post-hoc gaze calibration for pupil invisible? In our study, we've calibrated the glasses prior to each recording. The participants are about 1.5m apart, and as we start the recording, we ask participants to describe where they are looking at. However, there still seems to be an offset and we are wondering what are some ways to account for this. Thanks!

Hi Pupil Labs Team, I have 2 questions regarding the pupil cloud data download. Maybe you can help me out here: 1. After the Marker Mapper Enrichement, I downloaded the data and saw a file called "surface_positions.csv". Am I correct assuming that this is the data you described as "aoi_positions.csv" in you documentation (https://docs.pupil-labs.com/export-formats/enrichment-data/marker-mapper/#aoi-positions-csv) and that only the name is different? 2. I wanted to download the raw data export, but after clicking "download" the api closes (see screenshot). My project is quite large with 567 recordings of about 4 min each. Could this be the problem? Do you know a workaround for that?

Thanks in advance for your help! 🙂

Hi! I am starting using the new Pupil Cloud version and I am wondering how can I compute my analysis? I have created two Reference Image Mapper enrichment and run them . Now, in one of them I see a comment saying that the enrochment is ready to be computed, but how can I do that? And the second enrichment resulted in an error. Why could that happened? Thanks for any help you could provide 😉

I was hoping to get the roll-array out of the result extimu ps1.raw file. I found the x, y, z gyroscope and accelaration input-data. Is there "an easy way" to calculate this roll variable without using pupil player Or is there a way you can tell pupil player to spit out the csv files without making a new video in the output?

Hi @user-e40297 ! I assume that you don't want to use the Cloud? That's the easiest way to directly obtain roll, but if you want to programmatically obtain it please have a look at https://github.com/pupil-labs/pupil/blob/c309547099099bdff9fac829dc97b546e270d1b6/pupil_src/shared_modules/imu_timeline.py#L91

Hi Pupil Labs I am trying to undistort the world video using the Cloud, which works fine, but I don't seem to have accesses to the new gaze coordinates. Is there a way of getting those from the cloud?

Thank you for your management. Could you please answer the following questions? The following are Pupil invisible and neon questions.

Q1. In the csv named blinks, there are start_time and end_time. What are the definitions of these? For example, which of the following is start_time?　(1) the frame at the moment the eyelid closes (i.e., the blink is recognized from the eyelid border), or (2) the frame at the moment the pupil begins to be hidden by the eyelid.

Q2 Which of the following is the blink recognition method?　(1) Recognizing the blink from the eyelid border. (2) Recognizing the blink from whether the pupil is hidden or not.

Q3 If I want a timestamp or index of the turning point of a blink (the point at which the eye is completely closed), by what means can I obtain it? The blink turnaround point is the point where the eye is completely closed after the start of the blink, as shown in the right figure in the attached image.

Hi there! I managed to create a heatmap using the Reference image mapper. Now I want to distinguish between two different groups in the same project and create a heatmap with them. for example, I want to create a heatmap of the recordings called 'used before' and a separate heatmap of the recordings called 'never used'. I have already tried to do this by adding a filter but then it still seems to create a heatmap of all the images in the project. How can I create two different heatmaps in the same project? Can you help me out? Thanks!

Hi @user-5b371f ! There is currently no way to obtain the corrected gaze positions in Cloud. I have gone ahead created a feedback entry in our canny board to request this feature https://feedback.pupil-labs.com/pupil-cloud/p/provide-undistorted-gaze-positions

Feel free to upvote it. In the meantime, here you have a tutorial on how to undistort the video and get the new gaze data. https://docs.pupil-labs.com/invisible/how-tos/advanced-analysis/intrinsics/

Hi Jella👋 ! The heatmap generated in one project includes the data from all recordings present in this project. I’d recommend you split your recordings into two projects, e.g., “used before” and “never used”, assign the relevant recordings in each one, and apply a Reference Image Mapper enrichment in each project to generate a "used before" and "never used" heatmap. I hope this helps!

Hi @user-5c56d0 ! Here you have the description of all fields in the blinks.csv file that you can obtain from Neon and Pupil Invisible https://docs.pupil-labs.com/export-formats/recording-data/neon/#blinks-csv

Unlike with Core, Neon and Invisible employ a different blink detector operating directly on the eye video to detect the movement patterns of blinks, you can read more about this here: https://docs.google.com/document/d/1JLBhC7fmBr6BR59IT3cWgYyqiaM8HLpFxv5KImrN-qE/export?format=pdf

Hi! our android device for the invisible device broke down. We want to still do some (informal) testing (i.e. not proper datacollection) while we are figuring out how to replace it. I saw that the companion app only works with compatiible oneplus, which we dont have. Is there an option to use it with a non-recommended device for now?

Hi guys! Hope you're well. I am just wondering if it's possible on pupil cloud... I recorded some data on a different workspace, is it possible to move those recordings do another workspace... I am struggling to see if I can upload files rather than transfer them onto the workspace via the cloud. Thanks.

In the Pupil Invisible Companion App, when setting a manual exposure what units are used for this value? I am assuming milliseconds but can't find a definitive answer in the docs

Hi @user-cc819b ! Exposure Time (Absolute) Control is used to specify the length of exposure. This value is expressed in 100µs units, where 1 is 1/10,000th of a second, and 1,000 would be 0.1 second.

Note that high exposure values can reduce the frame rate of the scene camera. In auto exposure the frame rate is locked at 30 FPS but this is not the case for manual exposure

I'm having some problems with the Reference Image Mapper on Pupil Cloud. It's the first time I'm trying to add the RIM enrichment on the new cloud, but it doesn't seem to work. When I click "Run", I get an "internal error message" but when I run it again it says "Enrichment Running". I don't see a progress bar of the enrichment processing anywhere (as it was the case on the old pupil cloud). How do I know if the processing has actually started and is running? I started a few RIM enrichment yesterday, but I don't see the gaze mapping on the reference image, so I'm not sure if its working or not...

Hey @user-eebae9 👋 ! Thanks for bringing this to our attention. We addressed the issue and it should work fine now. Could you try creating/running the enrichment again? It usually takes a few minutes until the enrichment is completed but if you don't see any updates after a while, it's also possible that the enrichment is done, but the browser hasn't reflected that. In this case, try performing a hard refresh (command + shift + R). Hope this helps! Please let us know if you keep having issues with that

In one of our recordings, participants are asked to vocalize what they're looking at. What's confusing is the first time they look at something, the gaze tracking is quite far off. After looking at various other objects, they eventually come back to the same object but the gaze tracking is very good. The participant appears to be in the same position, and the position of the object hasn't really changed relative to the participant. Do know what could be happening there?

Hey @user-e66c8c - I'm replying to your message (https://discord.com/channels/285728493612957698/285728493612957698/1121410590754226246) here. To get the data in Pupil Player format, please go to the Workspace Settings section and then click on the toggle next to "Show Raw Sensor Data". There you can enable this option in the download menu. Please keep in mind that you need to enable this feature for each workspace you'd like to download the raw data in the pupil player format.

@user-480f4c can you tell me where is Workspace Settings ? I can't find them

ok, I've found it

@user-480f4c thanks for your help, I've managed to resolve the problem

Hey, quick question. Is there anyway to recover a recording that's been deleted from the Pupil Invisible app before it could be uploaded to the cloud.

Additionally, are there any plans for adding a confirmation button before deleting a single recording?

Hey, I have question. I accidentally deleted a recording on Pupil Cloud. I checked "show trashed" to restore the deleted recordings. However, I get "Table filter url is invalid","Go back to default filter" and could not find the erased recording. Can I access the recycle bin and restore deleted recordings?

Hi @marc , today I can't login in pupil cloud. Is there something wrong with the server?

Hey Anne ExoInsights 👋 Thanks for

Hey @user-480f4c I just checked and was able to view the deleted recording. I was also able to restore the recording. Thank you very much for your support!

Hi there, in some of our recordings, participants are asked to vocalize what they're looking at. What's confusing is the first time they look at something, the gaze tracking is quite far off. After looking at various other objects, they eventually come back to the same object but the gaze tracking is very good. The participant appears to be in the same position, and the position of the object hasn't really changed relative to the participant. Do you know what could be happening there? Thanks

@user-3437df this sounds a bit unusual. Would you be able to share a recording of that?

the charles this sounds a bit unusual

Hi, could you please tell me ASAP how long would be the gaze with world index 433? Is the difference between gaze timestamp in ms?

Or is it in [s]?

Hi Marc, I have a similar question. I currently have 2 OnePlus 6 Invisibles, but only one cable works. maria has suggested a new cable for me, but my next question is whether there is an upgrade path for handsets. it looks like I have to remain on Android 8.1 with my current handsets... (in the chat, Android 9 is not ok).... or is it possible to directly install Android 11? Alternatively, what are admissable new handsets to buy? (I can no longer use my work wifi because they only allow Android 10 or 11.) Thanks

Hi Nadia, thanks for that. I always tell my students, never upgrade anything (ever) 🙂 So my best option would be an 8T, I will see if the budget can stretch. Thank you.

Hi@marc,I would like to know if Invisible's real-time API can get some advanced information such as blink index and fixation index.

Hi, I'm experiencing difficulties by interpreting the .time files. I'm using the code provided by you guys earlier def read_timestamps(path: str) -> np.ndarray: """Read gaze timestamps from a binary file.gaze psX.time""" return np.fromfile(str(path), dtype="<u8")

And are interested in the world-cam time values (PI world v1 ps1.time?) gaze time data (gaze_right ps1.time?) and gyroscope times (extimu ps1.time?) But the time vales seem to vary from the values from the csv files...

Am I using the correct files using the correct snippit?

No I"m looking at the files stored on the phone

Hi @user-a98526 & @user-d5d71f ! Blinks and fixations are not immediately available via the real-time API. However, we plan to open-source the code for our blink and fixation detectors in the near future, including potential real-time versions that could be used.

Please note that our blink detector requires eye images, which are only available in real-time for Neon. Therefore, using our algorithm, real-time blink detection with Pupil Invisible is not possible.

On the other hand, our fixation detector uses the gaze signal and scene video, both of which are available in real time via the API for Pupil Invisible. This means that real-time fixation detection with Pupil Invisible will be possible once released.

Hi, I tried to install the real-time api by writing "RUN pip install pupil-labs-realtime-api" in my docker file. But it gets the error "could not find a version that satisfies the requirement pupil-labs-realtime-api (from versions: none)", "no matching distribution found for pupil-labs-realtime-api", can someone help me out? Thanks!

Hi, I want to try something with Pupil lab invisible. The idea is that the wearer would be seated in front of a scene and someone on the scene would move. How can I get categories of what the wearer is looking at considering tat the person on stage would be at different coordinates along time?

That's amazing @user-480f4c , thanks! Is it possible to also capture other object of interest that would be moving? Like a lamp a dancer would take with them?

Hi, I have a question regarding gaze/fixation overlay in exported video from pupil cloud. I intend to annotate world camera video in the Elan annotation tool. We've previously done this with recordings from older Pupil Labs glasses by simply exporting video with gaze overlay using the Pupil Player. However, I can't get recordings made with Invisible to work with the Player app, and I find no option to export video with overlay directly from Pupil Cloud. How do you suggest I approach this?

Hi @user-9894cd 👋 I wanted to ask if you've been exporting recordings from the Companion device or Pupil Cloud for uploading to Pupil Player. In the second case, in the latest UI version of Pupil Cloud you'll need to enable the download of the Pupil Player compatible format in your workspace settings by toggling the "Show Raw Sensor Data" option. By the way, you can also generate and download gaze overlay videos using the Gaze Overlay enrichment. For more information, please refer to our documentation at this link: https://docs.pupil-labs.com/enrichments/gaze-overlay/#gaze-overlay

The download button at the bottom of the screen was not the most visible, but it works 😉

Many thanx for the help!

I'll take the oportunity to also ask something else.

I've used the real time API in Python and it works very well, but for testing it would be convenient to also be able to playback recorded data. Can I do that through the API?

Specifically, I want to grab each fram and gaze data from a recording and use it in a similar was as I get frames and gaze through the api.

I realize it's not that difficult to just open the video using OpenCV and parse the gaze-csv-files manually, but a more uniform interface allowing me to write code that could interface with either the live stream or a recording would be very convenient.

(sorry for repeated message) Hi, I tried to install the real-time api by writing "RUN pip install pupil-labs-realtime-api" in my docker file. But it gets the error "could not find a version that satisfies the requirement pupil-labs-realtime-api (from versions: none)", "no matching distribution found for pupil-labs-realtime-api", can someone help me out? Thanks!

Hi dom @user-cdcab0 ! I think I'm using python 3.6.9 in the image. I checked the documentation it seems the latest release requires 3.7, can I assign some older version that still surports 3.6.9?

Ok, thank you!

Hi, we bought the Pulpil Invisible in old version. But we didn’t book corresponding mobile phone. We are informed that we can book the mobile phone respectively.

This one

Yes, I saw it! Appreciate!!!

Hi, is there maybe maintenance going on in pupil cloud? It does not allow me to login and I have tried multiple browsers already. I saw someone else had the same problem a few days ago.

Hi! Could I ask what is included with the hardware at level of software and warranty? The same question for the core model, otherwise I ask it directly in the corresponding chat

thank you so much!