Hi, I have a question that's important to me before I initiate a research project where loss of data would be a very serious problem.

I tested the pupil invisible and pupil cloud functions with a two minutes video and when i did the marker mapper enrichment it took about 5 minutes for the pupil cloud to do the calculations to add the features to the 2 minute video

The thing is, for our research project we want to two procedures of up to 75 minutes for each participant So I'm wondering if that will be a problem and what your experience and recommendation is? Can the cloud reliably process a 75 minute video (or does the app have a timer limit?) ? When adding the marker mapper enrichment, do i risk losing data/the video, if the enrichment addition fails? Does internet speed or other factors affect the ability of the cloud to add the enrichment feature? If a 75 minutes video isn't possible, what do you recommend? (our study kind of depends on the participants doing a procedure that long)

We want to have two sets of students (of 12 each). They all have to do two procedures in a simulator program on a computer (working with the mouse and a haptic device in their right hand, while viewing the surgical simulation on a screen).

All students must do two procedures of 75 minutes each. During those procedures we want to estimate their eye movements to observe gaze entropy and where they're looking. Ideally during the entire procedure. I guess it could be possible to pause the procedures and stop the recordings at several points during those 75 minutes if necessary (and then upload) and start them again. But it would not be ideal, since it would give the students time to "rest" which isn't ideal for the study.

Thanks for providing more details @user-f6ea66!

If you need to map gaze onto the screen/simulator program during the entire session, then you can apply the enrichment for the entire recording. Analysis in this case would take longer as you will be applying the mapping to 12 subjects x 2 groups for 75-min long recordings.

Alternatively, if you do not need the full recording data and you prefer opting for faster analysis, you could create Events to slice up the recording and apply the Marker Mapper only on a predefined section of e.g, 30 minutes.

Ok, thanks. So you don't think there is any risk of a 75 minute continuous recording not uploading correctly/safely to the pupil cloud?

And when it's uploaded, it shouldn't be an issue for the pupil cloud 'software' to add the marker mapper enrichment to the entire 75 minutes in one go? (provided that the 'apriltags' are recognized of course, which they were in my two minutes test) That it takes time shouldn't be a problem since it's done after the procedures safely uploaded to the cloud. But do you have a rough estimate of how long it would probably take to add the marker mapper enrichment to a 75 minute video?

If that is the case, then we will make 75 minute recordings to be able to see the entire process in one analysis, including how gaze entropy changes during the procedure.

Thanks!

(sorry if I'm repeating somehow - just want to make sure I've understood it right)

Hi @user-f6ea66 !

There should be no issues with uploading recordings to the Cloud, regardless of their duration. It will take longer to upload, but the data on the device remains unchanged.

Here are the details regarding your concerns:

• Marker Mapper Duration: Yes, Marker Mapper can handle recordings longer than 75 minutes, there is no timer or time duration limit.
• Note that when a recording is uploaded, that recording will be transcoded first in the Cloud and then the neural network will be run again to ensure you get the data at 200Hz, so you may want to consider this buffer time too.
• As my colleague @user-480f4c mentioned previously, the enrichment processing time depends on the duration of the recording and the current workload on our servers which is hard to estimate To be safe, allow a couple of hours for the entire process.

If you have further questions or need additional clarification, feel free to ask.

Fantastic - thanks! ^^

Hi,

Is there any way to trim the video I took? It's a reference video and is a little bit over 3 minutes. How can I cut that extra part? Thank you.

Hi again, Do the Pupil Invisible glasses record and output pupil diameter metrics/data? I see the parameter in one of the output files (3d_eye_states), but it is empty. Thanks!

Ok, thanks for the swift reply : )

Hi, The two phones that I have are updated automatically. I have a OnePlus 8 and a OnePlus 8T. Currently, after the rollback, the 8T is on Android 12 and the 8 is on Android 10. I am having problems downloading the app and capturing the data properly. Can you provide the correct steps to take? Note that the files you put on your website for the rollback are not working.

Hello, I am running a study using pupil invisible to measure gaze beaviour while walking. Many of the participants will be using a gait aid (e.g., walker, cane), so they won't be able to hold the device while walking. I want to avoid using pockets to keep the device if I can. Anyone have any ideas/things that have done to keep/hold the device while a particpant is walking? Thank you!

Hi @user-f8a8b8, thanks for reaching out 🙂 Have you considered something like a sports armband, a small bag, or a waist pouch? See https://discord.com/channels/285728493612957698/1047111711230009405/1257997423670722710

Hi, I am trying to create the reference image mapper, but I keep encountering issues. It either results in an error or displays the message: "Reference image mapper definition ID does not match the specified ID." My scanning video is of a historical building, which I took to test this enrichment. The video is about 1-2 minutes long, as recommended, and I also took a photo of the building. Could you please assist me with this? Additionally, I would like to know if the heat map will still work if our perspective changes and we are moving a lot while the object of interest remains static (like in golf). Thank you

Thank you, Miguel, for your answer. I would like to know if, for our future research in golf, where the player will move a lot, this enrichment will work?

Hi! A quick question! Does invisible have an offline desktop software similar to Pupil Cloud?

And is there an offline desktop data recording software?

Hi, I have a quick question. If you have an environment like a golf course with many different objects at various distances, for example, a tree which is 2 meters away from the wearer or the flag which is 5 meters away from the wearer, will the reference image mapper enrichment work? Thank you .

Hi @user-58c1ce ! Sy I missed your answer! It's hard to say whether it would work or not without seeing the environment as there are many factors that play a role here:

• Can enough features be tracked and matched?
• Is the illumination changing?
• Is the field of view occluded by any object?

My recommendation would be to read our Best Practices if you haven't done so and pilot the data on similar environment.

Thank you, Miguel, for your answer.

1.Yes, we have flags, bunkers, water, etc. 2.Not really; it will be outside in a normal environment. 3.No. Thank you again.

Thank you

Hello! Is there a better way to calibrate the gaze location than simply dragging it with a finger in invisible companion app?

Hey guys, If i wear the glasses and then shift my head and look relatively centered,.. is it the same as if my head not tiled and i look centered. Do i get the same coordinates out for gaze? Because both are technically focussed or directed towards the center part of the world camera frame?

I need to download the Invisible Companion App but it's unavailable - is there another way to get it?

@user-d407c1 Meaning. If i just take the coords from the fixation.csv it these coords account for the head movement. Meaning if i look straight in front of me. And later after 3 fixations i tilt my head to the left, but my eyes remain focussed on the middle, those will be different coordinates?

Hi, for our research in golf, we would like to identify areas of interest. I tried to create a heat map, but it did not work well because our participants are moving so much within their environment. I also tried using Pupil Player, but I noticed an offset when using Pupil Invisible with iPhone 8 and 8T. Do you have another method to determine the areas of interest? Thank you.

Being in the USA should not present any problem. What happens when you click on this link from the device https://play.google.com/store/apps/details?id=com.pupillabs.invisiblecomp&hl=en ?

When I use the link you provided, I receive this message from the Play Store:

Hey Guys anyone knows the new link for this? https://docs.pupil-labs.com/invisible/data-collection/compatible-devices/#oneplus-6-companion-device-setup

thanks

Much appreciated! @user-07e923

Hello! Is there a way to directly download Pupil Player format data from the phone directly for Invisible or Neon?

Hello. I have different recordings with areas of interest delimited by aruco markers. My participants were completing a task looking at the screen and the screen is an area of interest (delimited by the markers). In each recording I created some time of interest that indicate what subtask the participant was doing. Is it possible to generate heatmaps of a specific time of interest (subtask) and not the whole recording in pupil cloud? Can't figure out how. Also my recording are in a single project

@user-5fb5a4 on another note, I'm not familiar with the aruco markers you're using. The Marker Mapper enrichment on Cloud supports only AprilTag markers that are listed here: https://docs.pupil-labs.com/invisible/pupil-cloud/enrichments/marker-mapper/#setup

thank you!, very clear. I have one more question, in case the subtask I want to create the heatmap is repeated during the recording, for example multiple pictures of men are shown, and in between there are picture of women. The label should be the same man_1_start, man_1_end for each of the men's picture? can i label them e.g. man_2_start, man_2_end and so on and create a single enrichment and then create the heatmap from that?

Understood, so I can't select more than one start and end label in the temporal selection. for example in temporal selection: start -> man_1_start, man_2_start, etc. In my case I named them like this for many recordings that's why I am asking, I will have to re-label them. Thank you so much again!

Is there something I can do only with Pupil Cloud but not Neon Player?

Hello, sorry if I came back to this. I don't get how the image on the right, where the heatmap will be superimposed gets selected. In my case, that frame is not present in the temporal selection I selected, nor the frame in which I defined the aoi surface

what do you mean by scenes?

the scene changes slightly because the machine moves a bit

in this case it is pretty much what you see on the left

I created another enrichment from on another participant and i get the same image on the right

for the same temporal selection unload_1 and rev_u_1

@user-5fb5a4 could you maybe invite me to your workspace so that I can have a better look and give you more precise feedback? If yes, let me know and I can send you my email.

of course

thank you for the explanation, looking forward to the update!

Hi @user-5fb5a4 ! As a temporary workaround until the update, you may want to try the following:

• For the heatmaps: The data does still respect the section you selected and therefore you can either recreate it with the data or simply download the heatmap visualisation, which should contain one image with and one without the background, (i.e. transparent background). Then, you can simply re-apply it over your image, ensuring the scale is the same.

• For the AOIs: As for the areas of interests, the main issue you might be facing right now, is that you can not draw them over a different picture accurately, correct? Well, what you can do is to use any picture you want and follow this tutorial which will upload the mask to the Cloud, regardless of the original image, in Cloud. Again the only thing you need is to grab a picture that respects the scale of the original, and is cropped by the markers, as in the one selected automatically.

Thanks for your reply, I am not sure what I am using, I have just recorded videos on pupil cloud until now and I have the glasses, they are supposed be the latest version apparently according to my college professor, just some basic visuals which give very basic insights into fixations or other stuff

Just checked it out i think pupil invisible

I am unaware about any analysis features offered with the glasses and how to access them

If you could just let me know how to access any analysis or visualisation features and use them, it would be great

hello! I am working on the same project as the earlier discussion (I put an image for reference). I would like to create a scanpath from the screen (delimited by the markers basically) from one or more participant from again an event of the recording. Is it necessary (and possible) to create a reference image mapper enrichment and then use python script? Also, what if I am recording an event that is happening on a computer screen and it is moving? In that case, I can't really have the 1 minute video of the environment. Am I missing something?

I'd need the scanpath (from a whole scene of the recording) applied to a image that represents that scene. Something like this image. Even without the mapping.

As for the last question yes, I was referring to the scanning recording for the reference image mapper

Hi, I am using the PupilInvisible for the first time, and trying to get the option to stream to a laptop for a demonstration. However, I think I must have an older version of the companion app, as I do not have the 'Streaming' option available when I open the Companion app. Is it still possible for me to use the Monitor app and functions? Many thanks, Clare

Hotspot with Invisible

Hello, the data I uploaded is not being transferred to Pupilcloud. What could be the reason for this? How can we resolve it?

Is there any chance to cut recorded video?

I have another question. As a beginner in research, I'm unsure what data I will need for analysis. I will be studying performance in basketball players, and I'm having trouble understanding which data I need to analyze, especially the data I have obtained in CSV format. If you could share any beginner-friendly resources for analysis, I would greatly appreciate it.