🥽 core-xr


user-0eef61 01 September, 2021, 09:52:47

Hi, I am recording data in VR and I see that gaze_positions.csv file is always empty. Is this something that should happen or I have forgot to press a button in pupil capture? Same happens with: head_pose_tacker_model.csv and head_pose_tacker_poses.csv

user-5349b5 03 September, 2021, 10:05:05

Hi, I am making 3D contents for VR headset viewing by Unity 2020.3.16f1. I import 'Hmd-Eyes.VR.V.1.4.unitypackage' for the above program for eye tracking. I import 'Gaze Tracker' as same setting as the sample demo 'ScreenCastDemoScene'. The program can be built, but the marker of gaze tracking that does not work at all (the maker still stay the same position). Looks did not line to eye motion. Does anyone give me advise what I should do? Thanks

user-525b2f 04 September, 2021, 15:43:10

Attempting to connect. Pupil capture has view from both cams but cannot connect to the Pupil via Unity. Any ideas?

Chat image

user-73b616 05 September, 2021, 08:48:13

you should run pupil service not capture afaik

papr 06 September, 2021, 07:36:13

You can run both. You can check if the port is correct by looking at the menu of the network api menu in Capture

user-d1efa8 19 September, 2021, 04:31:47

sorry i hadnt seen this before

yeah, you set the camera to only render certain layers instead of everything. Probably the easiest thing to do is assign the gaze visualizer to the UI layer (top right in inspector, where it says "layer" then from the dropdown select "UI") then in your camera that renders to the HMD, there should be a dropdown called "Culling Mask" and by default, it should be "Everything" but you can just uncheck "UI" and that should work

user-d1efa8 19 September, 2021, 04:33:41

You can just go to the "Gaze Visualizer" object and assign it and everything under it to the UI layer and it should work

user-44af31 29 September, 2021, 07:17:56

Hi! For my research I am using pupil size measures. As we are moving tiwards vr study designs I piloted the Pupil Labs HMD tracker. However, it seems that the exported pupil data is very noisy and therefore unusable. Data quality did seem to be good during measuring (As seen by a high confidence and the blue/red circle being stable). Any ideas on possibilities to increase data quality? Sub-pixel (we're are looking at 0.1-0.3mm changes) accuracy seems to be required, is the system suitable to do So?

papr 29 September, 2021, 07:46:49

Have you frozen your model during the recording? You can read more about this here https://docs.pupil-labs.com/core/best-practices/#pupillometry

Also feel free to share an example noisy recording and we can let you know if you can get better data with post-hoc detection.

In general, the system should be able to capture such differences, assuming that you have enough trials to average over.

user-44af31 29 September, 2021, 07:59:28

@papr Hey thanks for responding! I didn't freeze the model at the time as I've been looking at the 2d data. I did notice that freezing the model helped with getting good 3d data. Let me prepare an example to show you what problem I have encountered.

user-0eef61 29 September, 2021, 14:12:30

Hi, following the last topic from @user-44af31 it was helpful to read the best practices before collecting the data. I had a question, how do we validate the calibration and where can we see the error?

papr 29 September, 2021, 14:13:57

Actually, I am not 100% sure that we support validation for the HMD choreography. 😕

user-b14f98 30 September, 2021, 19:38:16

Hey folks, I would like to automatically set the recording directory and then start recording upon connection to the local PL server from Unity. Can you provide me with a pointer to the appropriate topic subscription to listen for?

papr 30 September, 2021, 20:48:22

Just send the R <directory> command via Pupil Remote. 🙂

user-b14f98 30 September, 2021, 20:12:14

..I believe I found an example in calibrationstatustext.cs

End of September archive