Hi there! My team has found that the export button in the app appears to take files from a folder on the device (whose internal organization we can’t fully determine) and map them into the export folder where we can access them. We’re developing a video storage/transfer pipeline and want to streamline this step. Is there a system or documented pattern behind these exports that we can take advantage of? We’re not currently using your API, but we’re considering it if it’s the right way to mimic the app’s export behavior in our own pipeline. Is there a supported API workflow for this, or guidance on how to mirror export logic safely? Our concern with the real-time API is reliability over institutional Wi-Fi. Our main goal is to understand whether we can build a wrapper around Pupil Labs export behavior. If not officially supported, could you let us know whether the behavior is consistent enough across versions that reverse-engineering the pattern would be feasible and reliable?
Hi @user-bbbd3b , there is no API for exporting data from the app/phone. That function copies only the data that a user needs to a new folder and renames the folder to be more user-friendly. The main idea is to make it easier for users to keep track of the exported data and to always be sure that there is an untouched original copy of the raw data, in case anything goes wrong.
May I ask though if you are refering to the Real-time API or the Pupil Cloud API?
In any event, while copying the original unexported raw data is not a supported workflow and we cannot provided dedicated support for that if anything goes wrong, there is nothing in principle stopping you from doing it, as they are simply files on the Android filesystem. They will load into Neon Player or the pl-neon-recording library, for example.
If you are concerned about institutional WiFi, you can also use a local dedicated WiFi router, a hotspot, or even an Ethernet connection. However, re-creating the Native Recording Data from a Real-time API stream would be a substantial amount of work when you can just run a recording in the app and use the high-fidelity data directly with the tools in our ecosystem.
Hi! I am trying to visualize the 3d point of the gaze from world to camera coordinates. The given 3d point for the gaze is the following: 'gaze_point_3d': ( 97.62555499946531, -92.17455507256946, 398.59534061149367)
I would like to know which coordinate frame is the gaze point in, as I have a few world cameras and have successfully found a transformation between the world and eye cameras, but the coordinates I get in the world cameras is in this index: [ 0.012684360843151115, 0.22219709315722463, 1.4293409732234759 ]
is it possible that the gaze point 3d is not in meters or are the gaze_points_3d with respect to the eye camera? Is there any additional transformation that is necessary?
Hi @user-3d30b4 👋 ! I assume that you are using 👁 core based on the output. In that case note that the gaze_point_3d is output on the world camera coordinates, which follows OpenCV conventions . Let us know if that clarifies it.