Creating 360 Panoramas and Collecting Data on Them in VR Using AI and SightLab
June 27, 2023
Skybox AI uses AI to generate full 360 degree panoramic images. Once you have a Skybox generated you can immediately download it to SightLab (Worldviz’s experiment generator plugin based off of the VR software toolkit, Vizard). From there you can collect and output data as well as view a wide array of visualizations for analysis.
2. Simply type in a prompt in the prompt bar to generate a new Skybox.
For my example I created a few kitchen environments of various color schemes. We can then run these in SightLab and analyze attention distribution, fixation duration, pupil dilation, saccades, revisits, and physiological measurements such as heart rate through SightLab’s connection to Biopac Acqknowledge.
You can also choose a style from a wide range (Realistic, Digital Painting, Claymation and much more)
The Tool Bar to the left gives you optional brush tools and guides. Shapes and lines you draw will be followed in your new skybox alongside the prompt. On the left you can also access the various grid outlines.
Additionally you can choose to download a depth map that can be loaded into a program like Meshlab to generate a 3D point cloud.
“Remix” lets you give your scene a new style without altering its basic geometry, so you can change things like time of day, weather, materials, visual style, etc.
3. The Skybox will generate fairly quickly. Once it’s done you can click on “Download” to download the panorama.
4. Save your 360 image into the SightLab - utils- resources- media folder
5. Next, run SightLabVR_360.py and choose your newly created panorama (for running a series of panoramic images see this page)
6. We can also set areas of interest, see this page in the SightLab documentation on how to do that. I’ve set the stove and table to be areas of interest
After running a session with this panorama you can view a session replay by launching SessionReplay.py to view analytics such as heatmaps, scan path, fixations, etc.
7. After running a session with this panorama you can view a session replay by launching SessionReplay.py to view analytics such as heatmaps, scan path, fixations, etc.
8. In the utils/data folder you can view a summary of fixations, raw tracking data to view things like pupil diameter and gaze position and if you saved a video recording of your session you can find that in the utils/recordings folder
9. To synchronize with Biopac Acqknowledge, simply change the biopac flag in the settings.py file to “True” and this will then drop markers in Acqknowledge that you can use to skip to various focus points in the session (where a user might be dwelling on an area of interest). See this page in the SightLab documentation on running alongside Biopac Acqknowledge.
10. To add more complexity to your session see the documentation on both SightLab and Vizard.
If you want to see how you can analyze your 360 panoramas, as well as 360 videos, 3D models and more using SightLab, contact firstname.lastname@example.org for a demo or to obtain a trial version.