@kotohibi_3d So it’s not an effort thing, both are cheap. It’s results: more overlap = better matching and loop closure, but more frames to process. Comes down to how much your SfM leans on global matching. Do you see seam tracks dropping on tall verticals with cube faces?
@kotohibi_3d And either is a few lines in cv2: build a lon/lat grid for your target view, project to equirect uv, then cv2.remap. Cube = 6 fixed 90° views, ring = same call at N yaws + 2 poles. Same primitive, you just choose how many views and where to point them.
@kotohibi_3d Ahh I see, but easy 20 lines of python cv2 code can extract 12 pin hole camera images from spherical images to create dataset that many of these programs can ingest, right?
If one hasn’t heard of @Chobani CEO’s story, please do. What an amazing guy - Hamdi: came from the Turkish mountains, makes yogurt, employs mostly refugees, pays them twice the market wage with shares in company! Has vouched to give all money away to charity!
What? Really? Selfish much?
Fun fact: I soared through their recruitment rigmarole & at the end of final interview for software engineer role, told them to give my position to someone else as I don’t see to fit in there. So glad I didn’t go software engineering route!
@sreenivasan_ac@FairAmericaOrg Indians are the most raciest set of people I’ve seen, discriminating people based on caste (ie being born in a family that happens to be lower caste)! Now they cry when the tables are turned in a foreign country where they are the minority & don’t understand country based caps!
@aviseka@Tesla No, all manufacturers use GOM - Guess-O-Meter. Don’t take the miles left for granted. Depends on how heavy footed one is, outside climate, car climate settings etc. Always have a buffer to destination.