@kaylesk
For the external intervalometer tests what microSD cards did you tried? Maybe faster card will allow shorter intervals to set: maybe 2,4 secs or less… I use a Samsung 128GB Evo Plus. I will try different intervals with this card in the near future.
I already tried the emild Reach and worked mostly fine. Based on my test in urban area (not in urban canyons) but between the 10 floor block of flats I had +/- 25 cm precision with fix corrections and +/- 75cm with float corrections. The test was made walking. Sometimes Emlid Reach establishes the float or fix only after a long time of waiting (1-3 minutes), but once it happened that I could not get the position (not even single) for more than 15-20 minutes - being in an airplane. Otherwise I used a lot the Reach for aerial imagery collection.
When you make the stitching in App you will find some strange duplicates on the stitching line. Stitching it with PTGUI (I used a template with more than 50 control points) I had a strange stitching error. See in the attached centimeter graded image. The distance between the camera and the centimetric image was around 70cm, but the reference image was held more in a square shape not in a circle shape. See the stitching problem on the attached image (I also put there the InApp stitched image - the result looks quite the same): https://drive.google.com/drive/u/1/folders/184Ju0Pf-Pd4cqEzeHnYX9RiJ65u6utJx
This stitching issue may damage the result and the accuracy of the feature extraction from the Xiaomi panoramas.
First card was SanDisk Extreme 64GB, but write speed was weird (older card, write speed drop down and return and drop and… speed graph looks like “VVVVV”. Drop was under 7M/s)
Now Samsung Evo plus 64GB.
I think 2,5s would be OK. I did test it. Well, only indoors and with 2,6s delay, but 2,5s should be OK.
peewee32 above mentioned there is the bug (by design?) where it only updated the location every 10 seconds. This isn’t really any good for cycling/driving
Latest firmware has 1,3,5 and 10 seconds setting for GPS but not continues. Maybe 1 second would have a good result but I have not tried this because I use dedicated GPS to add location in post.
Great article! I have a suggestion that i am sure many photographers will love, since you can play in so many ways wirh the exposure time the mi sphere offers it would be awesome to release some kind of nd filters to allow daylightg long exposure shots! I had seen a post where a theta s was wrapped is some kind of nd “wrap” if developing filters is too complicated
regards
Yep, latest firmware has GPS update at 1 second interval and it surely writes locations to JPG exif. Tested on Nokia 7.1 Android 9, while simultaneously saving track in OsmAnd continuos interval. When comparing Mi JPG locations and if geotagged with GPX later on, some difference occurs, see GIF.
However, minimum image acqusition interval is 2 seconds (in settings) - reality is about 2-4 secs with U3 class card in JPG (non-stitched). Maybe using wired trigger to hit shutter every second could make it faster, need to try that. And faster card too.
Hello everybody , somebody can advise how to make better resolution more than 3456x1728 ,now on the pictures not see small letters for example street name Mapillary
Could it be your options in the camera settings? It’s supposed to support: 6921 x 3456 (when you see the preview over wifi on your phone, go into the setting and you can set the picture resolution(and video))
This happens when you upload the image before it is stitched.
Because of the meta data, Mapillary assumes it’s a 360deg photo, but it expects it as a equirectangular image, not as 2 spheres.
You can stitch them using the mi sphere camera software for a PC (windows only), in the app on your phone, or using the camera itself when you take the photo. Other 3rd party software is available.
If you try stitching on a phone that hasn’t enough CPU, I believe it tries to stitch it at a lower resolution instead, to make sure you at least get some image.
That aside, it’s a 24mpx camera, but that is spread over 360degree field of view. It does make reading text more difficult compared to a 14mpx camera but with only a 130deg field of view