Help / For whoever holds the camera
Filming and exporting 360 video
If you are uploading files someone else shot, you still want the export section, because that is where most problems start.
What the headset can play
| Container | MP4 |
| Video codec | H.265 (HEVC) |
| Resolution | 5.7K or 8K equirectangular |
| Layout | Monoscopic, or 3D top and bottom |
| Web optimized / faststart | On |
Vantura checks these when you upload and tells you in plain language if something will not play.
The one setting that trips almost everyone
Your camera already records H.265 internally if it shoots above 4K. The problem is the export. Phone companion apps, including the Insta360 app, commonly re-encode to H.264 when you share or export. The file looks fine on your phone and then fails the upload check.
Set H.265 in the export options before you transfer the file. Do that once and you will not think about it again.
Why H.265 and not H.264
This is a limit in the Quest's video chip, not a rule we invented. Meta's documentation is direct about it: Quest supports 4K in all formats, and supports 8K only on H.265 and AV1. So H.264 hardware-decodes up to 4K and no further.
That matters more for 360 than for flat video, because the pixels wrap around an entire sphere. A 4K equirectangular video gives you only about 960 pixels across a 90 degree view, which looks roughly like 540p once you are inside it. Dropping to 4K to satisfy H.264 is not a real fallback.
So the upload check is catching files the headset physically cannot play. Without it, the failure just moves from your desk to the classroom, in front of students.
Cameras
| Camera | Works | Note |
|---|---|---|
| Insta360 X4 Air | Yes | 8K, 7680 × 3840, H.265 native. Nothing to convert. |
| Insta360 X4 / X5 | Yes | Export as H.265 |
| Insta360 X3 | Yes | 5.7K, export as H.265 |
| GoPro Max, Ricoh Theta X | Probably | Confirm the export codec before you shoot a whole lesson |
| H.264-only bodies (for example AKASO 360) | No | Cannot produce a file the Quest will play at full resolution |
Before committing to a camera, shoot 30 seconds and run it all the way through an upload. That is a cheap test and it answers the question completely.
If you already have an H.264 file
You can convert it rather than reshoot:
ffmpeg -i input.mp4 -c:v libx265 -crf 20 -tag:v hvc1 -movflags +faststart output.mp4
If the only warning you got was about web optimization, you do not need to re-encode at all:
ffmpeg -i input.mp4 -c copy -movflags +faststart output.mp4
Converting costs time and a small amount of quality. Fixing the export setting is better.
Shooting notes for teaching
- Mind the horizon. Vantura lets a viewer turn left and right, but deliberately does not tilt the view. A tilted horizon stays tilted, and a horizon that is off level is one of the more reliable ways to make someone uncomfortable in a headset. Get the tripod level.
- Put the camera where the learner should stand. Head height, in the position a student would actually occupy. The camera is the viewer.
- Hold each shot longer than feels natural. People spend the first several seconds looking around rather than at what you want them to see.
- Say where to look.Narration like “behind you, on the supply cart” works well, because the viewer can act on it.
- Watch your own position. Everything within a few feet of the lens is where stitching artifacts show up.
How far back to put the camera
This is the single setting that most changes how a lesson feels, and it is the one people get wrong first. A 360 camera records true angular size, so whatever fraction of your vision a person filled at the lens is the fraction they fill in the headset. Place the tripod at conversational distance and the viewer has to look them up and down to take them in.
| Camera distance | How much of the view a standing adult fills | |
|---|---|---|
| 3 ft | The entire height of your vision | Overwhelming |
| 4 ft | Most of it | Still too close |
| 6 ft | About half | Workable |
| 8 ft | Under half | Comfortable |
| 10 ft | About a third | Person plus room context |
Aim for 8 to 10 feet for anything where the viewer should see a whole person and the space around them. There is a second reason beyond framing: monoscopic 360 carries no stereo depth, so a close subject reads as a flat wall looming rather than a person standing nearby. The comfortable minimum in a headset is further out than it would be in the room.
The exception is deliberate. If you are filming a procedure the student should watch over your shoulder, or a bedside interaction, close is correct, because that is where the nurse actually stands. Decide which shot you are making before you set the tripod down, rather than discovering it in the headset afterwards.
File sizes to expect
A 5.7K lesson runs roughly 3.4 GB. An 8K lesson runs roughly 6.8 GB. Both are normal. Uploads resume on their own if your connection drops, so a large file over a mediocre connection is an inconvenience rather than a failure.
Exact bitrates and sizes are in the encoding guide.