Benefits of cubic panorama formats?

I was working with a client this week who was asking about cubic and equirectangular panorama formats.

I had never heard of the cubic format before and but it seems I have been living under a rock because there is lots of info online.

Cubic

The client seemed to think cubic was a better format for panoramas than equirectangular, but I really don’t know why. And even if this is the case, I am not sure if my camera stitching software (Insta360 Pro2) can even create panoramas like this. Can anyone enlighten me?

Cube maps (cubic format) are more regularly found in video game tech (perhaps your client has some experience in this field?).

Most consumer camera stitching tools, inc Insta360, will output into equirectangular format simply because it is the most widely used in the world of 360 photography software.

That said, when performing editing or analysis (e.g. object recognition). the cubemap format can be useful in the world of 360 photography.

Equirectangular images are stretched in the ‘latitude’ direction. It is the reason for a considerable amount of data redundancy near the poles. When downsizing an image in an editor, the effective texture resolution is decreased as expected - except near the poles. This can often cause problems when viewed later in panorama viewers.

An equirectangular projection is thus suitable for simulating only those environments where the distortions of the texture at the top and bottom of the sphere are negligible.

If you do want to scale, blur or sharpen a panorama it is better to convert it to cubemap first, a less distorted projection, make the edits, and convert it back to equirectangular format afterwards, if necessary, to avoid these issues.

Out of interest, what work are you proposing/doing for this client?

I am thinking of using cubic by default… might it be because of the single file thing that equirectangular is a more popular system?

As you said, there are quite a few ‘redundant’ pixels at the top and bottom… I have the feeling that cubical is less noisy and as to ‘quality per pixel’ is going to win… also when blurring for privacy I think that cubical will have better results in the long run…(?)

dslr 6 cube shot and stitched in cube has the least distortion, as far as I recall reading.

equirectangular is the output from most prosumer 360 cameras. If you’re just rendering the photo in a web viewer I see little need to use an extra step to convert to cubic.

The early versions of pii-blur suffered with the distortions of the stretched north / south poles causing a variable detection rate of people / license plates. That said, it was still pretty good and could be modified to work to an acceptable standard using equirectangular projections.

HQ | Blog | Spotted a Trekker? | Become a Trekker | Facebook | Instagram