combining several 360 images to make single high resolution image

[Tanzeel -ur-Rashid]

Hi, Is there any way to combine several 360 images (Raw files) of insta 360 one x2 of a particular room taken from different directions to develop high resolution single file with more detail.

The issue is that insta 360 is a consumer camera and can't be used for high resolution virtual tours. In this way, one may use this camera to take several shots of small retail store to make a rich single 360 image.

[John Houghton]

On Monday, November 29, 2021 at 5:29:45 PM UTC tanzeel....@gmail.com wrote:

Hi, Is there any way to combine several 360 images (Raw files) of insta 360 one x2 of a particular room taken from different directions to develop high resolution single file with more detail.

 What do you mean by "taken from different directions"?  The camera already takes a shot of the view in all directions.  Do you want to combine 360 degree shots taken from different positions in a room?  If so, the answer is no, that's not possible.  You can combine multiple shots from the same position, rotating the camera to "point" in different directions, but that would not yield higher resolution.  But you could eliminate bad quality at image joins that way.

John

[Mike Cowlishaw]

John, on your comment:

  " You can combine multiple shots from the same position, rotating the camera to "point" in different directions, but that would not yield higher resolution."

I think it could.  Consider the pixel-shift modes offered by various vendors (see, for example:  https://www.bhphotovideo.com/explora/photography/tips-and-solutions/pixel-shift-shootout-olympus-vs-pentax-vs-sony-vs-panasonic).   These work by moving the sensor a little.   That cannot be done easily with current 360° cameras, but changing the viewpoint slightly .. even if more than one pixel .. would surely allow common data to be averaged out and extra detail to be determined, just as though the sensor had been shifted slightly.

It could be a little hit-and-miss (in the worst case the view would change by an integral number of pels over the entire image) but given the fish-eye nature of the images that would seem unlikely.

In short, this sounds like a very interesting idea.  Perhaps a bracketed series of shots at the same exposure after giving the camera a nudge would work.

Mike

--
You received this message because you are subscribed to the Google Groups "PTGui Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ptgui+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ptgui/1ac7f679-6f24-4408-8a98-6a7ac652533bn%40googlegroups.com.

[Erik Krause]

Am 29.11.2021 um 20:53 schrieb Mike Cowlishaw:

> These work by moving the sensor a little. That cannot be done easily
> with current 360° cameras, but changing the viewpoint slightly .. even if
> more than one pixel .. would surely allow common data to be averaged out
> and extra detail to be determined, just as though the sensor had been
> shifted slightly.

There have been lots of experiments to do superresolution this way, so
yes, in theory this might be possible. You can use PTGui to align the
images, but it won't do the calculation for you.

--
Erik Krause
http://www.erik-krause.de