What content would you create with a 100MP camera system?

Forum|Forum|9 years ago
March 3, 2017
7 replies
1911 views

Hi all! This is my first post here and I'm just a simple photographer, new to the VR world, so please correct me if I say anything dumb or violate any posting etiquette.

I happen to have access to a full lineup of Phase One 100 megapixel medium format cameras, and I was wondering if there was any interest in the VR community in creating content with such systems. It would take a significant amount of data processing, and most of the interest in VR seems to be in the video domain, but if you wanted hyperrealistic environment/object recreation, I feel like these systems might be fairly useful. Here are some sample 2D photos taken on the system (google drive link because files are too big to attach):

https://drive.google.com/drive/folders/0B4lJAQLztAD8TWdHWDY0U2lYQ1U?usp=sharing

In a nutshell, I'm wondering whether there's an interest in high resolution photography for VR. Looking forward to your feedback!

Thanks!

Best answer by

360° photo-spheres and panoramas are the first thing that come to mind ... with that kind of camera, the world detail would be amazing. Can you remember the 360° Panorama done from the top of Tokyo Tower, where you could pan around and zoom in multiple times, revealing more detail.

Can't remember the name of the process, but there's also a way of creating 3D objects and environments using photographic images. Steam and and few other companies were/are doing work in that area... The camera you have is a great tool, allowing for some great detail... and VR is about that up-close detail.

A

Anonymous

Answer

360° photo-spheres and panoramas are the first thing that come to mind ... with that kind of camera, the world detail would be amazing. Can you remember the 360° Panorama done from the top of Tokyo Tower, where you could pan around and zoom in multiple times, revealing more detail.

Can't remember the name of the process, but there's also a way of creating 3D objects and environments using photographic images. Steam and and few other companies were/are doing work in that area... The camera you have is a great tool, allowing for some great detail... and VR is about that up-close detail.

K

kojack

Honored Contributor

Paul_Blythe said:

Can't remember the name of the process, but there's also a way of creating 3D objects and environments using photographic images.

Photogrammetry.
I use VisualSFM (free), but others work too.
Here's a little test I did a while ago:
https://www.youtube.com/watch?v=u7Q9j7le5pY
The top left is the original photos, top right is the point cloud, bottom left is the generated mesh, and bottom right is the final textured mesh.

In particular, a 100MP camera would be good at fisheye lenses. I have a Sunex fisheye lens which is 185 degrees. In theory one photo with that would have around 47 pixels per degree. The CV1 rift has about 15 pixels per degree.
(I say "in theory" because it depends on the fisheye lens and camera sensor size. The sunex one is made for crop sensors, so on my full frame sensor it doesn't fill the image, reducing effective resolution)

Author: Oculus Monitor , Auto Oculus Touch , Forum Dark Mode , Phantom Touch Remover , X-Plane Fixer Hardware: Threadripper 1950x, MSI Gaming Trio 2080TI, Asrock X399 Taich Headsets: Wrap 1200VR, DK1, DK2, CV1, Rift-S, GearVR, Go, Quest, Quest 2, Reverb G2, Quest 3

HiroProtag0nist

Honored Contributor

Hi @kojack - i've just had a look at VisualSFM and watched the video here: http://ccwu.me/vsfm/

It says using pre-computed matches - how do you match a group of photos, let's say of a room - is there any manual work or do applications do the work for you?

Big PC, all the headsets, now using Quest 3

K

kojack

Honored Contributor

It can use pre-computed matches. But it can also auto match photos. The simplest workflow is:
- use the open multi option to open a bunch of photos.
- select reconstruct sparse. This will match photos, calculate camera transforms and generate a very low detail point cloud.
- select reconstruct dense. This takes the camera data from the sparse pass and uses it to generate a detailed point cloud.

I don't use it's built in dense option though. Instead I use CMPMVS. http://ptak.felk.cvut.cz/sfmservice/websfm.pl?menu=cmpmvs
I use a batch file that calls visualsfm to match photos and generate the camera data, then cmpmvs to generate the 3d model. It also generates that sample video I posted as a side effect.

The batch file:

@ECHO OFF

IF [%1]==[] (
	@ECHO Usage path\VisualSfM.exe imagesDirPath
	@ECHO Result will be in imagesDirPath\_CMPMVS_00_OUT
	@ECHO OFF
) ELSE (
	C:\Users\kojack\Downloads\VisualSFM_windows_cuda_64bit\visualsfm.exe sfm+cmpmvs %1 %1\cmp.nvm
	CMPMVS.exe %1\cmp.nvm.cmp\00\mvs.ini
	md %1\_CMPMVS_00_OUT
	copy %1\cmp.nvm.cmp\00\data\_OUT %1\_CMPMVS_00_OUT
)

Author: Oculus Monitor , Auto Oculus Touch , Forum Dark Mode , Phantom Touch Remover , X-Plane Fixer Hardware: Threadripper 1950x, MSI Gaming Trio 2080TI, Asrock X399 Taich Headsets: Wrap 1200VR, DK1, DK2, CV1, Rift-S, GearVR, Go, Quest, Quest 2, Reverb G2, Quest 3

HiroProtag0nist

Honored Contributor

Thanks kojack, i'll take a look at this.

Big PC, all the headsets, now using Quest 3

R

rjay2b

New Participant

What I'd love is for cameras like these to be taken to places of great natural beauty worldwide, and records made of what was once here, because so much of it will be lost over coming decades. Glaciers, coral reefs, rainforests, places of great biodiversity in animal and plant life. I'd love to see it as much as possible captured before it is all choked, poisoned, burned, stripmined or drowned. VR will eventually be the only place these things exist.

A

abcdtAuthor

New Participant

Thanks all for the awesome replies!

@kojack @andyring @Paul_Blythe@rjay2b
2b - is there a specific area in the forums that's geared towards photogrammetry/360 photo spheres?

Get Help

Sign up

Login with SSO

Login to the community

Login with SSO