Sorry your browser is not supported!

Let's talk about the idea, I had.

I am a photographer, and I could imagine, to create something like that:
https://petapixel.com/tag/miniature/

https://petapixel.com/2012/10/19/a-behind-the-scenes-glimpse-of-matthew-albaneses-magical-miniature-worlds/

I have the following games in mind:
Jack Houston And The Necronauts
http://www.warbirdgames.com/blog/jackhouston/
https://www.kickstarter.com/projects/warbirdgames/jack-houston-and-the-necronauts

Trüberbrook
http://trueberbrook.com/de/startseite/
https://www.kickstarter.com/projects/btf/truberbrook-a-nerd-saves-the-world?lang=de


Harold Halibut
http://haroldhalibut.com/
https://www.kickstarter.com/projects/slowbros/harold-halibut-a-handmade-adventure-game?token=58da2cd7


And the idea is, to have take only some pictures, create a deph map out of it somehow and use a shader-thingly, to create an effect like that:

https://tympanus.net/codrops/2019/02/20/how-to-create-a-fake-3d-image-effect-with-webgl/

Other examples:
https://blog.picjumbo.com/create-facebook-3d-photo-from-normal-photo/


https://fstoppers.com/originals/hacking-portrait-mode-create-3d-parallax-photo-facebook-189359

https://cdn.fstoppers.com/styles/large-16-9/s3/lead/2018/10/depth_enabled_3d_photo_facebook_0.jpg

So the question would be, how the pictures have to be taken.



I could imagine, that I have a stereoscoping picture. Two pictures. One from the left, one from the right and somehow a software would compare both pictures and spot the difference
into a third image, where the distance between the pixels from the left and from the right eye picture / image would be a deph map.

And the shader could use a pixel from the left or the right picture, depending on the viewing angle of the camera.

Most of the examples, I found, are only using one picture.


But what if, the camera looks from an angle, where the picture reveals no content? In some of the Facebook-App examples there is a blur, or a distruction or it looks too fake.
But in Trüberbrook they have done the effect by using photogrammetry for the complete environment. That would be to heavy on all aspects, because I would have real 3D objects and also would need to create 3D objects out of lot of pictures. And have to handle textures and so on and would have to a real 3D space in the game.

Since it is more like a better parallax effect, that could be the anwer. Only having 1, 2 or 3 pictures, a deph map and all is nice.
That would be also not so heavy for loading. And I could have some scenes in the memory.

I also wondering, if somebody understands parallax occlusion mapping and maybe knows, how that would work and if that could be something.

Another idea and question is, if I would take 2 or more pictures from the miniture, could I create a normal map out of these (asume that it are 2) pictures and would that help for the effect, I wanted?

An alternative would be to create a camera projection mapping with a 3D tool like Blender

So maybe we could find a solution, how this effect could be used in App Game Kit as a background or something

I am not sure that depth map is possible, but If camera is going to have only subtle movement, I would recommend camera projection technique.

Now, there is some limitation. If you do camera projection inside blender, you can't export THE camera itself inside AppGameKit, that is crucial, but only geometry. Second problem are textures - they are coming from camera, therefore, they are not applied on geometry, therefore, you will have only blank polygons.

To avoid these obstacles, you could try few things...

To solve texture problem you would have to bake textures on geometry from camera view.

To solve the camera issue is a tricky part. Use a geometry (cube) in blender as a pointer where your camera is placed. Do freeze transformation or whatever is necessary so you can have imported fbx elements inside AppGameKit at exact same position as in blender scene. Finally, experiment with camera POV to match same camera angle (35mm, 50mm, whatever)

Keep in mind that Camera projection is more efficient for dolly movement and less for panning - not impossible, you just can't move camera much.