Hi,

I have a slice, which shows inside of a real sphere.

I want to render this slice to make it look really inside the sphere.

I know the relative pose of the slice to the real sphere, so the view matrix is not an issue.

The problem if how to add some visual effect to make it looks real. 

If I just simply render both sphere and the slice, it give a feeling the slice is dangling in the air.

The way I can figure out is to add a hole geometry on the sphere to give us a feeling that we look the slice through a hole, so the slice looks like inside the sphere.

Any suggestion about this method and other suggestions?

Thanks a lot!

YL

I'm not sure if you explanation is detailed enough?

Do you want to draw the intersection between the slice and the sphere? You can do that in the pixel/fragment shader by discarding fragments with world positon that is in front (or outside) of the slice?

Can you show a image of the type of effect you are looking for?

Sorry for unclear description of my problem.

Actually, I do not have the geometry of the sphere and cannot render it. The only think I have is one slice (a png format image) of the inside of the sphere. This slice is an image cutting through the sphere, like ultrasound image. 

I use a real camera to take a picture of the sphere and then want to render this slice combined with the picture to give people a feeling that this slice is inside the sphere.

If I have to have more information, I can use a stereo camera to get the depth information of the sphere. With the depth information, Maybe I can create a hole as KosadinPetkow said.

14 hours ago, YixunLiu said:

I use a real camera to take a picture of the sphere and then want to render this slice combined with the picture to give people a feeling that this slice is inside t

If the real picture is a texture on top of a 3D model then you would manipulate the 3D model and textures.

If the real picture is an augmented reality image then it will be much harder. You'd need to turn the object into a model through whatever means you have available, such as back projection or object registration or an object recognition library someone has already developed.  If that is the situation then I strongly recommend using an existing tool.  This will create a 3D model and texture that you would manipulate the same way.

On 7/4/2017 at 1:32 PM, YixunLiu said:

The way I can figure out is to add a hole geometry on the sphere to give us a feeling that we look the slice through a hole, so the slice looks like inside the sphere.

Once you have the model and texture to manipulate, manipulating the graphics will require some knowledge of graphics and model processing, or some tools that can do it for you.

For something simple you could make a slice of the model's geometry, or perhaps use a clipping plane to cut across it, or perhaps using a shader to modify the geometry or change to a transparent texture for the cutout.  The details will depend on the effect you are trying to achieve and your level of skill with the tools.

For more advanced shapes you might use tools for constructive geometry to subtract sections, assuming tools are available to you.

14 hours ago, YixunLiu said:

This slice is an image cutting through the sphere, like ultrasound image. 

If you have volumetric data for the sphere and you want to show a proper cutout, there is a bunch of research on how to do that efficiently. The IEEE VIS conference has twenty years of research papers that can help as a starting point. 

Thank you so much for your valuable comments.

What I want to achieve is like the attached picture Hole.png, which I capture a frame from the Youtube.

As you can see from the picture, a virtual hole is added on the real wall to show the virtual sky. 

To reach this effect, I think I need to

1. Obtain the wall surface mesh by using some depth camera and then cut a hole in the mesh

2. Render the cut mesh to get depth value

3. Render the sky image with depth test enabled

Am I right?

Any suggestions are really appreciated!

Best,

YL

Oh, that was all you wanted. Here I was imagining  some kind of system where you filled a object with voxel like slices.

Your shader should have some kind of "Discard" command that will simply not render that fragment. You can use that with some kind of map to define what pixels shouldn't be drawn.

It's often used for those little bullet holes. If you want I could show you a example with Unreal or Unity.

This will be able to create a visual hole in any object. It won't have the depth filling and won't adjust collisions. If you are using rays for firing a gun it will be possible to fire trough a hole made with a shader like this. This is how a game like Counter-Strike would do it.

If your game is using realistic ballistics. Like in Rainbow Six Siege, you need to break the mesh and update the collisions.

That is to say it isn't some kind of cheap effect. From the jagged edges and the z depth you can see it's a mesh.

The important part is to use some kind of LOD or Octree so that only the part that is hit updates. Updating the collision for the full wall would slow the game to a crawl.

Thanks Scouting!

My understanding based on what you said is there are two methods. 

The first method is to use pixel shade to render the sky, but just render part of the sky based on a map.

This method looks easy to be implemented because it does not require the mesh of the surface.

For this method, does it looks real or provide enough depth information? 

It is better if you can send me an example. But I am not familiar with Unity and Unreal. I want to use DirectX11 to implement it.

The second method based on what you said is to build a mesh for the wall surface and then update the mesh.

It looks this method can produce more realistic result, right?

Do you know where can I get some examples of  similar implementation using DirectX?

The following is the link for this video:

https://hololens.reality.news/news/video-holelenz-adds-magic-windows-hololens-gives-portals-new-worlds-0176281/

Thanks.

What is showed in the video you linked is just sprites of holes rendered over the render result. There is no real holes or objects with holes in them.

Just use directX to render your sphere, then render the sprite over it. In screen space.

Here the concept is shown:

The problem is  I do not have the sphere model. What I have is the slice and the photo of the sphere taken by a real camera.

The margin of the hole looks very real. Is it produced by using shadow?