I am doing a video-fusion project. I need to project a video in 3D space( surface of meshes).
The first try is creating custom shader in plugin and I got a my shader inherited from global shader following some tutorials. However it cannot help me to put each pixel in video scene to 3D world, I guess I need materialshader or meshmaterialshader to do that.
I found some materialshader/meshmaterialshader in engine code and also got my own one, but no idea how to use it. Please could anyone give me a simple demo code how to create and use a meshmaterialshader or materialshader?
Decal is my second try(mesh decal and dbuff decal). Just wondering, how to custom decal to make it rendering correctly on several surfaces?
Well, I am also considering postprocess, could I get camera matrix/mesh info in postprocess?
(I have algorithms(shader) to map scene pixel to 3D space, but no idea how to use it in UE4)
What im getting from your question is that you’re trying to make a projector kind of thing? If thats the case look into tri-planar projection. Tons of tutorials out there. Its mostly used for “ocean floor sunwaves” of sorts but theres many cases for it. Lots of tutorials out there
I do have render target to store the video scene, but my problem is I dnt know how to project the scene to 3D space.
I do have the algorithms, just try to make it works in UE4. My algorithms in my GIS directly use vertices, camera matrix, project matrix, etc. transform the video scene pixel into geographical coordinates and render it.
UE is updated, and the latest version is 4.25.4. Download and install it via the Epic Launcher, which also needs to be downloaded and installed…use forum credentials to log into it.
Is the video going to be projected onto a flat, 2D surface inside a 3D scene? or is it going to be projected onto 3D geometry? Can you be more specific about what the final result is intended to be??
thx for ur reply. I know there is 4.25 version. they began their project maybe 2 years ago, and it’s pretty sad that I cannot ask they to update it.
well, actually both. It is a GIS project, I need to project the security camera’s video in a virtual world where as the same place in real world. Maybe one flat is enough, but if it’s indoor corners should be covered.
For indoor example, your office. Pretend that we already build simple 3D meshes of the room without details or textures. Now I get the security camera video data in your office, then I wanna rebuild your office in the virtual world.
