just wondering if we can move a particle emitting object around a video clip in Blender without using a plane with video texture. Please find a sample image below. You may also find this video for a better idea of what I am trying to achieve.
https://www.youtube.com/watch?v=fxAAr8w4VnA
You didn't ask for it, but what I know all comp. apps works in that way, with the same amount of "deterioration" quality of a video footage - all apps operates with a video texture on a plane in 3D scene.
In short – If you connect a Texture node (with video sequence) directly into output (without shader) than colour information stay the same.
In Color Management switch from Filmic to Standard.
Matte tools are primitive for this technique, so sorry for dirty way I found on the way (probably exists better), or you should separate foreground in compositor properly, render it with alpha and use as image sequence, but that's another topic.
Example with particles. Green screen is taken from internet, easily find by name (Green Screen Sample | Panasonic HMC150).
Example with Smoke Simulation. GIF compression is too much here, I will try post regular render later.
Another way (without Plane in 3D scene) is Composite - render Smoke separately with faked Z pass in Render Layer (black&white gradient as smoke material orientated in Camera Z direction), because blender doesn't handle volumetric depth and even if you want to emit smoke like a particles (plane with texture) Z pass does not handle transparency either.
Smoke for this version you can render from blend file above, but not as Render output, but enable Composite first (and delete or exclude plane from rendering). If you switch layout to Compositing you will see two Render Layers set in file - one is regular render, second with faked Z pass.
These images has to be saved as OpenEXR_Multilayer (32-bit) file format, so you would be able to get enough depth data to composite with original footage. I expected much worse result, but for smoke it worked quite well :) Smoke was rendered only with 6 samples = noisy, but expect some noise in composite anyway. These even smooth anti-aliased pixels will be used as depth data, which means - any pixel represents position in - depth. In this level there is not any anti-aliasing (if I'm right).
Result (also because of gif commpresion) is not exact, I probably changed some parameter for smoke material due testing mishmash.
For compositing is essential Z Combine node that is responsible for mixing depth data. Because video doesn't have any depth info and we need to define some position in depth for original footage, it's enough to set some grey value for Z input. If you try to move slider to white or black you will see how smoke appears in front or behinde the actor.
