Hi table-state,
Let me summarize here, which is meant as a neutral recap.
I have no idea what eight cameras you are referring to. With each post,, more details become available. I haven't received a storyboard that shows me what you like, just a few images with lines where I have to guess which line is which.
If the images are the "same", where comes the problem? Since you said video, but the building is (so far, I understand) still images, not video
I have shared the FFD option shape, based on the initial version, which resembles a movement in Z while turbulence causes particles to move up and down. Later, it is a 3D movement with up and down movements.
The use of Nodes did not receive any feedback either. So I try to understand what you like to have. I placed the two images on top of each other and calibrated the camera to see the difference. Which you exclude, as well as the solution to the problem.
I have said it often before, Simulations are great, but art directing them in detail is not always simple. Here, the manual approaches come into play, perhaps even utilizing the results from the last frame, converting the Tracer into Splines, and then shaping the spline segments accordingly.
Then, animate it with MoSpline. Now you have 100% control.
The time spent handling a simulation as an Art Director or manually setting it up is a balancing question.
An alternative would be an object (or several) that is moving along the spline and needs to direct the path of motion.
While animating those objects in size or even deformation, you have control over them.
The file and its curve, along with the larger or smaller distances to each particle, are very much aligned with a perspective scaling. Hence, the idea is to find the camera position and angle, lens, etc., to stop fiddling with adjustments and have it working properly, meaning along perspective options. However, even though all indicate that this is the target, the information given suggests that it is based on something different.
As a side note, objects and their associated information that are needed at one point must be positioned above the object that requires it. Priority.
/summary end. All good, moving on.
What I have now, start, middle, and end, have all particles coming close, among those "close areas" getting wide again.
Why not use one or several objects as suggested below?
CV4_2025_drs_25_ANfl_01.c4d
Enjoy the rest of your Sunday.