Hi register-fault.
Thank you very much for the kind words. Yes, those techniques are certainly my sweet spot for a long time.
.
I assume this is the inside "Screen Box" that your project is targeting?
Example
https://www.outernet.com/news/enchanted-forest-the-melody-of-spirits
.
Audience Perspective:
I assume you have the precise measurements of each screen. I might write more than you need, but others might like to follow your project.
Those VR cubes, if passive, have a single sweet spot. (Active would be like an LED wall, capturing the camera's orientation and position to define the projected image's parralax representation.)
The passive version suggests that the middle point of the "room" and the average eye level (~5 feet or ~1.50cm, or the ideal user eye level) may align with this sweetspot.
.
That means the project setup requires a space, ensuring no object is closer than the screen.
.
Interactive Action: the variations based on the audience are, of course, not clear to me, and I assume nothing you can share.
.
With that being said, I would create a five-camera rig that matches each "wall" or "ceiling" dimension.
.
For the walls, if the side walls are equally wide, that would be a horizontal 90º field of view.
For the ceiling, subtract your defined ideal user eye level from the ceiling height, and the field of view will be defined from that point and the dimensions of the ceiling
.
You suggested a VR camera, which would give you an equirectangular image, which needs then "converted. This process creates pixel movements that are not exactly one pixel up, down, left, or right. Meaning: there is a substantial quality loss from this process alone.
.
The next step is the resolution of the screens, to be precise, it allows you to render just enough, while keeping things in great quality. Larger means scaling, smaller means upscaling with loss.
.
If the space is created with four equal walls, that would be a simple resolution decision, and the ceiling would need to match this.
The problem is that you need to explore are PostFX steps, like blur or blooming and such, this has more often than not, limits on the border, as the blur data after the border is not given, or a light is not "seen" and the resulting flare or bloom is different from screen wall to screen wall and Ceiling.
.
If that is a problem or effect in your project, padding each shot with a wider field of view and higher resolution should help. When all is done, crop the image and check that the "connecting" edges are matching.
How much padding is needed? If you want to set a blur radius as a maximum, that pixel amount serves as your padding.
.
Does this need to be on the lower part of the floor? I would suggest this, as it maintains a higher level of illusion. I'm not clear with two walls, are they half in height, that would be half resolution and a Offset, if you don't like to render fully.
.
When you are done, rendering, and any post-production is applied, you can create a texture out of each of the five renderings. If padded, create a larger plane so the padding is outside the "room". Now you can check with a VR 360ºx180º camera if they work together. If they do, crop them, and then test again with room-size walls.
.
Yes, the Japanese-style board, which we had often discussed in v3 of this forum, is sadly down. There, the curved wall presented challenges, as it works only well from a single sweet-spot, particularly when 3D elements extend far beyond the screen. From a different point of view, they look horrible.
The necessary steps involved rendering from the sweet spot, then using UV to project and bake the previously rendered content, resulting in a loss. The higher up those effects, the less resolution is a problem.
.
I hope that helps.
Please let me know if anything is unclear in my description, or if I've missed any points. I'm happy to look into it.
Thanks for the question!
Cheers
