Hello. I want to share with you my workflow to produce rendering for exterior architecture. In particular the scene management, in order to be as fast as possible to produce final images. I’d like to know how is yours and a way, if possible, to improve mine.
Usually I need not more than 5,6 cameras, so I create 5 folders where I insert all the different variations. This variations include the sun ,that has to be in a different position for every shot, skylight, skyatmosphere, if I need to have stronger or lighter shadows, and heavy objects, cars and vegetation. I use nanite as much as possible, but for vegetation where I need two sides or alpha, is not possible, so my scene can be really heavy, so I prefer to hide everything Im not working on. When I have to produce the final renderings I create a certain number of level sequences based of the number of cameras. I hide actors in game in all the variations folders except the one I need to produce output. I add the specific sequence in the movie render queue and I make start the rendering. When it’s complete I do the same operation for every camera. In my worfklow I think the worst part is waiting initialization for every shot. It is very slow, at least 3,4 time slower than calculating to produce the image itself. Should be nice to have the initializazion start just one time and not for every camera, but I can’t add all the levels in the render queue because I need to hide and unhide actors in game manually for every variation. A way to manage variations, like twinmotion does, would be appreciate 
I see that the variation manager exist, but I think works good for simple variations, like for configurators.