Thanks for the image, julopez.
Please have a look at the file below:
https://www.amazon.com/clouddrive/share/MA6fWMgscSoGFBHOG3BIRQ2qlFu6Hmn4XNmsObBOczO
I have used the Camera-Calibrator, and with the camera placed from where the image was shot, I created a Camera Projection.
https://help.maxon.net/us/index.html#TCAMERAMAPPING
If you use the still-camera [11_UHD] provided in the project, You can pretty much place anything in the scene, and it should be OK.
File 01, is more an exploration in CinemaScope.
The file provide allows for a quarter of HD, 960x540, it is 8bit/channel. Typically that leads to banding in the sky, with the tiniest color corrections. Hence I created an alpha channel and replaced it with a Sky object. This object is roughly timed to the shadow director of the scene.
If you use file 01 and the animation exploration camera, you can see why some parts are working and others not at all. Like the two cars in the right lower corner. The Van on the left has a geometry (roughly), and it allows for some camera movement.
The image itself has gotten some treatment in Ps. If you work mostly with the camera movement, you will see clearly why these changes were needed, if you drag the original jpg onto the “main object” I hope my little screen capture can showcase that.
Screen-capture
https://www.amazon.com/clouddrive/share/mJqOE7rbRM1B82WCxy3WSf6Ci7sVeyfkXGFBkLOyKAh
A few tips:
If you shoot for camera projection, shoot many images from the same position, zero change of the camera. Anything that moves, with cover parts of the scene. With many shots, you can get the non-moving parts.
If you like to move the camera in C4D a little bit, take the images from the real location very close to where the camera move will end. This is the time when the audience had enough time to understand the scene. If you had shot for the beginning of the camera move, the distortion would be more noticeable one the end.
Take some shots with a polarization filter, to get the reflections out of the windows. It might even help pull a mask for the windows, so they can reflect the data in c4d, and the static reflections (which you would have otherwise) do not give the effect away.
If you shoot for HDRI, the rule of thumb is per stop one shoot if you use jpg, two stops with RAW. The Range should be from entirely dark to completely white in the image. One can’t get more information from a scene.
The image must be 1.5 times larger than the render settings while the camera is closest to the picture. Here the crop that you can see of the image counts only.
If your project aims for 4K, and you move the camera in, get used to stitching (Panorama) the image. HDR and Panorama might lead to quite a few exposures. I often have days where I end up with 60GB of data for a single image.
If you like to have the windows as reflections working in the scene, you need material for this. The camera needs to be placed where the camera seems to be in the window reflection. Which often causes problems, then stitching and camera projection is often your only option.
As often mentioned, shoot with a gray card. While 2/3 pointing to the primary light source, and 1/3 pointing to the lens.
Avoid automatic in the camera, except for the HDR bracketing series. Here use time only, not aperture or any form of ND filers.
Check for anything that changes in the location.
This is a shortlist, anything else is the links I have provided above.
My best wishes