last I rendered an exterior for a client and he is using Revit. He gave me some basic shots by himself and they had a special look. The front of the build was shot not from the middle of the building front, but the front lines was perfect horizontal. At the moment I can’t contact them and ask him, what kind of perspective he used.
Here an example. At the first image the Rhino 2-Point-Perspective. For the second one I turned my model, shot a 2-point-perspective view, turned the shot back and corrected the falling lines. I get a perfect straight view on the front of my simplified building like from my clients images.
Is this a 3-Point-Perspective where two point are infinite far away?
Maybe some one knows what special camera type can be used at Revit?
@pascal@nathanletwory Who of the McNeel team could develop such a camera type and I could ask for?
If you position the camera so it’s looking directly at the large flat surface you would get this kind of perspective. Since the camera is perfectly aligned with the surface, the surface itself has no perspective distortion. You can try doing it with OrientCameraToSrf, and then panning the view so the surface is no more in the centre of the view.
Right, this works, but if I need the view centered to the left “gangway” of the building, than I need it to render with a lot of wasted space. For example Enscape is limited to approx. 8000px, so I would lost some final resolution.
But this brings me to the next question for the next step. If Rhino would support special camera features like “keep horizontal lines” could render engines use this special view or is this limited by the render engines? Than, the new camera types wouldn’t help so much.
If you think about standing in front of the building with a camera, if you are directly face on this is what you would get. If you wanted to centre the building in your viewfinder you would turn to the right and you would no longer be seeing a single-point perspective, but a two point. Otherwise you would walk to the right and change your viewpoint to the centre of the façade.
Or you go home, develop the picture and crop it to an unnatural view. Sounds like that’s the effect you want?
I would assume this depends on the render engines. It is for example possible to do something similar in Blender, by changing vertical or horizontal camera offset, so that would be one way of reducing render time. But I’m sure there’s better solutions!
Right, 2-point corrected views are unnatural views but usual for architecture visualization. Shift lenses allow it at the real world. Good shift lenses allow a horizontal and vertical shift. The two-point camera of Rhino allow a 100% horizontal shift correction only.
the shorter lenses are used for architectural shots, to straighten up the horizon of course. the longer lenses are often used for product photography, here the shift function is used to align the depth of field to the upper surface to create a sharp image over the entire item, something which i believe has no digital pendant.
FWIW that is actually done with the tilt mechanism. It allows you to bring the plane of interest, the film (or sensor) plane and a plane perpendicular to the lens axis to a common axis, which ensures the whole plane of interest is in focus (while items further from the plane of interest are more blurred than they would be without the tilt.
There has been a lemming-like stampede in recent years to use reverse tilt when filming distant crowd scenes for TV, to diminish the depth of field and give the impression of looking at a scale model.
ups, thanks… I never use these terms, they sound so similar i can never distinguish them from each other. got any fancy examples for the model like clips?
ah, ok now i got it, the reverse tilt effect - intensifying the blur making it look like a small model shot with a low aperture. i forgot yes, has become popular maybe 6-7 years ago… not sure anymore i can remember one of my tutors babbling about it, i used it to make renderings look like models, fancy looking but artistically better suited for real life clips as you show imho.
All that shifting the lens on a real camera is doing is cropping the lenses’ field of view…
FWIW, shifts and tilts started with large format “view cameras” (4"x5", 8"x10", etc.) where either/both the front standard (lens holder) and the rear standard (film holder) allowed shifts, tilts or both off the lens axis. This allowed you to get more of the scene in focus and correct (or induce) perspective distortion. You did need a good lens with sufficient field coverage otherwise you would get vignetting. And the ability to “see” your images upside-down. Really fun stuff back in the day…
All that shifting the lens on a real camera is doing is cropping the lenses’ field of view…
Indeed, and it is the same for a renderer: you can render twice (or however much) as much as you need, and crop away what you don’t want – and you can therefore get it done somewhat efficiently in a renderer that supports only region render, and not shift, by doing exactly that, though it is worth noting that this may require the memory of the full resolution, depending how the renderer implements region render.
