Hi Bruno On Nov 24, 6:28 pm, Bruno Postle <[EMAIL PROTECTED]> wrote: > On Mon 24-Nov-2008 at 17:24 +0000, [EMAIL PROTECTED] wrote: > > > > >I know I have not delved into how this stuff works as i should. Could you > >explain conventional three-point perspective ? > > A building like the one in Tom's Panini has lots of parallel edges, > and generally with a rectangular building there are three different > directions at right angles to each other. > > Most early perspectives are drawn with one of these 'dimensions' as > lines radiating from a single 'vanishing point' and the other two > 'dimensions' parallel to the picture frame: > > >> >>http://www.metmuseum.org/toah/hd/frac/ho_52.63.2.htm# > > If you tilt the 'camera' away from parallel you get three vanishing > points (in this case two of them are outside the frame): > > http://www.flickr.com/photos/[EMAIL PROTECTED]/3056638979/ > > ..but I don't think this is what Tom is referring to. Here Seb has > take a spherical panorama and deconstructed it into six one-point > perspectives: > > http://www.flickr.com/photos/sbprzd/2322158588 > > This is a very neat way of displaying a panorama and it would be > great to be able to interactively place the edges in one of these > views within hugin. >
Thanks for showing us these examples. Would agree that your tilted Panini does not look as "right" as the original? A lot of that is just because slanting columns aren't as comfortable as vertical ones; but I think part of it is because your change reveals that the original doesn't really have the single point perspective it appears to do. Try the same thing with a rectilinear photo. The "cuboid" is not technically a multi-point perspective, because all 6 faces are viewed from the same point, but it is certainly a multiple perspective, as each face has a different projection axis. Presented flat, it can't be seen as anything other than 6 separate pictures, though I suppose it was derived from a seamless spherical panorama. In a pano viewer you could see the same views, and a continuum of others, as you swing the rectilinear projection axis around. But imagine a viewer that knew where the walls were, so it could generate views of the inside of the "folded cuboid". To a first approximation, that could show you how the room looked from places other than the camera location. Refining the geometrical model of the room leads toward the kind of 3D virtual reality where you can wander around among the objects, and only the the surface "textures" that make things look realistic come from photos. That may be the ultimate "multi-point perspective image" -- and clearly you need photos taken from lots of places to build one. But sticking to a single, flat image, how far can you really go in combining viewpoints? One answer is "as far as you like, as long as your trajectory is smooth". I'm thinking of strip panoramas, that can go on for miles (e.g. the banks of the Danube project), and "rollouts" made by moving the camera around an object. These images are often made with slit cameras, real or synthetic, but can also be stitched together from separate photos. Rollouts always look odd, but some strip panoramas are perfectly natural looking. You could argue that such a strip pano approximates a cropped telephoto view, or in the limit, an orthographic image -- a single point of view, very far from the subject. But what about curved trajectories, walking around corners, etc? I've seen very convincing looking multipoint panos taken from such trajectories. My point is just that the limits on what can be rendered "realistically" in a flat photo are a good deal wider than one might at first suppose. For pictures like the Paninis, you might only need to move the camera a few meters between shots, taking care to match the places where the seams will be hidden from shot to shot. Then some appropriate warping and blending, and Presto! an 18th century masterpiece! Or so I like to dream. Regards, Tom --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "hugin and other free panoramic software" group. A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ To post to this group, send email to hugin-ptx@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/hugin-ptx -~----------~----~----~----~------~----~------~--~---
