Bobu wrote:
This picture was not shot with an alt-lens, but it has, in my opinion, a 3D-look. What do you think?
Perhaps a bit much on the perspective control. Exaggerated intentionally? There is perhaps a touch of 3-D evident there. Did you also do some exposure blending/HDR processing?
Lotusm50 wrote:
Perhaps a bit much on the perspective control. Exaggerated intentionally? There is perhaps a touch of 3-D evident there. Did you also do some exposure blending/HDR processing?
Thanks for the answer. Yes, this picture is a HDR. Could you please explain, what you meant with (too) much perspective control.
Bobu wrote:
Could you please explain, what you meant with (too) much perspective control.
He means that you have made the verticals parallel, which invariably makes buildings look top heavy and looming. The eye expects to see a little convergence of verticals with a building, just not as much as many wideangles produce. The trick with a TS lens is to get it to make the building look as natural as possible, not necessarily as orthographic as possible. In this case about 2/3 of the shift that you have actually applied would have produced a more satisfying image.
brainiac wrote......... In this case about 2/3 of the shift that you have actually applied would have produced a more satisfying image.
Hi brainiac,
More satisfying for what?
99% of the architects want it that way - it s kinda different language....
my guess is the 17 TSE as well - and looking at it, I'm glad that I went for the TSE24...
not for IQ-reasons, but the UW-look, it's ok when used orthogonally, but diagonally, I'm not impressed....
in my book the UW-look is not 3-D at all...
Using the TSE-24 (II) since about two monthes, I don't find that 3 D-look, as with my CYZ..
Straying a bit deeper into how perspective influences what we perceive ... what effect does making the verticals parallel (or nearly so) have on the horizontals & diagonals (skewed depth)? As I look at the picture, it seems to be that the building corner is perceived as a non-90 angle (i.e. acute) that angles backward.
I've never shot with tilt yet (probably needs to be a thread of its own), so I'm interested in it, but trying to stay in context of 3D-ish and how we perceive things, what are the trade-offs when adjusting verticals (via tilt or in post)?
BTW ... nice pic (sometimes gets lost in the quest for technical)
I understand that parallel verticals are sought, but from an architectural point of view it seems to me that perfect correction of the verticals can really hurt the visual impression of the building. The building looks too wide and too elongated at the top, and so seems very top-heavy. The building isn't rendered in a way that is sympathetic to its proportions. A degree of under-correction gives a more balanced projection which seems truer and straighter.
RustyBug wrote:
Straying a bit deeper into how perspective influences what we perceive ... what effect does making the verticals parallel (or nearly so) have on the horizontals & diagonals (skewed depth)? As I look at the picture, it seems to be that the building corner is perceived as a non-90 angle (i.e. acute) that angles backward.
I've never shot with tilt yet (probably needs to be a thread of its own), so I'm interested in it, but trying to stay in context of 3D-ish and how we perceive things, what are the trade-offs when adjusting verticals (via tilt or in post)?
BTW ... nice pic (sometimes gets lost in the quest for technical)
brainiac wrote:
I understand that parallel verticals are sought, but from an architectural point of view it seems to me that perfect correction of the verticals can really hurt the visual impression of the building. The building looks too wide and too elongated at the top, and so seems very top-heavy. The building isn't rendered in a way that is sympathetic to its proportions. A degree of under-correction gives a more balanced projection which seems truer and straighter.
To you. Architectural photographers and illustrators generally disagree with this - much more often than not. I personally don't have an opinion on this (I like whatever looks good to me - unquantified ) - just saying. They seek absolutely perfect verts across the entire image.
Bifurcator wrote:
To you. Architectural photographers and illustrators generally disagree with this - much more often than not. I personally don't have an opinion on this (I like whatever looks good to me - unquantified ) - just saying. They seek absolutely perfect verts across the entire image.
I agree with brainiac in this matter. My own brain objects when parallel verticals suggest an elevated viewpoint, while there are other cues in the same picture that suggest otherwise. The problem at hand is of course that we are looking at such pictures in the 'normal' way, with our eyes more or less at same height as the center of the image. Ideally one should reestablish the perspective adopted by the photographer and view the image from a low viewpoint. I just did, and the building looks much more natural. Similarly, pictures taken with telephoto and wideangle lenses should be viewed from long and short distances, respectively. If not, there will be an apparent 'perspective distortion'.
Biotar ... thanks. I think a bit more DOF would have helped, but I didn't realize that until I was well into post.
It's interesting to compare the same pic in this forum as the WA forum. For me, the one in this forum is much more to my liking, whereas the WA is size constrained and doesn't render as well (IMHO). Hence the issues of viewing distance & size come into play with regard to our perceptions of 3D.
Does anyone have any insight to why this is, or the physiology involved?
Ideally one should reestablish the perspective adopted by the photographer and view the image from a low viewpoint. I just did, and the building looks much more natural.
Interesting ... a bit less concerning if I imagine laying on a slope, looking back over my head.
brainiac wrote:
I understand that parallel verticals are sought, but from an architectural point of view it seems to me that perfect correction of the verticals can really hurt the visual impression of the building. The building looks too wide and too elongated at the top, and so seems very top-heavy. The building isn't rendered in a way that is sympathetic to its proportions. A degree of under-correction gives a more balanced projection which seems truer and straighter.
I agree about the top-heavy.
But it's not the question of more or less vertical shift, but the cam's position:
Perspective is done only by the viewer's position!
If taken from the center front (at the right), you would accept the shift and the vertical lines.
UW's used diagonally just create this kind of °image-distortion° (not lens distortion) - what biotar described as non-90 angle, exactly: the horizontal lines are in that situation a bigger problem than the verticals - nobody knows exactly the height of a house, but everbody knows how a 90 deg-angle looks like.
Even more - special composition: we see the ear... .. the main house's façe is looking away from us - and °running° away - outside the image.
BTW: When doing archi shots - I often cheat a tiny bit in favour of what you said, so nobody realises it, but the lines are still a bit less strong...
brainiac wrote:
I understand that parallel verticals are sought,
Sought and demanded. I do know what you mean and as a normal person and observer of art I agree. In fact I personally see no reason at all to even try and achieve 90 degree verts. Why? But in the architectural industry it's simply a must. If you cheat even a little you don't get the big bucks or the next job or etc. It's usually in the specifications and "cheating" is a violation of the agreement. The only reason I even know about this is when I was developing for a popular 3D CAD vender (photo-realistic rendering software) there were a great number of users who were extremely vocal about this point. I guess I can understand where it comes from - I mean, if you spent THAT much time on a drafting table you might become obsessed with 90 degree verts as well.
So within that venue if you intend to extend yourself commercially you better get it right and not cheat at all. If you're just goring for the art thing then it's like you say and it's more like whatever works. And of course you need to use the proper tools. Obviously a 20mm lens and perspective correction software is going to look whack (to an architect) more often than not.
The Minolta A2 is a great camera with a superb lens. I bought it to resist DSLR's until I took the plunge on a Kodak SLR/c. I don't regret buying the A2 (or the SLR/c) for a second. I have some excellent images from the A2.
Bifurcator wrote:
The Konica Minolta A2 with it's genre famous GT lens. It's a 6 year old bridge camera.
$800 new - Sells for about $200 today (used).
Bifurcator wrote:
Sought and demanded. I do know what you mean and as a normal person and observer of art I agree. In fact I personally see no reason at all to even try and achieve 90 degree verts. Why? But in the architectural industry it's simply a must. If you cheat even a little you don't get the big bucks or the next job or etc. It's usually in the specifications and "cheating" is a violation of the agreement. The only reason I even know about this is when I was developing for a popular 3D CAD vender (photo-realistic rendering software) there were a great number of users who were extremely vocal about this point. I guess I can understand where it comes from - I mean, if you spent THAT much time on a drafting table you might become obsessed with 90 degree verts as well.
So within that venue if you intend to extend yourself commercially you better get it right and not cheat at all. If you're just goring for the art thing then it's like you say and it's more like whatever works. And of course you need to use the proper tools. Obviously a 20mm lens and perspective correction software is going to look whack (to an architect) more often than not. ...Show more →
Surely the situation is different with CAD generated images, in that you can get true orthographic, i.e. no vertical stretching of the upper parts of the building due to the central image axis not being normal to the object plane and the film plane? From an architectural point of view, shooting upwards with no corrections and stretching the top part of the image laterally ('keystoning') will give you a much more natural and architecturally correct image than what we see here. When lens axis shift methods are used, the further off-axis a point is, the greater is the distortion of its shape. That's bad from an architectural point of view, and you can see why in this case. I sincerely doubt anyone uses software in CAD imaging to produce that kind of shift-distorted projection, even if the tools are available.
The best way to satisfy architectural requirements ought to be to shoot upwards and stretch the image into a trapezoid in order to achieve parallel verticals. That way you avoid the obvious top-heavy stretching distortion that we can see in this image.
brainiac wrote:
Surely the situation is different with CAD generated images, in that you can get true orthographic, i.e. no vertical stretching of the upper parts of the building due to the central image axis not being normal to the object plane and the film plane?
Yeah, but mostly only because the camera is placed virtually. Theoretically you could place the camera in space and photograph the building with an EXTREME focal length. In fact in 2005/6 or so that's what had to be done - almost. Thus all the requests for different kinds of cameras in the software.
From an architectural point of view, shooting upwards with no corrections and stretching the top part of the image laterally ('keystoning') will give you a much more natural and architecturally correct image than what we see here. When lens axis shift methods are used, the further off-axis a point is, the greater is the distortion of its shape. That's bad from an architectural point of view, and you can see why in this case. I sincerely doubt anyone uses software in CAD imaging to produce that kind of shift-distorted projection, even if the tools are available.
They do. And more. Here's some older screen shots from the same renderer - as the tools were being developed:
These are actual 3D models with interactive rendered shadows and lighting.
The camera even incorporates film grain.
But probably they don't actually do so for jobs in the architectural industry, just as art or whatever.
The best way to satisfy architectural requirements ought to be to shoot upwards and stretch the image into a trapezoid in order to achieve parallel verticals. That way you avoid the obvious top-heavy stretching distortion that we can see in this image.
I think I understand what you're saying. That sounds like correct thinking. Anyway just a few points here. CG is mostly only used when either the building doesn't exist yet (visualizations) or physical camera placement is too difficult or impossible. The rules for the two (from the architect's office) are the same concerning verts - when required - and they typically are. Now obviously a low profile real-estate agent won't have or know about these requirements. But from custom million dollar homes on up throughout skyscrapers this is the typical case - probably even for architectural magazine shots assuming it's not too consumer oriented. For photography it's just using the right lens, positioning the camera correctly, and using corrective software when needed. Photoshop comes with a pretty nice tool for these kinds of corrections.
) - just saying. They seek absolutely perfect verts across the entire image.
Like this:
