Questions arose in the perspective thread I stated, that maybe are worth a separate thread. To rephrase the main question. Can a geometrically corrected picure (in post or with a shift lens), be closer to what we experience with our eyesight, compared to what the camera and lens record on a sensor.
Here is a badly key stoned shot, corrected in post. The first shot is geometrically correct. Which is closer to what you apparently see?
I think most of us agree that our visual experience will seem similar to the second shot. Why?
A reply in the previous thread spoke about how young people find portraits taken with a mid telephoto strange, as they are used to the wide angle rendering of the cell phone selfi. How much does general culture and visual literacy count in a viewers appreciation of a picture?
I think everybody will agree that the perspective effect in the picure below is exaggerated and unreal. It was made with a 15mm (FF) lens. What focal length (FF) will render a more "realistic" rendering?
Can the added drama of the exaggerated perspective in the shot above, give the viewer a better sensation of actually being in the building?
Yes, the exaggerated perspective effect is unnatural. The exaggerated arches in the top right and left are distracting eye-magnets for me tending to lead my eyes out of the image.
A more realistic display of the perspective I would expect to see would give me a much stronger sensation of actually being there.
This is a pretty cool presentation. But I would call it the extreme implementation of the panoramic photograph. I would imagine that a lot of software correction is applied to the end result.
The move from the printed photograph to the computer screen as the medium where we look at images, has opened up a lot of possibilities.
When viewed from a comfortable viewing distance (say twice the length of the diagonal), the first image shows more obvious perspective distortion than the second. However, I think that is not the real reason that I prefer the second image.
Suppose I view each image from its centre of perspective. In that case, both images look very realistic to me with no signs of perspective distortion.
Nevertheless, I still prefer the second image. I think it is because the second image is much closer to the real shape of the building. If it was enlarged to life size, it could be pasted on the front of the building and the building would look much the same. The is not true of the first image, which shows obvious foreshortening because it was taken at an angle to the facade of the building rather than perpendicular to it.
Similar, yes, but definitely not the same because I reckon that the shooting angle (from the near surface) and (for buildings) the shooting distance are significant factors. Nothing new about the shooting angle - the common lore being to avoid key-stoning by holding the camera level, i.e. cos(90) = no correction required.
One would certainly not "correct" the classic railroad lines shot.
To me at first glance, the corrected image almost looks reverse-key-stoned! Indeed, I prefer not to correct a key-stoned image all the way.
When I process photos, I always go for "pleasing" as opposed to "realistic". First of all, by the time I process the photo, I've long since forgotten how the scene actually looked. Secondly, my idea of "pleasing" is often (but not always) to make the photo look "realistic".
On a couple of occasions, I have paid close attention to the color and contrast of a scene as my eyes recorded it and my brain processed it, and thought, "That would not be a good photo as-is, but I could process it to be good." And the thing is, the scene did look good to me -- it was only when I concentrated on the specifics that I found it lacking and understood that it would need to be processed in a way that was "pleasing" rather than "realistic".
So, thinking about this more, I came to realize that when I look at a photo, it takes up a very small portion of my field of view. It's also 2D. The fact that the scene takes up the entirety of my field of view, is 3D, and my eyes dart around the scene in real time focusing all over it creating a 3D map of the image, makes for something that is radically different than a small 2D photo. And by "small", I mean even a photo that is displayed fully on my 32 inch monitor with me viewing from 30 inches away to a 24x36 inch print on my wall.
Now, if I were doing video for a domed IMAX theater with 8K (or higher) resolution, then, perhaps, the processing wouldn't matter as much. When one is not fully immersed in the photo, it is art based on reality -- it is not realistic in and of itself. Even when one is fully immersed, it's still 2D. Maybe VR will reach a point in the relatively near future where you will not be able to tell that you even have a headset on, and then the "rules" for processing will change.
What we "see" is the image interpreted by our brain and that is the reason why at times what we think we saw and what we did is not the same thing.
The main reason why people become convinced they have seen ghosts, UFOs , Big Foot and so on. Most often they really do believe that is was what they saw....
Along these lines..."But in fact the brain is deciding what you’re going to perceive, and it’s processing all of the information and then it’s determining what’s the best interpretation.” www.psychologicalscience.org/news/releases/your-brain-sees-things-even-when-you-dont.html
A good introduction to how we perceive the world around us is Zakia's "Perception and Imaging". It is a very thorough investigation of how our eyes and brain process visual information.
Our eyes move around and process small areas of a scene or picture. How they move around a picture can vary according to what we are looking for, or told to look for. Here is just one scan I made from the book, that examines how we look at a picture.
It is a pity that modern books on digital photography do not include a good explanation of perspective as in some of those older books, such as the Manual of Photography by Jacobson et al., which I quoted from.
Pretty true of a lot of text books. My 1974 edition of the Steel Designers Manual, is the one I reach for when something unusual crops up. This was the Bible of structural steelwork design, later editions dropped most of the hand calculation methods, and my more recent editions are gathering dust,
So far as I can see, modern books on photography don't include a correct explanation of any of teh basics of photography. I'd really like to find one that does.
I think it applies to every aspect of photography, if you wish to dig a little deeper past the basics.
Principles of composition in photography by Feininger, Andreas published in 1972 is now long out of print. No other book about composition comes close to this slim volume. The tired old S curve and Rule of Thirds, is still the staple of books on this subject. Has no author bothered to study how the eye really looks at a picture, since Feininger mentioned studies of how we see and debunked all those "rules".
As for the technicalities. Well film sensitometry is not too difficult to explain and a knowlege of the materials you were using was much more important to photographers using film. Digital capture is much more difficult to explain clearly and concisely.
But I have found for anything I want to study seriously, two or three text books must be used in tandem.
... or maybe more than two or three - so as to avoid common myths like "ISO adjusts the sensor sensitivity" found in so many descriptions of the infamous Exposure Triangle ...
However, the unique point about perspective is that it has been well understood for a very long time and used to be explained correctly, but no longer is.
Other aspects of modern photography that are not properly explained in recent books are usually aspects of digital photography and you would not expect to find those topics in older books about film photography. The topics themselves are relatively new.
This is human cognitive function. Basically our eyesight has developed to be of the most help to us. Part of that is to present a world that remains consistent as we walk through it, and part of that is that human cognitive function corrects vertical key-stoning as it's quite distracting when you walk through a 3D space on a single plane. If you work out the maths of linear geometry you will find that it presents a world where shape distorts far more than what we as humans experience as we move through it.
We do not see as a series of "stills" defined by ray tracing but by building up a picture through scanning the scene as we move around it, add into that mix the sheer level of data processing and you will finfd that some things are actually from memory and others from 1 second old data when you were standing a little further back. this gives us the ability to build a picture that's far more of a "global overview" than a viewpoint from a single point in space (retinal image) would allow. This is constantly updated and persented so we can act on it with absolute confidence, so we believe it is the absolute truth.
You do not see the world as described by linear geometry though human eyes, cognitive function prevents it, therefore you cannot predict what we see using linear geometry, only what the camera sees, or use it as an approximate framework. There are a few inconsistencies between the world of linear geometry and the world as we understand it.
And that in the real world relative shapes and distances constantly change with position, in an image they are fixed and remain fixed and constant at all viewing distances. For an eye that is tuned to make sense of the fluid nature of the real 3D world this is quite significant.
We seek to make sense and understand, so it is a temptation to try to describe phenomena by the metrics we understand, fit it into the framework that we know and are familiar with, then we know and are familiar with the phenomena. So if things like vertical key stoning and telephoto compression are entirely a function of the image then they can be explained entirely by linear geometry.
But there is a massive problem with this:
It requires us to see that geometry correctly. As soon as you allow just a 1% distortion of that geometry through human cognitive function then you must allow for that and so it therefore cannot be entirely a function of linear geometry.
If we must see the perspective correctly in images then we must also allow that we see it correctly in the real 3D world. And if this is the case then why does it look odd in an image? If you tilt your head looking at buildings you would see exactly the same key stoning as the camera, you have previous memory of it “in that context” and so it would be recognised as correct and not look odd.
If we see the perspective in the image correctly, as we must if these things are a function of geometry alone, then why don’t we recognise the absolute truth in that geometry?
Instead we do something different, we choose to interpret that geometry relative to our viewing position of the image. And that would require us to use our memory of the sizes of recognised objects and angle subtended, but not our memory of the correct geometry in context.
When we view an image out of context, outside the centre of perspective, and it looks odd it is suggesting that we have no direct memory of that particular perspective. If we theorise that the one place we do not see linear perspective correctly is in the real world, and in images viewed from the centre of perspective, then we could say that the reason the correct perspective looks odd out of context is because it is showing us the differences between linear geometry and human vision in the real world in context.
