Adding extra depth to conventional stereo.

Armed only with ignorance and enthusiasm, am experimenting with adding extra depth to conventional stereos.

Disheartened with hyperstereo's lack of intelligible foregrounds, am now playing with a method which adds extra depth to conventionals or, vice versa, adds foregrounds to hyperstetreos giving scale clues and removing 'model village' look.

Xeyed pairs below have extra depth versions on top and conventional versions underneath.

Can you see any difference? Is it worth persuing?

The technique used did not give good results.

Hi, Tony,

That’s an inter3sting series!

Here’s my take on it (Collectively, excep for the flat, round stones in water which i found difficult to perceive /evaluate compared with all the rest).

The depth distinctions, although definitely discernible from foreground to distant background, appear  distinctly magnified in the far  middleground - to the interesting extent of imparting a sense of elevation of the terrain in that middleground region.

I have encountered something of the same effect when using a counterposed technique of re-joining sky and terrain in merged bizonal perspective images.

this is all “for what it’s worth”, of course, given the individual nature of illusory perceptions.

I would certainly like to see more as you further develop your technique!

Best regards,

Dave Graham

uuglypher wrote:

Hi, Tony,

That’s an inter3sting series!

Here’s my take on it (Collectively, excep for the flat, round stones in water which i found difficult to perceive /evaluate compared with all the rest).

The depth distinctions, although definitely discernible from foreground to distant background, appear distinctly magnified in the far middleground - to the interesting extent of imparting a sense of elevation of the terrain in that middleground region.

I have encountered something of the same effect when using a counterposed technique of re-joining sky and terrain in merged bizonal perspective images.

this is all “for what it’s worth”, of course, given the individual nature of illusory perceptions.

I would certainly like to see more as you further develop your technique!

Best regards,

Dave Graham

Thank you for taking the time to examine them.

I have noticed that elevation you refer to which was worse at first but I thought it was acceptable now.

Evidently, it needs more work. Your feedback is welcome and useful.

tony brown wrote:

uuglypher wrote:

Hi, Tony,

That’s an inter3sting series!

Here’s my take on it (Collectively, excep for the flat, round stones in water which i found difficult to perceive /evaluate compared with all the rest).

The depth distinctions, although definitely discernible from foreground to distant background, appear distinctly magnified in the far middleground - to the interesting extent of imparting a sense of elevation of the terrain in that middleground region.

I have encountered something of the same effect when using a counterposed technique of re-joining sky and terrain in merged bizonal perspective images.

this is all “for what it’s worth”, of course, given the individual nature of illusory perceptions.

I would certainly like to see more as you further develop your technique!

Best regards,

Dave Graham

Thank you for taking the time to examine them.

I have noticed that elevation you refer to which was worse at first but I thought it was acceptable now.

Evidently, it needs more work. Your feedback is welcome and useful.

Here's a re-work of the first image pair, again with lower image the original conventional stereo and top one is the rework.

tony brown wrote:

tony brown wrote:

uuglypher wrote:

Hi, Tony,

That’s an inter3sting series!

Here’s my take on it (Collectively, excep for the flat, round stones in water which i found difficult to perceive /evaluate compared with all the rest).

The depth distinctions, although definitely discernible from foreground to distant background, appear distinctly magnified in the far middleground - to the interesting extent of imparting a sense of elevation of the terrain in that middleground region.

I have encountered something of the same effect when using a counterposed technique of re-joining sky and terrain in merged bizonal perspective images.

this is all “for what it’s worth”, of course, given the individual nature of illusory perceptions.

I would certainly like to see more as you further develop your technique!

Best regards,

Dave Graham

Thank you for taking the time to examine them.

I have noticed that elevation you refer to which was worse at first but I thought it was acceptable now.

Evidently, it needs more work. Your feedback is welcome and useful.

Here's a re-work of the first image pair, again with lower image the original conventional stereo and top one is the rework.

Hi, Tony,

Here’s my take when comparing the three steps:

conventional S3D, your first re-work, and the latest rework)  I see each as composed of two regions: the foreground and middle ground ending at the near side of our view of the water, and background all beyond that.:

your first rework produced an apparent precipitous drop-off of background water level beyond the middle ground, while in your latest re-work the drop-off seems less severe- less  precipitous.

The delineation between middleground and background does, however, remain distinct.

How does that jibe with your intent?

Best regards,

Dave

uuglypher wrote:

tony brown wrote:

Here's a re-work of the first image pair, again with lower image the original conventional stereo and top one is the rework.

Hi, Tony,

Here’s my take when comparing the three steps:

conventional S3D, your first re-work, and the latest rework) I see each as composed of two regions: the foreground and middle ground ending at the near side of our view of the water, and background all beyond that.:

your first rework produced an apparent precipitous drop-off of background water level beyond the middle ground, while in your latest re-work the drop-off seems less severe- less precipitous.

The delineation between middleground and background does, however, remain distinct.

How does that jibe with your intent?

Best regards,

Dave

Hi Dave,

What you observe is what I have tried to reduce. BTW I searched for 'bizonal perspective' without sucess but can imagine what it is.

I started stereo only a year and a half ago and used B&W film in 1950s cameras to make images whuch were mostly landscapes and old architecture. Finding the conventional spacing could not distinguish depth much beyond 15 yards, I then used Cha-Cha technique with perhaps 1 to 2 metres separation between captures. The depth continued as far as I required into the image but the foregrounds were irreconcilable. Not only that but I couldn't have animals, birds or moving people because of the interval between exposures.

Thus I assembled two Fujifilm XP60 cameras on a bar with the lenses ~100mm apart and was able to download images and recharge their batteries without removing the cameras from their mountings. Using 'finger synch' I was able to get exposures to within about 1/4 second and often better which allowed more motion. However, the depth at distance was gone again.

So I use the Fuji pair to take a conventional stereo and now follow with ONE additional capture, usually to the right, at 1 to 2 meters as before, to capture background depth separation. Thus I produce two stereo images with the left image common to both pictures and the right two are both used to create stereos with the common left one.

These are assembled as Photoshop layers, lined up in Anaglyph mode (by switching off the RED channels in both right hand images) to get matching lateral displacements at the horizontal joining area and the right hand, wide image masked to avoid interfering with the foreground. The mask can be quite unsharp and the viewer's eye does most of the work. I attempt to get the lateral displacement of the wide (rear) view the same as that at the rear of the front (foreground) view as it is at the joining area. That can be achieved by moving the rear view horizontally with respect to the left hand image which remains as it was captured without masking or alteration.

Having got the positioning right as above, I then restore the RED channels, switched off to display the Anaglyph for working, and then move the layers to display the full colour, Side by Side or Xeyed versions you see. I enjoyed Anaglyphs when only taking B&W images but feel so much improvement with SxS in colour using Loreo hand held prismatic viewers. Colour Anaglyphs, as in the reworked image containing the red car, require use of DuBois treatment by which time the end result is far from the real life 'as you see it' 3D image.

I assume 'bizonal perspective' is a draughtmanship involving combining just such two images with different perspectives? Any link to same would be appreciated to give me a superficial insight.

It is a moot point whether my Tri-shot process adds extra depth to conventional stereo or enables rational foregrounds to be added to hyperstereos to provide scale and avoid the 'model village effect'. The price to pay is some visual discontinuity at the join which must be balanced with the extra depth information provided.

tony brown wrote:

uuglypher wrote:

tony brown wrote:

Here's a re-work of the first image pair, again with lower image the original conventional stereo and top one is the rework.

Hi, Tony,

Here’s my take when comparing the three steps:

conventional S3D, your first re-work, and the latest rework) I see each as composed of two regions: the foreground and middle ground ending at the near side of our view of the water, and background all beyond that.:

your first rework produced an apparent precipitous drop-off of background water level beyond the middle ground, while in your latest re-work the drop-off seems less severe- less precipitous.

The delineation between middleground and background does, however, remain distinct.

How does that jibe with your intent?

Best regards,

Dave

Hi Dave,

What you observe is what I have tried to reduce. BTW I searched for 'bizonal perspective' without sucess but can imagine what it is.

I started stereo only a year and a half ago and used B&W film in 1950s cameras to make images whuch were mostly landscapes and old architecture. Finding the conventional spacing could not distinguish depth much beyond 15 yards, I then used Cha-Cha technique with perhaps 1 to 2 metres separation between captures. The depth continued as far as I required into the image but the foregrounds were irreconcilable. Not only that but I couldn't have animals, birds or moving people because of the interval between exposures.

Thus I assembled two Fujifilm XP60 cameras on a bar with the lenses ~100mm apart and was able to download images and recharge their batteries without removing the cameras from their mountings. Using 'finger synch' I was able to get exposures to within about 1/4 second and often better which allowed more motion. However, the depth at distance was gone again.

So I use the Fuji pair to take a conventional stereo and now follow with ONE additional capture, usually to the right, at 1 to 2 meters as before, to capture background depth separation. Thus I produce two stereo images with the left image common to both pictures and the right two are both used to create stereos with the common left one.

These are assembled as Photoshop layers, lined up in Anaglyph mode (by switching off the RED channels in both right hand images) to get matching lateral displacements at the horizontal joining area and the right hand, wide image masked to avoid interfering with the foreground. The mask can be quite unsharp and the viewer's eye does most of the work. I attempt to get the lateral displacement of the wide (rear) view the same as that at the rear of the front (foreground) view as it is at the joining area. That can be achieved by moving the rear view horizontally with respect to the left hand image which remains as it was captured without masking or alteration.

Having got the positioning right as above, I then restore the RED channels, switched off to display the Anaglyph for working, and then move the layers to display the full colour, Side by Side or Xeyed versions you see. I enjoyed Anaglyphs when only taking B&W images but feel so much improvement with SxS in colour using Loreo hand held prismatic viewers. Colour Anaglyphs, as in the reworked image containing the red car, require use of DuBois treatment by which time the end result is far from the real life 'as you see it' 3D image.

I assume 'bizonal perspective' is a draughtmanship involving combining just such two images with different perspectives? Any link to same would be appreciated to give me a superficial insight.

It is a moot point whether my Tri-shot process adds extra depth to conventional stereo or enables rational foregrounds to be added to hyperstereos to provide scale and avoid the 'model village effect'. The price to pay is some visual discontinuity at the join which must be balanced with the extra depth information provided.

Hi, Tony,

I ought start by making clear that my interest in 3D is not via “stereo photography” but by means of 2D-to-3D conversion using disproportionate geometric transformation of a copy of an original 2D image (the left eye image) and transforming the copy to b3come the right eye image.

That said, I did read through your explanatory disquisition, then realized I’d better get a cup of coffee and give it another, more contemplative, read! Which I did, and I’m beginning to get the gist of it.

Basically, I think you and I are both less than totally happy with the results of traditional S3D photography. My major problem with it is the early extinction of rendition of the 3rd dimension leaving a 2D background looking like a painted stage backdrop.

For the past 8-10 years I’ve been working on techniques of 2D-to-3D conversion. By the way, “Bizonal Perspective” is a technique of differentially transforming the sky above the horizon and the terrain below the horizon so that when counterposed at the horizon, the result serves as the right eye image of the 3D image pair. But this brief summary was by way of explaining why I have little experience with the “two simultaneous image” sort of 3D imagery and why you won’t (yet) be able to find “bizonal perspective” in a literature search.

Here is an example of the technique as applied to a 2D image I made In AZ some years ago.

Anyway, You are in the right place to get lotsa help with S3D photography. I hope you will find help from those far more experienced in traditional “stereo” photography than I am.

Now, I’m back to another read of your explanation of your problem.

Best regards

Dave

uuglypher wrote:

tony brown wrote:

uuglypher wrote:

Hi, Tony,

......

Basically, I think you and I are both less than totally happy with the results of traditional S3D photography. My major problem with it is the early extinction of rendition of the 3rd dimension leaving a 2D background looking like a painted stage backdrop. ........

Best regards

Dave

Yes, I go along with that. We seem both to be trying to extend the depth effect somewhat. A little imperfection is nothing compared to what the viewer accepts with colour Anaglyphs.

BTW, I use cheap Loreo viewer for Side by Sides as here:-

https://www.3dworldshop.com/epages/544034815.sf/en_GB/?ObjectPath=/Shops/544034815/Products/4001

tony brown wrote:

The price to pay is some visual discontinuity....

A little imperfection is nothing compared to what the viewer accepts with colour Anaglyphs.

The problem with the presented techniques does not consist in "small discontinuities" but in observable and disturbing discontinuities if the images are viewed at a reasonable size, and if the resulting 3D images are viewed at large sizes, the discontinuities are very disturbing and the 3D perception is totally compromised.
Good results can be obtained only with a lot of work with very precise local distortions in the junction areas of several images with different depths to ensure a seemingly continuous passage between the different depths.
...Or with advanced techniques for 2D to 3D conversions.

But is fun to try and experiment with 3D images.

3D Gunner wrote:

tony brown wrote:

The price to pay is some visual discontinuity....

A little imperfection is nothing compared to what the viewer accepts with colour Anaglyphs.

The problem with the presented techniques does not consist in "small discontinuities" but in observable and disturbing discontinuities if the images are viewed at a reasonable size, and if the resulting 3D images are viewed at large sizes, the discontinuities are very disturbing and the 3D perception is totally compromised.
Good results can be obtained only with a lot of work with very precise local distortions in the junction areas of several images with different depths to ensure a seemingly continuous passage between the different depths.
...Or with advanced techniques for 2D to 3D conversions.

But is fun to try and experiment with 3D images.

Hi, Gunner,

you refer to “discontinuities (that are) very disturbing”  and that “the 3D perception is totally compromised”.  These indictments are quite subjective; can you be more specifically objective? To what specific features of the presented images are you specifically objecting?

Thanks,

Dave

I am referring to the steep transitions between different planes in the image, which should be continuous, as well as to the improper placement in space of some elements from the immediate area where the change of perspective begins.

This technique can be also applied vice versa, to compress the background perspective. For example in the case of macro or close-up images that have a background that is too far away, which bothers the eyes.

3D Gunner wrote:

tony brown wrote:

The price to pay is some visual discontinuity....

A little imperfection is nothing compared to what the viewer accepts with colour Anaglyphs.

The problem with the presented techniques does not consist in "small discontinuities" but in observable and disturbing discontinuities if the images are viewed at a reasonable size, and if the resulting 3D images are viewed at large sizes, the discontinuities are very disturbing and the 3D perception is totally compromised......

In which of my initial samples do you pecieve disturbing discontinuities? It is good to know which image is not an acceptible 3D to your eyes but constitues such disturbance to your brain as to prevent the conventional 3D illusion.

Then I can decide whether your comments are constructive or can be ignored (no offence intended). I find with landscapes that the very limited foreground depth provided by, say 65mm lens separation of the human eyes, plays no part in such views while Cha-Chas on a much longer baseline render the foreground unintelligible. Now the latter foreground is when I find 3D images very disturbing and 3D perception is totally compromised, the eyes/brain being quite unable to reconcile the two halves.

In all of them. You chose to insert the depth change at the water level, where theoretically it would be harder to observe and easier to correct, but it is visible in all images.

The image inside the locality shows 4 areas with distortions and processing artifacts. In addition, the separation of plans has become stronger in remote than in nearby areas, which is counterintuitive.

I hope you're not upset about my remarks.

3D Gunner wrote:

In all of them. You chose to insert the depth change at the water level, where theoretically it would be harder to observe and easier to correct, but it is visible in all images.

The image inside the locality shows 4 areas with distortions and processing artifacts. In addition, the separation of plans has become stronger in remote than in nearby areas, which is counterintuitive.

I hope you're not upset about my remarks.

Not at all. I welcome comments but don't intend to revert to convention rather than experiment further. 20 years of DPReview have taught me the variety of views expressed.

Now in my eighties, time is of the essence and waiting for perfection is not an option with a topic that is well over 150 years in progress. There is amazing archive stereo work taken during the American Civil War so development is hardly racing ahead, though not at a standstill.

Your intention in perfecting this technique is commendable and the results can be amazing.
I am not a native English speaker, so I am looking for expressions that are as easy to understand as possible, which can induce the idea of slight aggression in expression. Sorry about that.

3D Gunner wrote:

Your intention in perfecting this technique is commendable and the results can be amazing.
I am not a native English speaker, so I am looking for expressions that are as easy to understand as possible, which can induce the idea of slight aggression in expression. Sorry about that.

Your criticisms are welcome and why I posted here. I have altered the technique slightly and here is a rework of that particular example:-

As I said before, the result can be very good if you pay attention to details. Because I use a large PC monitor (55 "4k), the imperfections are easy to notice, and are even easier to notice if the images will have high resolution (which would require even more attention).

In this particular image, only two small imperfections remains to be corrected.
Good work!

rmexpress22 wrote:

I didn't read the other comments here but while I do see much more of a 3D effect on the first version, the different sections are more obvious and almost seem to look like cutouts. With the second version, which is the original, the foreground looks better but the background tends to look at bit flat.
I think they both have their uses and regardless, I think it's cool to see 2d images in 3d and I appreciate the exaggeration of the first version anyway.

You perceive them correctly. The first (top) version has the background exaggerated by using wide gap Cha-Cha or hyperstereo, hence the 'cut out' look. It is masked in during processing in Photoshop or similar. Only three images are captured; a conventional stereo being two and a single additional capture (not stereo) at some distance which is then paired with one of the former stereo pair. Usually, I use Stereo Photo Maker to create two Side by Side stereos from the three images above to be layered and masked in Photoshop.

While not a real view it is intended to add some depth effect to landscapes while still allowing properly perceived foregrounds. The degree to which the background is exaggerated lies with the photgrapher's decision on the separation distance of the 'Cha' shot. The 'Cha' shot is moved while editing to a) line up with the original stereo pair and b) then adjusted latterally to start the front of its entry to approximately the same Anaglyph displacement as the foreground stereo image has where it meets the new background.

The editing is done in Anaglyph mode but, after lining up, I convert back to Side x Side or X-eyed as I don't enjoy colour images in Anaglyph as much.