Getting the Right Depth in 3D Photography

written for the stereosite by David Kuntz, USA

A 3D pho­to­graph usu­al­ly con­sists of two images of the same sub­ject tak­en from dif­fer­ent view­points.  One of the most com­mon ques­tions amongst both expe­ri­enced and new stereo pho­tog­ra­phers is how to deter­mine the right dis­tance (typ­i­cal­ly called base­line, stereo base, cam­era sep­a­ra­tion or inter­ax­i­al dis­tance) between the left and right cam­era posi­tions that should be used when tak­ing the pho­tos.  The pho­to­graph illus­trates what base­line is; again, it’s just the dis­tance between the cam­era posi­tions where the left and right images were made. 

Why Does Baseline Matter?

What dif­fer­ence does base­line make? Who even cares?  Well, to see why it’s impor­tant, let’s look at a series of 3D pho­tographs of the same object made with pro­gres­sive­ly larg­er base­lines.  In the sequence of cube pho­tos, #1 starts with the small­est base­line, and the base­line is steadi­ly increased through pho­to #9.

Depend­ing upon how you’re view­ing these images (which are made for par­al­lel view­ing), you’ll prob­a­bly see almost no 3D in the first cou­ple images.  Then, some­where in the “medi­um base­lines” set, maybe at image #5, the cube will look fair­ly “nat­ur­al” and have a sat­is­fy­ing amount of depth.  In the last few images, the shape of the cube may start to look dis­tort­ed, and it may even be hard for you to view it in 3D. 

Don’t wor­ry about exact­ly which 3D of the cubes looks best to you – the take­away is that too lit­tle base­line gives you almost no 3D effect, and too much base­line makes the image hard to view, or dis­tort­ed look­ing. The right base­line gives you an image that has a sat­is­fy­ing amount of 3D, isn’t uncom­fort­able to view, and doesn’t dis­tort the object (assum­ing you want to avoid that). 

What is the Right Baseline?

So, how do you know what the right base­line is for a spe­cif­ic pho­to­graph that you want to take?  The answer is, “It depends.”  And it depends on a lot of things.  Some of these are relat­ed to the sub­ject mate­r­i­al you’re pho­tograph­ing. These include:

• Sub­ject size (espe­cial­ly sub­ject depth)
• Sub­ject dis­tance (from the camera)

The oth­er things that influ­ence what base­line will work best have to do with the way you’re going to repro­duce or view the image.  These include:

• View­ing screen or print size
• View­ing dis­tance (from your eye to the dis­play or print)
• View­er type (par­tic­u­lar­ly mag­ni­fi­ca­tion if you’re not just freeviewing)

In par­tic­u­lar, it’s impor­tant real­ize that the larg­er you’re going to repro­duce your 3D image, the less over­all depth it should have.  Sim­i­lar­ly, the far­ther back you are from the dis­play (or print), the less depth you should (typ­i­cal­ly) have.  So, an image meant to be viewed on a large screen TV should have less depth than one intend­ed for view­ing on a cell­phone display. 

All this may seem like an awful lot of fac­tors to con­sid­er.  And, exact­ly cal­cu­lat­ing or even esti­mat­ing the right base­line val­ue could be quite com­pli­cat­ed, espe­cial­ly since you may not actu­al­ly know all the infor­ma­tion you need. But, there are a cou­ple of ways to make it simpler. 

Practical Methods

If you’re tak­ing sequen­tial pho­tos (that is, using a sin­gle cam­era to take the left and right images one after the oth­er), then, you can just take sev­er­al pho­tos, grad­u­al­ly mov­ing the cam­era in the same direc­tion between each expo­sure. This is illus­trat­ed in the photo. 

After you’ve tak­en this series of pho­tos, you view var­i­ous com­bi­na­tions in 3D – for exam­ple, the left­most pho­to with the one tak­en imme­di­ate­ly to the right of it (#1 and #2), or the left­most pho­to with the pho­to tak­en a cou­ple of places far­ther to the right (#1 and #4).  Then, you just pick the pair of pho­tos that look most sat­is­fy­ing to you. 

That’s exact­ly what was done with the cube pho­tos.  It’s 10 pho­tos of the same object, in which the cam­era was moved slight­ly to the right between each expo­sure.  Then, nine 3D pairs were made.  In each case, the left­most orig­i­nal pho­to is on the left, and it’s then paired with ones on the right which were made at suc­ces­sive­ly larg­er and larg­er base­lines (#1 and #2, #1 and #3, #1 and #4…).  And again, you can just pick out the 3D pair that looks good to you. 

This may sound like a lot of work, but once you’ve had some expe­ri­ence you’ll get a feel for how much base­line works in a giv­en sit­u­a­tion.  Then, you’ll prob­a­bly find your­self tak­ing around five pho­tos of a sub­ject and get­ting one that works well when you start match­ing them up.

As I men­tioned before, a spe­cif­ic 3D pair may look per­fect on Insta­gram but might not be usable for a large screen pro­jec­tion. So, anoth­er ben­e­fit of shoot­ing a sequence of images is that it enables you to cre­ate mul­ti­ple 3D pairs hav­ing var­i­ous amounts of depth for dif­fer­ent pur­pos­es. And, even if you don’t want to do that right now, it’s worth sav­ing all your images in case you want to go back lat­er and pro­duce anoth­er 3D with a dif­fer­ent amount of depth. Per­haps your stereo pho­tos will be part of a pro­ject­ed show some­time in the future.

If you’re shoot­ing with a two-lensed 3D cam­era, like a Fuji W3, then your base­line is already set, and you can’t change it.  In this case, you con­trol the depth in your pho­to by posi­tion­ing your­self (and your cam­era) rel­a­tive to the sub­ject mat­ter.  In oth­er words, you might get clos­er to or far­ther from your main sub­ject, or choose to keep ele­ments that are very close or very far out of your composition. 

Specif­i­cal­ly, when shoot­ing with the Fuji W3 or W1, which have a stereo base of about 75 mm, arrange your pho­to so that the clos­est and far­thest objects in it were at dis­tances cor­re­spond­ing to the val­ues in the table.  This will reli­ably pro­duce images which are suit­able for large screen view­ing.  If you’re dis­play­ing your images on a small screen, you can stretch the depth range in your com­po­si­tion a lit­tle far­ther than this. 

Composition

It’s also impor­tant to real­ize that view­ing com­fort with a 3D pho­to depends on both total image depth, as well as the way that depth is dis­trib­uted through­out the image.  The next draw­ing illus­trates that.  It shows three series of cir­cles.  In each case, the total depth between the orange and pur­ple cir­cles is the same.  How­ev­er, for me per­son­al­ly, I find the top set, in which the orange and pur­ple cir­cles over­lap, slight­ly unpleas­ant to view.  My gaze keeps shift­ing between the orange and pur­ple cir­cle, and the jump in depth between them is a bit annoy­ing and dif­fi­cult for me to take in.

In the set of cir­cles in the cen­ter, the orange and pur­ple cir­cles still have the same rel­a­tive depth rela­tion­ship.  But, now there are three oth­er cir­cles in between them that act as step­ping stones for my eyes, so that my gaze doesn’t have to shift sud­den­ly from the fore­ground all the way to the back­ground.  I don’t find it at all unpleas­ant to view this set of circles.

In the bot­tom set, the total depth between the orange and pur­ple cir­cles is again the same as before.  But, now they’re spaced far­ther apart hor­i­zon­tal­ly.  I don’t have any trou­ble view­ing this set either, prob­a­bly because my eye has enough time to read­just as it moves over the large blank space in between them.

The les­son of this is that it’s not just the total depth that deter­mines how com­fort­able a stereo image is to view.  The way that depth is dis­trib­uted also plays a roll.

Let’s look at a real world exam­ple to see why this actu­al­ly makes a dif­fer­ence.  This pho­to, tak­en at the Roy­al Palace in Munich, has a huge amount of depth.  But, I have no trou­ble or dis­com­fort in view­ing this image, because my eye is led grad­u­al­ly from the fore­ground to the back­ground.  It’s just like the sec­ond set of cir­cles in my first image, where I had a bunch of step­ping stones to allow my eye to make the tran­si­tion from clos­est to far­thest object. 

The next pho­to of the flower also has a large amount of depth (too much!).  But, the real prob­lem is that it’s dis­trib­uted like in the very first set of cir­cles, with the clos­est object (the flower) right in front of the far­thest object (the wall).  Plus, there’s a dis­tract­ing orange cir­cle right next to the flower.  For me, this back­ground ele­ment com­petes with the fore­ground, and my eye keeps shift­ing between them with­out know­ing where to rest.  So, I find this less pleas­ant to view than the Munich Palace image. 

So, the les­son is, don’t just think about the total depth in your image when you’re com­pos­ing a pho­to.  Also, con­sid­er how that depth is dis­trib­uted and arranged.  And, try to avoid jar­ring tran­si­tions from very close to very dis­tant objects. 

Get Creative

The dis­cus­sion so far has most­ly focused on view­ing com­fort.  In oth­er words, cre­at­ing images that are easy to view in 3D.  But, the depth of an image is also a cre­ative tool.  Base­line can be employed in an artis­tic way, to pur­pose­ful­ly make things per­cep­tu­al­ly appear larg­er or small­er than they actu­al­ly are, or to dis­tort or exag­ger­ate depth.  For exam­ple, the next pho­to of the cas­tle was tak­en with a fair­ly large base­line.  Most peo­ple look­ing at this image will see the sub­ject mat­ter as “minia­tur­ized.”  Going in the oppo­site direc­tion, it’s pos­si­ble to take close-ups of flow­ers and insects at a very small base­line in order to make them appear gigantic.

Conclusion

Hope­ful­ly, you’re now aware that you need to con­trol the over­all amount of depth in your image to make it view­able.  And, that you have two pri­ma­ry tools at your con­trol to accom­plish this – the base­line you use when you take the pho­to­graph, and the amount of sub­ject depth you have in your com­po­si­tion.  Plus, your view­ing expe­ri­ence and com­fort depend upon how you’re view­ing the 3D image.  And, final­ly, that adjust­ing base­line is also a cre­ative tool you can use to achieve a spe­cif­ic per­cep­tu­al effect. 

It’s a lot to think about. But, don’t be par­a­lyzed, try­ing to cal­cu­late or fig­ure it all out before you take a 3D pho­to. The Stereo Sher­iff isn’t going to arrest you if your results aren’t per­fect.  Just become aware of what you’re doing.  In par­tic­u­lar, the depth in your scene and the base­line you’re using.  Could you move slight­ly or point the cam­era dif­fer­ent­ly to include or exclude some­thing close to you in your com­po­si­tion (like an over­hang­ing tree branch)?  These are the things you need to be aware of in 3D that aren’t an issue in flat photography. 

Just play around with it and take note of your results.  And, have fun!