Rule of Thirds?

On Thu, 4 Dec 2003 14:13:48 +0000 (UTC), Toke Eskildsen wrote:

----
If you adopt my point of view for a short while and accept the
premise that rules of thumb has value, then you aren't right. They
are rules of thumb and as such a based on what most people prefer.
That means they can be judged against each other (the Rule of Thirds
and the Golden Section at least, since they work on the same level).
----

Where do you disagree? If it's the part with "That means they can be
judged...", then please explain why one can't weight rules based on
statistics (eg. "what most people prefer") against each other.

Where I disagree is that by trying to decide between the rule of thirds and
the Golden Section, you are trying to introduce a level of precision where
no such precision is possible.


Since one rule says 33% and the other say 38% (rounded values), they
are not the same. Period.

Wrong, wrong, wrong. I've never met anyone who sensibly considered the rule
of thirds to mean *exactly* 33%. It could easily mean 28% or 38% or
whatever. This is why your whole premise is based on pseudo-precision.

It's highly likely that the rule of thirds evolved from the Golden Section
as a less-precise rule of thumb more appropriate to visual media (the
Golden Section having been developed in Ancient Greece for architecture).

You need to look again at the definition of a 'rule of thumb': it it not
something that is supposed to be used to an accuracy of 5%. No such
accuracy is possible - or even the slightest bit relevant - to taking
photographs.

If your point is that both rules are very approximate, then why did
the Golden Section arise at all? It's quite hard to calculate, yet
both architechts, photographers and painters have found it worthwhile
to use in their creations instead of the much easier "roughly a
third".

It arose in Ancient Greece. Give me examples of photographers who have
found it worthwhile instead of the rule of thirds.

Please explain what you mean by pseudo precision.

Insisting on a level of precision where non is ossible or relevant.

If I compose a picture according to Rule #1, the same picture
according to Rule #2 and I show them to 100 people, asking them to
choose the one they find more pleasing, will they be split in two
groups of roughly equal size?

Who can tell. The answer will most likely be entirely meaningless. Now
you're adding pseudo-science to pseudo-precision.


You need to stop worrying about the difference, because there is no
difference.

I understand that is your opinion.

Yep. And I've wasted enough time on this, so this is my final post on the
subject. If you really want to waste your time on something as trivial and
pointless as deciding between the rule of thirds and the Golden Section,
then go ahead. That's your problem. My advice is - choose one and go with
it. Flip a coin. If you choose the rule of thirds, I guarantee that
*no-one* will ever look at your pictures and say "Hmm, it's okay, but it
would have been better if you'd used the Golden Section".

This whole distinction between the two rules is false. But hey, it's your
life.

Steve wrote:

[...]

Where I disagree is that by trying to decide between the rule of
thirds and the Golden Section, you are trying to introduce a level
of precision where no such precision is possible.

You did not answer my question.

I presented my reasoning. You seem to disagree. I ask you what is wrong
with my reasoning and instead of pointing out my error, you repeat your
claim.

Since one rule says 33% and the other say 38% (rounded values),
they are not the same. Period.

Wrong, wrong, wrong. I've never met anyone who sensibly considered
the rule of thirds to mean *exactly* 33%.

Are you talking about the rule itself or applying the rule to a
specific picture?

My understanding is that most people regard the Rule of Thirds as
"Place x aproximately a third from the edge". The "Place x
aproximately" is the part about applying the rule and my oh my, it
_does_ say aproximately and not "exactly". The exact part of the
sentence is "a third from the edge", that's the abstract part, the
ideal or goal that the rule prescribe.

It could easily mean 28% or 38% or whatever.

Yes, when you apply it.

This is why your whole premise is based on pseudo-precision.

I presented my point about statistics, but you've refused to comment
directly on that.

It's highly likely that the rule of thirds evolved from the Golden
Section as a less-precise rule of thumb more appropriate to visual
media (the Golden Section having been developed in Ancient Greece
for architecture).

Ah, but this is very interesting. The Rule of Thirds might indeed be
more appropriate for photos. I would appreciate more pointers on this.

[Snip Why the Golden Section at all]

It arose in Ancient Greece. Give me examples of photographers who
have found it worthwhile instead of the rule of thirds.

Ask a random danish photographer. I asked in the group dk.fritid.foto
and they told me Golden Section and not Rule of Thirds. If you look
back in this thread, you'll see that I am curious as to why there's
this difference here. I invite you to verify my claim - questions in
english are quite welcome in the group.

It does not mean that they whip out a ruler and measure, but instead
that they roughly aim at the Golden Section instead of rougly aiming at
the 33%. If they think about the rules at all.

[Snip Perform statistics]

Who can tell. The answer will most likely be entirely meaningless.
Now you're adding pseudo-science to pseudo-precision.

Using statistics to examine statistical based rules is pseudo-science?

Yep. And I've wasted enough time on this, so this is my final post
on the subject.

You have wasted a lot of time yes. However, if you stopped repeating
your point in different ways and instead focused on answering my
arguments directly, a lot of energy would be saved.

My advice is - choose one and go with it.

I stated several posts ago that I'd do that.

If you choose the rule of thirds, I guarantee that *no-one* will ever
look at your pictures and say "Hmm, it's okay, but it would have been
better if you'd used the Golden Section".

I agree completely. You've said that before though (in a slightly
different way) and I also agreed then.

Toke Eskildsen apparently said:

try and answer me so
that we can move forward instead of just repeating ourselves.

You are a troll. Go out and take some pictures and figure it out.

Dean Van Praotl wrote:

Toke Eskildsen apparently said:

try and answer me so that we can move forward instead of just
repeating ourselves.

You are a troll.

Actually, the sentence that you quoted is quite the negative of
trolling. Trolls are not interested in getting the discussion to move
onwards.

The Jargon File has a definition, if anyone should be unfamiliar with
the term: http://jargon.watson-net.com/jargon.asp?w=troll

Troll is also a convenient name to call someone, if one finds them
annoying.


As for me being interested in the subject instead of just in trolling,
well... http://ekot.dk/programmer/jpegcrops/guides.html

Of course one might argue that being interested in the subject doesn't
leave out trolling tendencies, but maybe we should take that discussion
by email or in another newsgroup?

Go out and take some pictures and figure it out.

That's what I do for my personal photos.

Toke Eskildsen wrote in message .. .
Michael Scarpitti wrote:

Photographs CANNOT be composed according to the golden mean, which
is a ratio. Only rectangles, etc., can have ratios, of their
sides.

A great deal of the webpages I've seen on the Golden Section start by
defining it on a line and the moves on to rectangles and sometimes
triangles or pentagons. Examples fresh from Google a

http://www.perseus.tufts.edu/GreekSc...im/Golden.html
http://www.gkmac.co.uk/golden.php
http://cage.rug.ac.be/~hs/polyhedra/dodeca.html

|------|----| - Line divided using the golden ratio.

The very nature of photography is such that objects in the
world are in the images. We divide the space between objects in
photographs to form a pleasing composition, which approximates, in
***some*** cases, the ratios of the golden mean. That has nothing
to do with the golden mean itself.

If the ratio fits then I don't understand how you can claim that it's
unrelated to the golden mean (which is defined by that ratio)? That
sounds very illogical to me!?



The golden mean is aspecific ratio of lengths. Period. The
distribution of objects in the real world cannot be made that precise.
The whole thing is absurd.

Michael Scarpitti wrote:

The golden mean is aspecific ratio of lengths. Period.

Err.. Yes? That's what I, as well as the webpages say!? No disagreement
here.

|------|----| - Line divided using the golden ratio.

|------| - Length 1
|----| - Length 2

The distribution of objects in the real world cannot be made that
precise.

I also agree with that.

The whole thing is absurd.

Yes, if someone wants to hit the exact ratio (which is quite
impossible) when performing composition. Which I don't.

(maybe I should put this in my signature)


Let me try another way:

If we show a bunch of pictures to a great number of people and ask them
to pick out the most pleasing ones, we should be able to extract
certain trends. One of these trends is "don't place the interesting
elements in the center". Further investigation would give us an
aproximate position to place those elements.

At least that's the case if we accept using rules for composition.

That aproximate position could be 1/3, it could be at the Golden Mean
or it could be somewhere else entirely (although I doubt it'd be far
off). I find it safe to assume that there'd be a lot of people pointing
at positions near the mean and fewer people pointing at positions
farther off.

A quick Google search gave me this page, which has a fine illustration.
Don't bother about the mathematics, it's just the curve we need:
http://www.robertniles.com/stats/stdev.shtml

Going with the graph on that page, we have our aproximate position at
the top of the curve. Aiming for that position is our rule of thumb.


Let's assume that the Rule of Thirds points to the position at the top
of the curve. When we apply that rule of thumb by the eye, we're pretty
sure not to hit 1/3 exactly. But that's fine, as long as we're
resonable near it: There'll still be a large number of people that
think it's a good spot.

Since we assume thet the Rule of Thirds is the mean, then the Golden
Section is somewhat off the top of the curve. If we performed our
census, the judgement would be "it's fine, but maybe a bit to the right
would be better".

When we try to apply the Golden Section to a picture, it'll be an
approximation again. Maybe the result is a bit to the left, maybe it's
a bit to the right. If it's a bit to the right, then that's fine: Then
we're closer to the Rule of Thirds, which we assume is the better
choise (most popular). If it's a bit to the left, then that's not so
fine: We're sliding down the curve, so to speak.


It's just a matter of chance. The better our ideal is, the greater our
chance to hit a position close to the top of the curve.

Toke, my apology for the "troll" comment.

Back to business.

Toke Eskildsen apparently said:

Since one rule says 33% and the other say 38% (rounded values), they
are not the same. Period.

You state later that they are the same "for all practical purposes"
and that is the interesting point here. This is where you haven't
convinced me yet.

In a very real sense, they ARE the same. In the series of
fractions: 1/1, 1/2, 1/3, 1/4...., 38% is obviously closer to
1/3 than 1/2. In this case it might as well be the same,
since the visual effect of 33% vs 38% is generally trifling.

Seriously, prior to this thread, I hadn't considered using
the golden section for composition in photos, but looking
through my library I find that I have at least one book that
discusses it:

"Portrait Photographer's Handbook" by Bill Hunter

In it, he defines the "Golden Mean" for composition:
"The golden mean is found by drawing a diagonal from
one corner of the frame to the other. A second line is
then drawn from one or both of the remaining corners
so that it intersects the first line perpendicularly."

This is obviously not using the same golden section
that we've been beating on here, since the ratio in
Hunter's definition will vary depending on the aspect
ratio of the frame; in a square 6x6 frame it will be in
the exact center, and as the aspect ratio changes,
the point(s) will move closer and closer to the limiting
case of an infinitely wide (or tall) frame in one dimension,
where the points would be on the very edge.

So.... here's another wrinkle to confuse the issue. If you're
looking for a "rule" to apply -- and I think the biggest mistake
anyone ever made was to call these ideas "rules," then go
with the rule of thirds. At least it doesn't change.

Dean Van Praotl wrote:

Toke, my apology for the "troll" comment.

Apology accepted.

In a very real sense, they ARE the same. In the series of
fractions: 1/1, 1/2, 1/3, 1/4...., 38% is obviously closer to
1/3 than 1/2. In this case it might as well be the same,
since the visual effect of 33% vs 38% is generally trifling.

I've tried and clarify my take on that subject in my reply to Michael:


However, given the development below, it might not matter at all.

[Snip "Portrait Photographer's Handbook" by Bill Hunter]

That's the same recipe that Vaidd pointed out:
http://www.psppower.com/2002may/crop2.htm

This is obviously not using the same golden section
that we've been beating on here, since the ratio in
Hunter's definition will vary depending on the aspect
ratio of the frame;

* Thinking hard *

I think I understand more now. The page
http://www.bbc.co.uk/radio4/science/5numbers3.shtml
has a fine illustration of applying the golden rule "forever" to a
golden rectangle. If we use the rule from above, the crossing point is
where the spiral of rectangles vanishes in the picture.

So the Rule of Thirds and the Golden Section is quite different: The
first uses a ratio directly (1/3), while the other uses a ratio
recursively.


There's some other questions though.

Assuming af golden frame and assuming one wants to use the golden
section and assuming a photo of a landscape, where should the horizon
be? I'd suspect the second dividing line (which is the 38% from
before), as that would split the image in two parts with the golden
ratio.

It seems that the Golden Section only works in framings that are golden
themselves? That seems very limiting and it makes me a bit suspecious:
When the Rule of Thirds seems to work on all aspects, I would assume
that some variant of the Golden Section does the same?

If you're looking for a "rule" to apply -- and I think the biggest
mistake anyone ever made was to call these ideas "rules," then go
with the rule of thirds. At least it doesn't change.

PSPPower seems to give the same advice, but I'm getting rather curious
as to when and how the Golden Section can be applied at all.

Toke Eskildsen wrote in message .. .
Michael Scarpitti wrote:

The golden mean is aspecific ratio of lengths. Period.

Err.. Yes? That's what I, as well as the webpages say!? No disagreement
here.

|------|----| - Line divided using the golden ratio.

|------| - Length 1
|----| - Length 2

The golden mean is as follows:

A/B is the same as A+B/A

That's it, completely. The ratio is digital form is discussed he

http://www.vashti.net/mceinc/golden.htm

The distribution of objects in the real world cannot be made that
precise.

I also agree with that.

The whole thing is absurd.

Yes, if someone wants to hit the exact ratio (which is quite
impossible) when performing composition. Which I don't.

(maybe I should put this in my signature)


Let me try another way:

If we show a bunch of pictures to a great number of people and ask them
to pick out the most pleasing ones, we should be able to extract
certain trends. One of these trends is "don't place the interesting
elements in the center". Further investigation would give us an
aproximate position to place those elements.

Usually, but not all subject matter lends itself to this distribution.

At least that's the case if we accept using rules for composition.

That aproximate position could be 1/3, it could be at the Golden Mean.....


If you don't put it in the center, and you don't put it at the extreme
edge, what else is there?

PSPPower seems to give the same advice, but I'm getting rather curious
as to when and how the Golden Section can be applied at all.

I remember stuff I read as a kid, where the golden mean was
mentioned as a ratio that made a "more pleasing" shape for
a rectangle, i.e., an improved aspect ratio. I was taught the
rule of thirds over thirty years ago when I first delved into
serious photography, but never heard of using the golden
section/mean until recently.

I dunno.... there are lots of folks out there with way too much
time on their hands. Maybe one of them started the whole
golden section thing as a way to amuse him/herself....

As for a horizon when using the golden section, I would
assume you can choose either the upper or lower point
to locate your horizon, depending on whether you want
to emphasize the foreground or not.

BTW, I mistyped the author's name in my last post, it
is Bill Hurter, not Hunter. Mea culpa...