Home
Updates
References
Student papers
week 1, 2 Sept.
Defining languages of designs
week 2, 9 Sept.
The Yingzao fashi
week 3, 16
Sept.
Defining style formally: shape grammar
week 4, 23 Sept.
Grammatical versus stylistic correctness
week
5, 30 Sept.
The Yingzao fashi, shape grammar, and extant
buildings
week 6, 7 Oct.
AL away
week 7, 14 Oct.
Paper topics
week 8, 21 Oct.
Special topic, consultations
week 9, 28 Oct.
Special topic, consultations
week 10, 4 Nov.
Presentations
week 11, 11 Nov.
Presentations
week 12, 18 Nov.
Presentations
week 13, 25 Nov.
Review week
|
Monday 29 September 2003
I know that some of you haven’t started the assignment yet, so it’s not
too late for me to send a little message.
-
The section grammar is included in the Penn paper (Li forthcoming),
which you can download here (text,
images). The grammar
is shown in figure 6, but don’t overlook the constraints on rule application,
on page 6.
-
If you think there are stylistically correct designs that the grammar
is not producing, you can “loosen” the grammar by changing the rules
or the constraints on rule application (or both).
-
How do you know whether there are designs as mentioned in (2) above?
We assume that:
Therefore, since we are dealing only with 6-rafter buildings, we
have 7 possible locations for a column. With 2 positions always “filled,”
there are 2 ^ 5 = 2 × 2 × 2 × 2 × 2 = 32 combinations
of columns. Your grammar may generate fewer than 32 sections, so you
may want to consider the remainder of the 32 as candidates for (2)
above.
-
If, because it is producing unstylistic designs, you want to “tighten”
the grammar, you can change either the rules or the constraints or
both.
-
How do you know whether the designs produced by the grammar are stylistically
correct or not? Short answer: You don’t. Long answer: If your only
information is the corpus of 18 sections, then all you know for sure
is that those 18 sections are stylistically correct. Everything else
is speculation. What we are trying to do is speculate in a way that
both is precise and provides the 3 benefits mentioned by Stiny and
Mitchell.
-
Don’t forget the descriptions. In particular, since a design consists
of a section and a description, two designs with the same section
but different descriptions are not the same design. Take for example
the design D, with the description “a 6-rafter building, centrally
divided, with 3 columns,” and the design E with the description “a
6-rafter building, a 3-rafter beam abutting a 3-rafter beam, with
3 columns.” D and E have the same section, but different descriptions.
Therefore they are different designs. One may be in the style, and
the other not. One may be grammatical, and the other not.
Monday 22 September 2003
-
I am wonder about how we define a design is legal or illegal. For
a given design. If we can creat it with only one rule of the grammar
which contains more than one rules. Is this design legal or not?
…
For the square grammar. Does all the dots which have the same positions
with any of the squares have to be erased? Is it the rule?
-
Dots first. The dots in the square grammar are for controlling
the application of the rules only. They are not part of the final
design. So when you’ve finished adding squares, you should erase
all the dots. The dots are like the construction lines when you
make a drawing. You need them to make the drawing, but you don’t
necessarily want to see them in the final drawing. So you might
erase them, or even make an entirely new drawing by tracing over
the old one.
About rules. They are optional. If a rule can be applied to the
current shape (by can be applied is meant
matches under a Euclidean transformation),
then you, the user, may apply it or not apply it. The choice is
yours. You don’t have to apply any rules. So in the square grammar,
you could do nothing, and have one square. Of course, you should
still apply the dot-erasing rule. So it would be more precise to
distinguish rules that do something to the design (like adding a
square) and rules that manipulate labels (like erasing dots). See
the difference between distinctive design
and final design, mentioned in the Penn
paper.
How to define whether a design is legal or illegal? Actually there
are two kinds of legal here, and I really
should think of two different words for them. But you are getting
right to the essence of how we should think about problems like
ting tang sections.
-
One kind. A design is legal if it can be produced by the grammar
in question. Now that we’ve gotten clear on the rules about
applying rules, determining this should be straightforward.
-
The other kind. A design is legal if it appears similar to
the designs in the corpus. When you say things like “a beam
may not be longer than 6 rafters because it is too weak,” you
are proposing criteria of similarity. Where do these criteria
come from? From all your knowledge – architectural, structural,
historical, cultural, etc. – which is not contained in the corpus,
because it is just a dumb list. Now we can decide that the knowledge
is relevant, but this is an act of interpretation, so evaluating
this kind of legality is subjective, while evaluating the first
kind of legality is objective.
-
But I am still confused by the assignment. If I choose the sqaure
grammars. Call it grammar A. It also has two rules, rule 1 and 2.
So I add rule 3 to creat grammar B.Now I can easily produce a design
by grammar A. Precisely speaking, I produce it by applying rule 1
and 2 successively to the initial design.
Now, here is the question. Is this design legal or illegal in grammar
B? Probably, I consider it to be legal because I can say I produce it
by grammar B although I didn’t apply the rule 3. But it is my choice
to apply rule 3 or not.
If my discussion above is right,how can I create a design that is legal
in grammar A but illegal in grammar B?
Wednesday 17 September 2003
- Dennis Leung pointed out a mistake in figure 4 of my Penn paper (Li
forthcoming). After the 6th and 7th applications of rule 1, a dot should
be added inside one of the existing squares. He is absolutely right.
The derivation shows what I expected to happen, but I didn’t check carefully.
My mistake. So, something to think about: what rules would give the
derivation shown?
Tuesday 16 September 2003
Sunday 14 September 2003
- The complete and final version of Li (forthcoming) is now available.
Download the text and images
(PDF, 254 and 241 KB).
Saturday 6 September 2003
Thursday 4 September 2003
-
A subassembly is an assembly that is part
of a larger assembly. An assembly is a set
of manufactured parts assembled to make a completed product, especially
a machine (adapted from the American heritage
dictionary). In the context of the assignment, the complete
building is the assembly; building sections and roof sections are
two kinds of subassemblies. So, do tasks 1, 2, and 3 for building
sections. Then do tasks 1, 2, and 3 for roof sections. Finally, answer
the last question.
-
For building sections, it is sufficient to draw line diagrams (like
the one I drew on the whiteboard). All we need to see are the salient
features (number and location of columns, beams, rafters, etc.). We
do not need to see bracket sets.
-
For roof sections, please draw to scale. Actually, you will find
it easier to do so, because it is a graphic method of calculation.
-
To make a PDF file on a Mac, you simply prepare to print the document.
But instead of sending the document to the printer, you save it as
a PDF file. In OS X, this feature is built into the system. In OS
9, you need first to install a piece of software from Adobe (I don’t
remember the name). I don’t know how to do it on a PC, but I’m sure
it’s similar.
-
A sharp-eyed student has asked a number of questions about the 18
ting tang sections and their descriptions.
I should explain in more detail.
-
p. 313, above. This is dian tang,
so it is not one of the 18 ting tang
sections.
-
p. 315, above. “Bing rufu 並乳伏”
means “two rufu touching.” In other
words, “qian bing rufu, hou bing rufu
前並乳伏,後並乳伏.”
-
p. 315, below. “Qian hou ge 前後各”
means “in both front and back.” “Ge”
is optional. Compare with p. 315, above.
-
p. 316, below. “Dui 對” means “abutting.”
In other words, there is no unspecified bay between the two beams.
Compare p. 315, above and below.
-
p. 318, below. The description should be “bajia
chuan wu, qian hou zhaqian rufu, yong liu zhu 八架椽屋前後劄牽乳伏,
an 8-rafter building, 1- and 2-rafter beam in front and in back,
with 6 columns.”
-
p. 320, above. The description should be “liujia
chuan wu, qian hou rufu, yong si zhu 六架椽屋前後乳伏用四柱, a 6-rafter
building, a 2-rafter beam in front and in back, with 4 columns.”
-
p. 320, below left. The description should be “sijia
chuan wu, zhaqian dui sanchuan fu, yong san zhu 四架椽屋劄牽對三椽伏,
a 4-rafter building, a 1-rafter beam abutting a 3-rafter beam,
with 3 columns.”
-
p. 321, right. The description should be “sijia
chuan wu, qian hou zhaqian, yong si zhu 四架椽屋前後劄牽用四柱, a
4-rafter building, a 1-rafter beam in front and in back, with
4 columns.”
-
p. 321, left. The description should be “sijia
chuan wu, tong yan, yong er zhu 四架椽屋通檐用二柱, a 4-rafter building,
clear span, with 2 columns.”
|