Would you like to inspect the original subtitles? These are the user uploaded subtitles that are being translated:
1
00:00:00.06 --> 00:00:02.02
- [Instructor] When utilizing generative design
2
00:00:02.02 --> 00:00:04.07
in our projects, there is a typical format
3
00:00:04.07 --> 00:00:07.08
that a generative design process will follow.
4
00:00:07.08 --> 00:00:12.05
This process is generate, evaluate, and evolve.
5
00:00:12.05 --> 00:00:15.04
The generate phase of the generative design process
6
00:00:15.04 --> 00:00:17.08
is the first step which consists of generating
7
00:00:17.08 --> 00:00:21.01
potential design solutions to a design problem.
8
00:00:21.01 --> 00:00:22.09
This set of designs is what we call
9
00:00:22.09 --> 00:00:26.07
a population of solutions, which is the starting point
10
00:00:26.07 --> 00:00:29.03
of any generative design process.
11
00:00:29.03 --> 00:00:31.04
In this phase we, the designer,
12
00:00:31.04 --> 00:00:34.07
specify the set of parameters that the generative algorithm
13
00:00:34.07 --> 00:00:38.09
can adjust to generate different permutations of a design.
14
00:00:38.09 --> 00:00:40.09
Say, for example, we had a tower
15
00:00:40.09 --> 00:00:42.09
which we had parametrically developed
16
00:00:42.09 --> 00:00:44.07
to use four different parameters:
17
00:00:44.07 --> 00:00:47.02
length, width, height, and twist.
18
00:00:47.02 --> 00:00:48.05
In the generate phase,
19
00:00:48.05 --> 00:00:51.01
the algorithm would generate different design options
20
00:00:51.01 --> 00:00:53.00
based on the variables provided.
21
00:00:53.00 --> 00:00:54.07
The important thing to understand here
22
00:00:54.07 --> 00:00:57.05
is that we set the parameters that the algorithm
23
00:00:57.05 --> 00:01:00.05
can then use in the generation process.
24
00:01:00.05 --> 00:01:02.02
Once the design has been generated,
25
00:01:02.02 --> 00:01:04.06
the next step is to evaluate the design
26
00:01:04.06 --> 00:01:07.04
to determine whether or not the generated design
27
00:01:07.04 --> 00:01:09.05
is objectively good or not.
28
00:01:09.05 --> 00:01:12.00
To do this, the algorithm needs some method
29
00:01:12.00 --> 00:01:13.06
of scoring the designs,
30
00:01:13.06 --> 00:01:16.09
which is a numerical-based evaluation of the design.
31
00:01:16.09 --> 00:01:18.05
In the example of our tower,
32
00:01:18.05 --> 00:01:22.05
it may be the amount of floor area generated and cost.
33
00:01:22.05 --> 00:01:25.04
Both of these values can be determined by analyzing
34
00:01:25.04 --> 00:01:27.00
the generated geometry,
35
00:01:27.00 --> 00:01:29.04
such as the net floor area available,
36
00:01:29.04 --> 00:01:31.08
and some cost multiplier.
37
00:01:31.08 --> 00:01:34.04
Based on these values, the algorithm can then rank
38
00:01:34.04 --> 00:01:37.00
which designs in the population
39
00:01:37.00 --> 00:01:41.02
are better or worse than others based on these metrics.
40
00:01:41.02 --> 00:01:44.01
The designer specifies what these metrics are
41
00:01:44.01 --> 00:01:47.07
and the algorithm uses them to rank the designs.
42
00:01:47.07 --> 00:01:50.01
The last phase of the generative design process
43
00:01:50.01 --> 00:01:51.08
is evolving the design,
44
00:01:51.08 --> 00:01:55.02
which is where the magic of generative design happens.
45
00:01:55.02 --> 00:01:57.01
Based on the design rankings,
46
00:01:57.01 --> 00:01:59.06
the algorithm can determine the best designs
47
00:01:59.06 --> 00:02:02.04
and the parameters that make up those designs.
48
00:02:02.04 --> 00:02:04.04
From here, the aim of the evolve step
49
00:02:04.04 --> 00:02:06.03
is to improve these designs,
50
00:02:06.03 --> 00:02:08.09
and it does this by combining and randomly changing
51
00:02:08.09 --> 00:02:11.06
the parameters that make up the best designs
52
00:02:11.06 --> 00:02:14.01
to get them to a specified goal.
53
00:02:14.01 --> 00:02:16.08
In this example, this might mean trying to maximize
54
00:02:16.08 --> 00:02:19.06
the amount of area while reducing the cost.
55
00:02:19.06 --> 00:02:22.00
So in the evolve phase, the algorithm will take
56
00:02:22.00 --> 00:02:26.04
the best generated designs and from them create new designs.
57
00:02:26.04 --> 00:02:29.05
Each time a new set of designs are generated
58
00:02:29.05 --> 00:02:32.01
this is considered a new generation
59
00:02:32.01 --> 00:02:35.08
in which the original population has evolved into.
60
00:02:35.08 --> 00:02:39.04
And with each new generation there is a new set of designs,
61
00:02:39.04 --> 00:02:43.08
which over time get closer to the specified design goal.
62
00:02:43.08 --> 00:02:47.01
This process often uses what we call genetic algorithms,
63
00:02:47.01 --> 00:02:49.05
which are type of optimization algorithm,
64
00:02:49.05 --> 00:02:52.03
allowing us to evolve a set of solutions,
65
00:02:52.03 --> 00:02:56.04
define the best or fittest solution to a problem.
66
00:02:56.04 --> 00:02:59.01
Over time, the generative design algorithm
67
00:02:59.01 --> 00:03:02.05
is able to explore a multitude of design options,
68
00:03:02.05 --> 00:03:05.05
improving them to reach a design goal.
69
00:03:05.05 --> 00:03:07.01
While generative design in Revit
70
00:03:07.01 --> 00:03:10.06
does include different types of generative design processes,
71
00:03:10.06 --> 00:03:13.09
including Randomize, Crossover and Like This,
72
00:03:13.09 --> 00:03:15.05
which we'll learn more about later,
73
00:03:15.05 --> 00:03:17.03
this process is the most common
74
00:03:17.03 --> 00:03:19.03
found in generative design tools.
6133
Can't find what you're looking for?
Get subtitles in any language from opensubtitles.com, and translate them here.