Archinect
anchor

Modelling: How did Ito do this?!

RossM

Hi all,

I wonder if there are any hardcore modellers here that would be able to help me?I am trying to model a form similar to Ito's Ghent Forum/Taichung Opera House in 3ds Max but have had no luck in working out the technique used!I have posted some images/screen-shots on another forum...the link is:

http://www.pushpullbar.com/forums/showthread.php?t=5636

If anyone knows the specific technique used I would greatly appreciate any information!

Thanks, Ross

 
Mar 3, 07 10:53 am
Hasselhoff

Looks like Rhino. Would be pretty easy with lofting and then blending surfaces in Rhino.

Mar 3, 07 11:39 am  · 
 · 
BlueSteel

Looks like the basic geometry was a scripted, (i would use rhino script).. Multiperiodic minimal surface of some kind... (sherk possibly)

otherwise you can slog through building it up with lofts and sweeps....

Mar 3, 07 11:54 am  · 
 · 
Hasselhoff

You made me curious so I played for maybe 15 minutes in Rhino. I think you could do it without too much difficulty in Rhino. The geometry actually looks fairly simple if you study the isocurves. The tubes all curve to a common surface and then the triangluar holes are patched. Take some work and just getting do the exact technique, but I think it's fully possible in a reasonable time without scripting as well. I've seen people at school model complex forms like that without any scripting.

Mar 3, 07 12:10 pm  · 
 · 
filo

it was done in rhino as pathwork or lofted surfaces, it could be done in most of the nurbs modelling

Mar 3, 07 12:27 pm  · 
 · 
binary

NURBS......

Mar 3, 07 12:45 pm  · 
 · 
RossM

Hmmmm....thanks for the relies....well i don't have access to rhino but i suppose its gonna be nurbs in max then!thanks!

Mar 3, 07 12:51 pm  · 
 · 
FOG Lite

ah, but you do have access to rhino! You can go hereand get a fully function version of the just released version of Rhino 4 good for 25 saves. You could do make that for fairly easly within 1 save. The new V4 filleting will rock your world.

Mar 3, 07 2:22 pm  · 
 · 
le bossman

i'm not as impressed with the digital models as i am with the resin model. the resin model is amazing.

Mar 3, 07 5:06 pm  · 
 · 

i haven't read it, but may be some clues in here

knowing cecil's working method i had thought the actual forms were based on mathematical models, and hence scripting of some sort, but never know ...

Mar 3, 07 7:08 pm  · 
 · 
mdler

lasers

Mar 4, 07 1:20 am  · 
 · 
ooid

its not a nurb..
have a look at cecil balmond's architecture and urbanism book it tells there this project a bit..
it is polygonal geometry,after having subtracted differentiated spheres from the global shape, they tesellated to polygonal geometry so that you achieve this smoothy seamless surface.

Mar 4, 07 8:04 am  · 
 · 
RossM

Junkieonur that sounds very interesting!...have you tried this technique before?!

Mar 4, 07 8:33 am  · 
 · 
ooid


Mar 4, 07 8:54 am  · 
 · 
aspect

for CG ppl, nurbs/polygon dun really matter, because they can always convert after the desire command being applied.

i think Ito's model is a script, because it looks analyitcal unlike nox, zaha stuff which looks kind of arbituary... since scripting requires strict discipline of understanding mathematics formular.

Mar 4, 07 9:00 am  · 
 · 
ooid

hi aspect,
you are right the orientation of the spaces is scripting but to achieve such a smooth and seamless surface requires additional tools like in Maya..
I dont know so much about Nox's work but Zaha's works especially latest ones, fully generated on scripting and other parametric softwares..thats why they build so fast,

Mar 4, 07 9:05 am  · 
 · 
aspect

yes, after the script, always need some massaging, haha...

i also guess Zaha latest museum at dubai (branching texture) also scripts too!! http://www.tropolism.com/adzaha1.jpg

do u know what program she or any of those architects is using? i myself a big fan of houdini, and sometimes maya... i'm thinking to explore rhino since u guys are mentioning it... i usually transport the object to different programs for different massaging!!

Mar 4, 07 9:17 am  · 
 · 

not usually into this swoopy digital pyrotechnic stuff, but that zaha project is a delicious little jewelbox.

Mar 4, 07 9:36 am  · 
 · 
RossM

Hmmm this is a very interesting thread now....everyone feel free to add there own opinions, interests....I imagine these spaces must feel amazing to be inside...like an inate inuitive caveman like response or something....nature has no straight lines...(except for the nazca line!) :D

Mar 4, 07 10:02 am  · 
 · 
Hasselhoff

That Zaha building looks cool as a model, but I imagine it losing it's sexiness at scalle and becomeing flabby and silly. Have you see the renderings with the people for scale? THe thick parts are like 30 feet tall, so all the little delicat stuff is actually huge. I don't think it will look too great from the eye's point of view. Zaha always does aerial trickery and we almost never see the "5'5" at 50 degree view" of her work.

Mar 4, 07 10:43 am  · 
 · 
filo

i think there is somekind of misunderstading, ito's office is not really equipped with fancy animation software, or GC, and their design process will not really start form a script, and certanly not from a digital enviroment, or from minimal surfaces and math considerations but there is a necessicity to draw this object... soo 1+1=?

Mar 4, 07 11:14 am  · 
 · 

no, ito probably didn't start with a cg model, but cecil is quite adamant about rationalising the system at some point...

in any case, question is about how to make cg model, not ito's office design process.

Mar 4, 07 11:30 am  · 
 · 
anotherquestion

i think you can look out there on the internet and find some minimal surface objects (nurbs) generated by scripts for various 3d programs.

you can use these as components to model and distort, cut, loft curves etc.

Mar 4, 07 11:35 am  · 
 · 
filo

jump, since when design process and final output is not related? design process translates into how they designed it, i can probably think of a 1000 ways to get the same output, but that is why it makes a difference to know how they went there ...

Mar 4, 07 12:19 pm  · 
 · 
A Center for Ants?

damn. that zaha 3d print looks so nice. i can't stand the starch based ones. looks like your object is a sandcastle...

the one issue i have with the opera house is the skin. ito's projects are so beautifully done from the design of the skin reflected inwards. however, this is backwards and i feel like it's a cop-out to just truncate the volume with this shearing slice of glass. it's like a section model. resolve those end conditions!

Mar 4, 07 12:53 pm  · 
 · 
holz.box

oh man... where have i seen that hadid image bfore?!?

Mar 4, 07 1:19 pm  · 
 · 

Ito - build the model and render in photoshop
Hadid - blow nose spray with canned air, repeat as per previous

Mar 4, 07 1:29 pm  · 
 · 

lol, holz.

no argument with you filo. only, the question asked was how the office built the cg model. how he designed it is another issue altogether...

Mar 4, 07 8:38 pm  · 
 · 
mdler

lasers

Mar 4, 07 10:21 pm  · 
 · 
aspect

to get the right answer and method of doing it, i suggest RossM go to Maya chatroom or any CG forum and ask, and u usually got pages of instructions from the CG guru!!!!!!!!!!

that's what i usually do!! architect is always the last one who would give u a "practical answer" to the problem.

Mar 4, 07 10:25 pm  · 
 · 
o d b

junkinur: hate to burst the bubble on cecil and toyo, but i think the "smoothing algorithm" may be none other than the subdivided surface command in maya

make a polygonal object, convert to sub-D, and have fun =D

i think it's actually difficult to achieve that kind of topology in rhino, unless you start w/ a surface, cut some holes in it, and blend between the holes. it would be difficult to construct out of curves and lofting i think.

Mar 4, 07 10:29 pm  · 
 · 
o d b

after looking at the other site featuring ito's images, it very well could have been done using revolved surfaces as somebody mentioned, except that they are not all perfect circles. the sections might have been swept along 2 rails, probably the top and bottom edge of the trimmed portion of the flat surface.

maybe i'm behind on the modeling question, but it's an interesting thing to think about.

Mar 4, 07 10:32 pm  · 
 · 
grahambarron

i saw that ito bldg at a gallery in tokyo. at first i was really impressed. then i realized that those tubes are just hollow circulation columns, and all the floor plates are planar.

ito is still working in grid format. same with sendai mediatheque.

why doesn't he work more with section? the only sectionally interesting places are were the column/tubes begin to sloped down, but you can't inhabit those. smt likewise looks interesting from afar, but standing on each floor plate the space is quite bland.

i would guess maya rather than rhino. in the animations in the gallery there were uv lines that looked more maya-like.

here's how i would do it:

1. make cylinder nurbs "columns" that are on common plane top and bottom
2. push and pull uv's on plane
3. loft between uv's and attach seams
4. draw cube around bldg envelope

the one thing i do like about taichung is that the curvy lines of the columns are extended into the landscape. nice blending of inside/outside.

graham

Mar 5, 07 1:57 am  · 
 · 
PerCorell

Hi all you fancy graphics modelers , It is great what the computer offer and Gee how nice it would be with one of those houses but, tell me one thing though.

How do one bring the smooth forms from screen out into reality, how do you transfere the visionary designs into real work drawings Now do you make sure that the house now don't cost four times as much and need to be fiddled in wiremesh and ferrocement.

Now don't take this as a crit, as I realy love that architecture still, architecture combine ,and it is difficult to emagine how this combine with any building method at all.

Mar 5, 07 7:57 am  · 
 · 
PerCorell

What I know from zaha designs are, that the concrete formwork , just became extreemly expensive and outmost difficult to make. "A real challance that asked our most experienced engineers ,just to do , just making the formworkit was only possible due a lot of innovation from our side" -- that was the comment from the danish engineering firms , after Zaha Årdrupgård append.

Mar 5, 07 8:13 am  · 
 · 
grahambarron

in the same gallery exhibit in tokyo, ito had a set of actual form work that was used for one of his curvy structures. you could walk on it and see how it was made.

no doubt it's difficult and costs a lot more than standard orthogonal buildings. really good engineers and craftsmen love that kind of challenge though. and the results are incredible.

graham

Mar 5, 07 1:38 pm  · 
 · 
PerCorell

What the everyone in copenhagen have an oppinion about are the Opera and the new playhouse ;

http://www.kglteater.dk/site/OmTeateret/Scenerne/Skuespilhuset.aspx

A lot of things happen around the harbour , still the city as such fail a decent solution to trafic problems , one suggestion is to plaster a light bridge between the Opera and the new Playhouse , but basicly this work against the best quality of the copenhagen city, the fact that you can sail right in to the center of the town.

Mar 5, 07 4:28 pm  · 
 · 
autarchy

use evolver > thats what it was developed in.

mathematica also generates 'em.


If you want to fake minimal surfaces, head here >

http://www.susqu.edu/brakke/evolver/examples/periodic/periodic.html

and component one of the surface typologies (using rhino work with networkSrf and loft).


Mar 21, 07 1:25 am  · 
 · 
garpike

sweep1
sweep2
network surface

and secret weapon:
blend surface

It's up to YOU to keep things tangent.

The best part about Architecture is "minimal surfaces" are a starting point. Purists need not apply.

Mar 21, 07 2:08 am  · 
 · 
garpike

(loft is bush league)

Mar 21, 07 2:09 am  · 
 · 
AndrewBarrie

Here's the low down on the Ghent model:
Ito's office is very low-tech software wise, and only very recently has begun to use sophisticated tools like Maya or Rhino. The vast majority of modelling in the office is of the physical, paper-and-foam kind.

The Ghent project was was planned in two dimensions using (I think) Vectorworks, and the structure modelled only physically using clay products. The 2D plans set out the curved shapes at the mid-points of the "columns", and also a ploygonal set of intersecton lines on the "floors" where the structural surface became horizontal. At a late stage in the preparation of the scheme, a new staff member (Florian Busch) joined the office who was able to model the structure using (as I recollect) Maya in much the way described above.
The physical model was the output in resin (by a local agency), but the curves were very rough. It then took days of sanding and repeated paintingto smooth out the surface. The base and other elements of the physical model were made is a farly standard way.

Click on this link and then "office toyo ito" for a number of photos of the model under construction:
http://robertbaum.de/japan/index.htm

Robert Baum is a cool German guy who worked as an intern for six months. I worked in Ito's office for several years, and made (with Robert and a number of student interns) the model of the Ghent competition that was used in the exhibitions mentioned above.

Mar 21, 07 5:29 am  · 
 · 
psycho-mullet

This is a little late but I just saw this and was interested so I took a stab....

Hopefully these links work.




I don't think I went about it quite the same way as the images posted, those look like Maya and they look constructed not scripted.

- I started with an 8 sided cylinder
- Distorted it into a sort of barbell shape
- Deleted every other of the eight faces on the head and tail
- Copy around like crazy attaching and welding at the openings created in step above
- Trim down barbells cutting end ones in half.
- Apply Tubosmooth
- Apply Shell
- Make pretty rendering



Ito's is obviously distorted and not so uniform but I didn't have time to do the final tweaking.

If you look at their model there are planar triangles at the floor levels between the openings, I tried modeling it that way initially but it was much more tedious, this achieves the same effect more quickly, but ultimately you might need to use the more meticulous and intentional approach.

Jun 8, 07 3:58 am  · 
 · 

computing is sooo cool, no?

i remember when up-front was considered so cutting edge... ;-)

Jun 8, 07 6:05 am  · 
 · 
rayray

psycho - what program did you use and how long did it take (also, what's your skill level familiarity with the progam?)

very cool object

Jun 8, 07 11:32 am  · 
 · 
psycho-mullet

- 3d studio max.
- Modeling the final thing took about 10 minutes
- Figureing out the best way to do it out took the better part of some documentary on the 1980 US olympic hockey team.

- Skill level... hmmmm... well I do 3D visualization for a living now...

Dare I say I don't know what up-front is?

Jun 8, 07 2:56 pm  · 
 · 
FOG Lite

psycho-mullet is a bad-ass

Jun 8, 07 3:57 pm  · 
 · 
aspect

psycho-mullet> you are my man!

Jun 9, 07 10:30 am  · 
 · 
filo

na ja, cool

Jun 9, 07 10:36 am  · 
 · 

Block this user


Are you sure you want to block this user and hide all related comments throughout the site?

Archinect


This is your first comment on Archinect. Your comment will be visible once approved.

  • ×Search in: