Planning before Executing a Project

[Stan Zaslavsky]

great responses so far and lots to learn from for me too ..

 

my workflow is similar to Justin's - modelling comes first and cameras later.

my modelling i've organised to go from ground up and then back down again - meaning that i will quickly progress from ground level to 1st to 2nd to any other floors wihtout filling in windows or any other details until i get to the roof.

 

once i've done with the roof - then i follow this path back down again.

 

this way i can deal with small details in the modelling during the flow down and generalyl the massing gets done pretty quickly because i dont touch details until later on in the piece.

 

similarly with naming and object conventions - everything is on a separate layer which makes it easier down the track with materials - so i can select all the windows on every floor in one hit, etc etc

 

just my 2c worth

[Crazy Homeless Guy]

I used to be terrible at naming when I first started out, but now I am solid and well organized.

 

I remember naming a file Turtle 01, Turtle 02 circa 1999 just because we had got a African Spur Thighed tortoise that week. I think the project was Phoenix Children's which obviously has nothing to do with turtles. I also remember working quickly before deadlines and just tapping a handful of keys at once when I needed to save a file. That random tapping of the keys makes still makes a rare appearance, but if it does, it is with a file that is used for a quick transfer, and will be deleted less than a minute later.

 

I can only count 1 or 2 design models that we started over the last year. This has its ups and downs. Typically our designers work with a model as they are designing. When we start the model is massed, and contains a low level of detail. The models are typically coming from Rhino / Revit / Skethup / (occasionally) Max / or AutoCad. It all depends what the person building the model feels most comfortable working with. So it helps that it is started, but a good deal of time still needs to be sent in order to make sure it is organized properly, and detailed properly.

Edited October 13, 2010 by Crazy Homeless Guy

[pand0r]

A logically-layered and properly-named scene is indeed bliss to work with. In a multi-lingual company like ours (but I bet there are more multi-lingual than uni-lingual companies in this industry), english is the standard. There is nothing more frustrating than opening a scene and finding out my polish colleague "forgot" to use english :-) Then again, his spelling mistakes aren't pretty either...

[Dave Buckley]

Just wanted to post this link as it shows his process really well documented and he doesn't hold back on explaining how and why he does things in certain orders and ways http://www.ronenbekerman.com/forums/hover-challenge-submissions/483-benjamin-brosdau-tora_2097-a.html

[Dave Buckley]

Travis:

 

One question, if you split you model into various files such as Architecture.max and Site.max

 

What is in you file called lighting and views.max when you establish your views.

 

Is it simply a mass of the building and a roughly shaped landscape/flat plane? then you break off an model the site in much more detail and when finished you merge into your master file and replace the rough landscape?

 

also do you model everything per view? or are there somethings where you will model all of it regardless of the viewpoint such as the site?

[Crazy Homeless Guy]

The views and lighting file contain camera and usually all the individual lights. This serves as the master file that the other files are Xref'd into. There may be multiple lights and camera files if I am doing day shots and dusk shots, where the dusk shot needs a lot more lights than the day shot. If I am doing an exterior where I only need the architecture and site, then those two files are Xref'd into the lighting file.

 

Typically I use whatever base model has been started, and reference that into my master lighting file, unless it is already to heavy of a model, then I may only use the walls, or build a quick mass to test the lighting.

 

We try to only add extra details to the model in the views that you will see them, but all views do share the same model. If the model is coming from Revit, chances are the design team has a decent level of detail already modeled in, at least for exterior shots. Interior shots with Revit models are whole different story. They require lots and lots of refinement.

 

I try not to merge unless it is needed for final rendering passes.

 

We keep the pieces separate so multiple people can work at once, and also because several of our projects are just to large and cumbersome to both manage efficiently in one model, and for Max to keep up when spinning and rotating.

[Dave Buckley]

ok i'm with you now.

 

I'd completely missed the bit where you'd said you'd xref things in

 

let's say you've got two different views that show 2 different parts of the site.

 

is this where your layers come in?

 

essentially both parts of the site would be modelled but if you can't see some of it in a specific view you could have that stuff on a layer and just turn it off for that view?

[Crazy Homeless Guy]

Maybe.... Cars and trees, yes turned off if they are not seen. Or possible and entourage xref file that only Xref's the entourage in for each view. The rest of the site I don't worry about much, I usually just leave it on.

 

We have the luxury of having nearly 50 machines with hyper threaded quad core Xeon's and 12gb of RAM. Our workstations are dual hyperthreaded quad core xeon's. That helps a lot when working with large projects. Our typical project may run between 100,000 sq. ft to 4.1 million square feet.

[Dave Buckley]

any particular reason for using xrefs rather than containers?

 

i don't use containers myself, just being inquisative as always

[Crazy Homeless Guy]

Mainly just comfort level. I tried containers when they first came out and had a few problems. The problems may have been entirely user error, I can't remember exactly what it was though. I should probably try them again and see how it works.

[Dave Buckley]

fair enough.

 

last question from me i think but . . . going back to the way you split your files up.

 

all the architecture in one file, then another file called cafe.max for example.

 

do you model everything in the cafe and then detail the objects that are directly seen by the camera?

 

is your .dwg present in each file? apart from the master?

 

or when splitting into different files do you save selected from the original model that you've been given?

[Crazy Homeless Guy]

Kind of...

 

Usually when we are getting towards this type of breakdown of spaces the architecture team has built a detailed model of the entire building in Revit, a hospital in this case. Nearly every space is modeled, and assembled together. For the cafe I simply drew a section box around the cafe, and exported, and then deleted any overlapping items that the architecture model already had in it. Now I have 2 models that fit perfectly together.

 

Then we go back and enhance and detail the cafe geometry per the view we are going to render.

 

The DWG or FBX is present only when imported, and then converted to a Max model and not linked. I wish they were linked, but we need to refine them to much. If an additional DWG is needed for adding detail (typically site) it is used to add the detail, and then removed from that file inorder to try and keep it light weight.

[Dave Buckley]

another quick question travis, are you using Xref scene or Xref geometry?

 

Being a SketchUp modeller, my current scene breakdowns are a little different hence the question about save selected. Don't have the luxury of the 3d section box in revit. So it's saving selected out to various files. The only problem with that when having modelled all architecture in sketchup is that when you are at the stage to split into model files, you would end up with blank files, as I import into max when I am ready to model furniture.

 

the advantage you have with the revit file is the geometry already in it. where as my shell makes it kind of pointless at this point splitting into files.

 

let's say for example, i have modelled the building (walls, floors, ceilings, openings, doors, windows), they would go into the architecture.max file, but i also want a view of the cafe for example. The only geometry I have representing the cafe is the walls which is part of the architecture file. So i'd have a blank file called cafe.max and no way of knowing if furniture is being modelled in the correct place. Unless I had the dwg present on a separate layer in each .max file so I could see where I need to model the furniture so it's positioned correctly in the camera I have setup in the lighting_views.max file.

 

Hopefully that makes sense, but this is why I ask about xref scene, xref object, and whether you have a copy of .dwg file in each scene that is just turned.

 

Also when hiding things not in visible in the view you are currently working with, how do you go about hiding them? turning xref's on and off or turning layers on and off in the file being xref'd in?

[Justin Hunt]

Xref scene is best if you have lots of object in the file where as xref object is where there one (or only a few) object in the file.

 

In your case use xref scene. I have found that hiding and unhiding Xrefed Scene layers a bit hit and miss. Especially when you forget to unhide layers that should be on in the "master" file.

 

One thing to bear in mind is with reletivly simple scenes splitting the model up too much can be counter productive. You spend more time jumping between files when making updates. Splitting is a good idea when working on large scenes with many building/ rooms etc and when more than one person is working on the project.

 

Try setting up a layer and object naming convention that names the building and the part, eg for Building A , the layers (and objects) have a preface of "A_" and building B would have "B_" etc. This way it is easy to track what is in the Master file and where it came from should you need to mach changes.

 

jhv