Graphics VS. Design, and what I learned from video games

24 01 2011

A few weeks ago I posted a blog asking for critiques on some 3D models components I had made.  I posted it on IGN, WordPress, and Land8Lounge.  The response was fairly low, and I tried posting it as a discussion thread on Land8Lounge as well.  This got some helpful comments, but it also possibly inspired a thread that brought up the issue of rendering VS. design in Landscape Architecture.  This question came up a few weeks ago at a networking event for NYC landscape architects, and I talked to a few different people about it.  It made me wonder: Why is it that designers often have a negative reaction to “high end” renderings?

If I said I had never seen a rendering and assumed that the graphic quality was hiding something I would be lying.  I think this often comes from past experience, either in firms as CAD monkeys, or in school.  We, I think, have all had a presentation where we worked our ass off to the last minute perfecting a design, leaving little time to get perfect graphics, so we go out with what we can, trusting our peers and other trained professionals to see through the pencils lines to the heart of the design beneath, only to be disappointed.  We have also all had the time when we have seen the person slap together a half-assed design, but with pretty pictures, and when we are counting on our bosses and teachers to see it, they instead get “blinded” by the crisp lines, hypnotized by the texture work, and in a trance from the shadow quality.  I think it’s this shared experience that leads so many of us to distrust pretty drawings: we all know just how easy it is to lie with graphics.  Whether it is drawing a plan that doesn’t show steps because they client didn’t want them – even though it’s not physically possible, screwing with perspective sizes, or hiding views that you don’t want noticed.  We know the tricks all too well; in part because we use them to some extent ourselves.

We choose a rendering style based on what we want to show and not show: Computer graphics traditionally show a Utopian version of everything – where it all looks fresh from vacuum -packaging, but it makes the space look more contemporary to clients; Hand graphics hide views by simply not drawing the far background, but show more flow and life in the space.  We choose views that may not show the space in the best ways, but show a feature we want to emphasize over others.  It’s a limit of non-physical models – you have a finite window onto the world, you are going to be careful about what you choose to show to show off what you want to be seen.  The problem is, we need a way to easily communicate to clients, and in forms other than physical models, and clients who have not been trained as designers get the best feel for a space through perspective drawings.

-Utopia?

So the issue becomes not graphics VS design, because it never really was outside of designers’ heads.  The issue is how can we make graphics that communicate effectively, and manage to not fire off alarm bells in designer’s and client’s brains.

Personally, I choose to concentrate on computer graphics over hand graphics not due to a lack of skill (believe me, if I put the time into drawing that I have put into rendering I’d be pretty good with a pencil), but because in my mind, it is the most honest form of representation.  If I model everything the way it is – which I can with no extra effort – I know exactly what I will see from a given view.  I can put a camera inside your eyes when you are sitting on a bench, and if you would actually see a sliver of that utility box, it’s going to be in the rendering.  The problem is, it would still set off alarm bells, all because of an effect that is becoming well-known in the media worlds of movies and video games.

The “Uncanny Valley” is an issue that became most well-known a few years ago when the movie “The Polar Express” came out.

The Uncanny Valley

It’s the theory that as things become more realistic, they become more familiar, but only to a certain point.  Once things become TOO realistic, small things that are wrong make us cringe and react negatively.  In “The Polar Express” the thing that set people off was the eyes.  For all the realism the characters had, from mannerisms created by directly copying from actors in Motion Capture suits, to careful texture work, the fact that the eyes didn’t glisten correctly gave them a dead look, and made some people instantly see, instead of a heartwarming children’s story, a movie about a train filled with zombies… (note to self – Make a movie about a train full of zombies.  “Brains on the Orient Express”?)

IGN.com

In the ensuing years movies, and video games, have dealt with this issue in one of two ways.  Some have gone the route of making themselves, while beautiful, purposefully unrealistic.  This brings thoughts of movies such as “Up”, and games such as “Little Big Planet 2”.  Neither of these tries to be realistic – the characters look like cartoons, but with realistic flesh-tones (or woven sack-tones in the case of Little Big Planet), elements.

Little Big Planet

The other method is to continue to push the boundaries of realism.  Movies like Avatar skirt this method by having realistic aliens that we have no internal reference for as digital characters.  Where true realism is pushed the hardest today is in gaming.  Games like Uncharted 2, Heavy Rain, and Read Dead Redemption push realism in gaming past where it has been before.  As some of the best looking games widely available, they all have one thing in common.

Grime.

Uncharted 3

None of these games are set in pristine areas.  They all are set in places that have been lived in, that have wear and tear, grease and grime, chips and gashes.  I think this is one of the essential things to making a convincing 3D rendering, and one that doesn’t make you think you are being tricked by graphics.  The splinters out of the wood crate, the dirt on the boots, the powder burns on the pistol, all give it a realistic feel.

Red Dead Redemption

If that detail is paired with a render engine that has more than 1/30th of a second to output, it would give amazing results.

And those are the kind of results that I think we, as designers/graphic artists have to work for.  Gone are the days of the pristine landscape with rows of identical, perfectly pruned trees.  If we want buy-in from clients, and from other designers, we need to show all the blemishes on the face of our designs.  Whether it’s the ugly light industrial building that is visible through the trees, or the mottled color in the bricks, these are some of the things that A) will affect the spirit of place and B) that will create a sense of life in our renderings.

Are hand graphics still valid? Of COURSE they are.  Neither hand graphics nor computer graphics is inherently “better”.  They are just different.  The grime and dirt of the world in computer graphics is no different then adding a little of every color to a tree in a Mike Lin style render.

BeLoose.com

It adds a little depth, and a spark of randomness that is what makes the world what it is.

Uncanny Vally image from http://ntlkdesign.co.uk/blog/

Little Big Planet 2 image from littlebigplanet.com

Mik Lin image from Beloose.com

All other images from IGN.com





An Advanced Course in PC Hardware: Choosing Components

8 01 2011

The first step in picking what components you want in your computer, whether you are building a custom setup, or buying one off the shelf, is looking at what software you expect to use.  I was looking for a rig that would work for Photoshop, AutoCAD, SketchUp, Thea Render (3D-Rendering Engine), and of course, games from time to time.  The next thing is to figure out what each one of these software packages rely on most heavily- Processor, RAM, or Graphics Card:

Photoshop: Here are the system requirements:

  • Intel® Pentium® 4 or AMD Athlon® 64 processor
  • 1GB of RAM
  • 1024×768 display (1280×800 recommended) with qualified hardware-accelerated OpenGL graphics card, 16-bit color, and 256MB of VRAM

The newer Photoshop versions make use of graphics cards more and more, but in bigger file sizes they also rely heavily on Processor and RAM.  The amount of layers, filters, and file sizes you normally work with will determine how much of either of these you need.  As I am looking to do plan graphics, printable high quality at 36″ x 48″, with many (40+) layers, I, in short, need a LOT of both RAM and Processor.

AutoCAD: The stated system requirements for 64-bit AutoCAD11 (2D) are:

  • AMD Athlon 64 with SSE2 technology, AMD Opteron® processor with SSE2 technology, Intel® Xeon® processor with Intel EM64T support and SSE2 technology, or Intel Pentium 4 with Intel EM64T support and SSE2 technology
  • 2 GB RAM
  • 2 GB free space for installation
  • 1,280 x 1,024 true color video display adapter 128 MB or greater, Microsoft® Direct3D®-capable workstation-class graphics card

First, you can run a PC at either 32-bit or 64-bit, 32-bit can only see up to 4 Gb of RAM.  Generally if you are building a new system you will go with 64-bit to increase the amount of RAM you can use now, or could upgrade to in the future.  None of these are amazing stats, but the RAM is slightly more powerful than the rest of the system- if all the components were equally important I would expect to see a Pentium 4 paired with 1 Gig of RAM and a 256 MB Graphics Card, or a P4 3.0 GHz (processor speed)/Dual Core Pentium 2.0GHz, 2 Gig RAM, 256 Meg Graphics Card.  This basically tells me that for 2D CAD the RAM is a bit more important than the Processor (which comes more into play with CAD’s modeling tools), and that while you need a Graphics Card, it does not need to be a great one by any means.

SketchUp: Recommended resources:

  • 2+ GHz processor.
  • 2+ GB RAM.
  • 3D class Video Card with 512+ MB of memory or higher. Please ensure that the video card driver supports OpenGL version 1.5 or higher and up to date.
    *SketchUp’s performance relies heavily the graphics card driver and it’s ability to support OpenGL 1.5 or higher. Historically, people have seen problems with Intel based cards with SketchUp. We don’t recommend using these graphics cards with SketchUp at this time.

While Processor and Graphics card are needed, RAM is the biggest limiting factor, in my experience, when you have large scale and or detailed models.

Thea Render: I could not find any system requirements, likely because the program is still in beta (testing phase), and while it currently only uses Processor and RAM, they are adding Graphics Card based rendering in a future update.

Games: For this I took the example of one of the newer PC games that has come out, and one that I wanted to be able to run at full bore- Civilization 5.

Minimum system requirements are:

  • Intel Core 2 Duo 1.8 GHz
  • 2 GB RAM
  • 256 MB nVidia or AMD Graphics Card

When running it with these specs on a laptop I could play it, but far from maximum settings.  The recommended settings are:

  • Quad Core 1.8 GHZ
  • 4 Gig RAM
  • 512 Graphics Card

Gaming is generally more Graphics Card intensive than productivity software, in part because the architecture of the graphics card is better at drawing faster – for better frame rates, while processors can do more math faster.  So in an enclosed system like a game where it is working with a small set of parameters, Graphics Cards shine.  But when you have many layers of images all affecting each other, the processor takes the lead back.

What am I left with after all this?  To get the best out of all my software, I need a good Processor, RAM, and Graphics Card- I can’t save on one to improve the others.  I also know I use a lot of this software more intensively than most, and I want to be future proof for a bit, so I need to exceed these system specs.  One place where I can save a little money now is in the RAM and Graphics Card.  That is because these are fairly easily upgradeable- with the right mother board you can add RAM to your existing RAM without replacing it.  Also, with the right Motherboard and Graphics Card, you can run in SLI or Crossfire – a method of tethering two Graphics Cards together in your system, and having them act like one, much more powerful card.  The nice thing both both of these is that you can spend, say $300 now on a Graphics Card, and in a year spend $200 on another card, and get nearly the same performance that you would have gotten by spending $700 now.

That being said these are the basic specifications for my new computer:

Processor- Intel Core i7 (Quad or 6 core) with a speed of at least 2.5 GHz and I want the ability to overclock in the future to upgrade slightly.

RAM – 6 Gig (Most motherboards can now handle in the range of 24-36! Talk about future proof!)

Graphics Card – 1 Gig at least, unsure of if I am going to go nVidia or AMD, it depends on the exact card (more later)

Hard Drive – 1TB I use some HUGE files (200 Megs for a single photoshop file), and I don’t want to worry about space for a while.  I may do a solid state boot drive down the road.

Optical Drive – DVD RW for sure, I probably don’t need a BluRay drive, as I could add one once burners get cheaper

Motherboard – Needs to fit my processor, and at least 2x my graphics card

Case – needs to fit everything, have great cooling, and have lots of room for additional Hard Drives, burners, and maybe even water cooling (if I overclock in the future).

Next time I’ll start looking at each individual item, and how I picked which to go with.

Ignoring the odd artifacting, this is what Civ 5 Looks like with the laptop:

And this is what it looks like with the new build:





Tuesday Tutorial: Time to get your hands dirty

25 05 2010

Dirt is one material that, while rarely fully exposed, is often a base material that other elements are placed over.  Many times these elements will have gaps between them where there is exposed dirt, whether it be in a mulch-less planting bed, or in areas like a dense forest as we have here.  You cold also combine the dirt with another layer of material, like a scattered moss or leaf-litter, to add some depth and realism.  I have used this in 3D models before with good results, but in a 2D rendering it can add slightly more detail than is useful, creating a cluttered look that makes the drawing a little harder to read.


The first process for making dirt is essentially identical to how you make turf.  However, the images I selected for this tutorial have some major imperfections that will allow me to show one of the powerful tools in the Photoshop box: spot heal.


You can see here that the first image I am using, from a residential lot, has a large pipe with a string attached that sits in an area that otherwise I would want to use.  I could give up, and use a different image, or only use a small section.  I did not take the photos intending on using them for this purpose, and therefore only had a few, and I also was only using a small piece from my second image due to the angle and size of the bluestone patio in the image.

As I need some flexibility in size to reduce the tiled look, I opted for the slightly more challenging method of healing the pole and string out of the image.  First, however, I adjusted the perspective of the image to minimize scale changes throughout the dirt.

Once I did that, I was able to use my heal brush to remove the pole from the image.  In an image such as this, it involves a relatively simple process.  First, select the spot heal tool from the main drawing toolbar.

For an image such as this, its best to use the spot heal tool instead of the heal tool.  Heal is better when large areas need to be covered using techniques I will cover later in the series.  The spot heal tool works by essentially blurring the area.  To demonstrate the basic idea, I created a colored grid.

I then used the spot heal tool on the path seen in purple:

This is the result:

So by using this tool, and painting over the pipe and string, you are able to essentially eliminate them from the drawing with very little work, and you now have a much larger area to pull your texture from.

Next you have to pull an area out to use as your pattern.

As you can see, there are some areas in the bottom of the drawing that are much lighter than the rest of the dirt, so you should again use a burn layer and around 8-9% opacity black to paint the areas darker.

This light amount of paint evens out the textures greatly, giving a pattern that will look fairly uniform on its own.

Once you have that, it’s the same method of setting up two buffer areas with guidelines, copying them, erasing them, and then cropping the image down to creating tiles without hard edges.

This material looks pretty good, but when rendered it will likely look a little to cool and dark, almost like rough asphalt.  By shifting the image’s saturation, you can get an image that it warmer and a bit lighter.

I took the same steps with the second image, giving me a second, smaller texture that is equal in quality to the first.

By layering these images in the same way as I did with the turf textures, and with the same masking layer pattern, you get a seamless topsoil layer.

Next week, we will tackle the granite outcropping on the west of the site.





Tuesday Tutorial: Turf, and basic techniques

18 05 2010

As it is with many things, my technique for turf, and the similar techniques I use for other materials, were born of other needs.  Namely, I hated the way my turf looked in 3D renderings.  You could use a created texture, but it never looked organic.  You could use a photo, but you got TERRIBLE tiling effects.  You could do a slight color overlay to soften the impact of the tiling, but the color was always off, and you would lose to much of the internal texture.  I tried all of these, before scouring through tutorial sites and stumbled on the idea of using two separate textures, with a masking layer to blend the two.  This is used commonly in 3D models when you have to put a non-square image onto a model.

Say you have a beer… sorry I got distracted there for a second.  So, you have a bottle shaped model, and you want it to be beer.  The first thing you will do might be to fill the bottle with liquid, and make the material into tinted glass.  You then have a very pretty blank beer bottle.  It needs a label to look right.  Many beer labels are not square, but no matter what the image you import onto the model will be square.  You handle this by making two images.  One is the image you want shown, the other is a mask, like we make for the hand drawn lines- with what you want shown in white, and what you want to have invisible in black.  Now, you normally just place a single image, so you get one label.  But there is no reason you couldn’t tile the material so it would repeat hundreds of times over an open field. And there is no reason this technique cannot be used in photoshop to make detailed, photograph based textured that don’t show any tiling.

The best way to get good, organic texture, and realistic color, is by using photos.  These can be photos you took expressly for texturing, or everyday photos that have a good amount of the material you are looking for.  The best option is images takes specifically for texturing, as you can take a picture that will be fairly evenly lit, and taken from above, reducing the amount of perspective in the image.  For this tutorial I am using two photos I took of a small residential project I designed last summer:

The first thing you do is crop the non-grass areas out of the image, so you have a decent base to work off of.

This gives you a clear area of turf to work with.  Having other distracting elements removed from the photo also allows you to see inconsistencies in color.  These inconsistencies will show clearly when tiled.

Looking at the image, you can see how much lighter and smaller the blades of grass are on the top and right of the image.  By cropping those areas out you easily eliminate some trouble spots without resorting to the slower techniques next.

There is still a darker area to the left of the image.  The way I will address that is by adding a second layer above the turf image.

By setting the layer to linear burn, and using a round, soft brush, set to black, with around 8% opacity, you are able to paint the image slightly darker in the areas that need it to even out the over all texture.  By using linear burn you darken the tone of the image without adding gray to it, muting the color and texture.

That is all the paint I needed to lay down to change the texture to this:

The image is a little more flat then i would like, so next I increased the contrast to give more sharpness to the texture itself.

As you can see, much of this process is trial and error- looking until you see a problem, fixing it, then looking for the next problem.  After doing all these adjustments, I noticed that the perspective makes the foreground far more coarse than the top of frame.  To fix this, I used Edit->Transform->Perspective to both shrink the foreground.

Once cropped, you have a nice smooth texture that still has a realistic amount of character.

This seems like a fairly regular texture, with a good amount of color, so its time to make the image tile without a hard edge that will always show clearly.  The first step is to lay some guidelines around 1/10th of the way from the edges.

Select the areas between the edge and the guidelines and the outer edge, and copy it on a new layer to the opposite side.  This makes it so you have a smooth transition from the left side to the right, with no edge.

You will still have hard edges where the new layer hits the old, but you can now make a transition between the two.  For this tutorial I used the most basic method, with is to use a soft eraser to fade the transition layer from the edge in.

Now by cropping down the image to where you have the guidelines you will make a tile that will not have a hard edge.

Now with a smother transition, preventing a hard edge your ready to make a pattern.  Hit Ctrl+A to select all, then go to Edit->Define Pattern and name your new pattern (Something like Turf Layer 1V1.0) and save the image in a folder- mine are saved to the desktop->Render Patterns->Turf->Turf1V10.jpg.

This way you can back up all our render patterns, and move them to other computers just by opening the image, and defining the pattern again.  Now give your new pattern a shot, using it to fill the turf areas in whatever rendering you are working on.

Now you repeat the same steps with the second image.

Crop

Burn

Perspective (correct for off-center image here as well)

Crop and increase contrast, copy edges and erase to eliminate straight edges

This is a key point in the creation of your second pattern. To make an organic, non-repeating pattern, you must make sure the two final patterns sizes will not allow them to line up often, creating a repeating overlap.  You do this by looking first at the size of your first pattern.  My layer one is 280×224.  That means you will have an edge (left to right) at 0,280,560,840,1120,1400 etc, and (top to bottom) at 0,224,448,672,896,1120 etc.  You want to pick sizes for your second layer that will not line up with those.  I resized mine to 600×415.  That means that left to right my two tiles will not repeat until 4200, and top to bottom until 92960.

Once you resize your pattern, save it and make it into a pattern just as you did with layer 1. And paste it as a second layer in the same rendering.

The last step is to create a clipping layer, which acts basically like a masking layer, for the grass layers that will allow the two patterns to mix. First pick a noisy, random pattern brush.

Then go into the settings for the brush, and max out the size jitter, and turn off the brush presets other than Shape Dynamics, Scattering, and Smoothing.

Then turn up the scattering, and turn the count as low as you can.

Now make a new image, and make a second layer in the image.  Make a scattered spray of black over the second layer, getting fairly even, but random coverage, then turn off the background layer.

Use the same method of Guidelines, Copy, Erase to reduce edges, than use the same idea as above to resize the image to reduce overlap of the edges.  Last save the image as a PSD (to preserve transparency), and with the background layer off, make the image into a pattern (Turf Mask v1.0).  Spray the mask into the same area as you had the turf, into a new layer.

Put all the layers into a Turf folder in Photoshop, with the mask sandwiched between the two turf pattern layers.

Now select the top-most turf layer, and click on Layer->Create Clipping Mask

This will tell the selected layer to be masked by the layer behind it.  This mask is not based on Black Vs. White, but opacity, which is why you had to make the background transparent for your masking pattern.

This finally gives you the final results you have been looking for:

A smooth, lineless, textured, natural colored pattern.  You can also look at the overall image here, and – using hue-saturation/contrast – tweak the colors until it looks a little better (This seems a tad brown and muted to me).  Write down what tweaks you made, reopen the saved patterns you made, and make the same tweaks, save the version 2.0 of each, and re-define the patterns as well.

These two patterns don’t seem like they would work together with the different darkness levels, but they are what gives you the following results:

While this process is time consuming and a pain, the key is to remember that from now on, all you have to do is paint an area with three patterns, make one a clipping mask, and your done.

Tune in next week for the next key landscape building block: dirt.





Tuesday Tutorial: Now in 3D!

25 08 2009

Now that you have a completed CAD base map, with all of the elevation edits needed, you have just one final step before you can Import your design into SketchUp.  If you used a construction line hidden layer, turn it on, and every other layer off.  Then delete all information off of the trace layer, and delete the trace layer itself.  The main reason to do this is because when CAD files are imported into SketchUp, if a line exists in the same spot on two different layers, the line will be assigned to one layer, and the other will have the area blank.

As a quick software note: One of the main differences between SketchUp standard and Pro is that while CAD can be imported into either version, you can only Export non-CAD files in Pro.  So if you are planning an rendering in SketchUp, or using photo editing software to tweak, either version of SketchUp will work fine.  However, if you want to export into 3D Studio Max for higher end light and material rendering, you will need the Pro version.

Now its time to fire up SketchUp and get cooking.

Step 1: Importing

Once you have SketchUp running, the actual importing of the CAD file is pretty straight forward.  Open the File drop down, and select Import.

ImportMenu

Now navigate to the area where you saved the CAD file, change the file type to import to .dwg, and import your file.  If you get a message about your CAD file being invalid, its possible that your CAD release is newer then your SketchUp, so just save your CAD file back to 2004 or 2000.

One of the nice, and not so nice, things about SketchUp is that it imports layer states along with the layers, so if you had all of your layers turned off when you saved your CAD file, they will all be turned off now.  To turn them back on, go to the Window and select Layers.

layerdropdown

This is your layer navigator.  This will be a window you want to keep open, as what ever layer has the radio button checked is the layer that what you are drawing will be on. Turn on all your layers, and then zoom to extents and select the top view, and you should see something like this:

newimportallon

Congratulations, you now have an Imported SketchUp model, its time to save, and move on to the topography.

Note: Save early and often in SketchUp.  For as fun, quick, and easy, as it is, it is buggy as all hell, and crashes fairly often, so save yourself the grief of losing 45 mins of work.

Step 2: Let the technology do the work

The first thing to do is turn off all your non-topography layers.  Once you do that, Take a quick look around in your model.  You should be able to get a good feel for the topography you have created, and the type of landscape you have.  You may notice a few areas that you want to tweak.  You should do that now.  I noticed, when looking out from what would be my train station, that the two hills that flank my view are lower then I would like, and that the topo is not going to carry as far out as I would like.  You can either jump into cad, fix the problems, and re-import, or add and move lines to fit your needs.

needtofixlines

You also may notice problems like I did, where I have a hill top that is supposed to have two peaks.  I made two separate contour loops as seen below (blue are the higher contours).

needtofixvalley

The problem is that SketchUp will simply create topo straight across between the two, as there is nothing pulling the form back down between them.  This can be fixed by quickly adding a line or two that will outline where you want the low point of the pass to be.

Now its time to have SketchUp do some work.  First open the Window drop down, and select Preferences.

PrefMenu

Then in the left side select “extensions”, and turn on sandbox.

ExtentionMenu

That will add the sandbox toolbar to your display.  ensure that you have all layers off except topo, and that you have the topo layer selected with the radio button, and do a quick save in case the program crashes during the next step.  Now click on the button on the far left of the toolbar- “From Contours”.

sandboxtoolbar

At the bottom left edge of the screen, there will be a progress bar.  Depending on your processor and the size and detail of your model, it may be quick to fill, or take quite a while.  If you notice the bar suddenly stop moving, don’t touch anything.  SketchUp is very touchy, and its likely that your computer is using so much power to create the topo that it doesn’t have enough left to move a —->—- further to the right.  I know, it sounds scary, but in SketchUp, if something starts to freeze up, don’t start clicking, just put the mouse down and go get a sandwich.  If you click you may cause a crash, but if you leave it alone, you have a good chance of the computer catching up with itself and then you can save ASAP. Once its done, turn on shadows to get a little better lighting, and it should look something like this:

newlytopoed

Step 3: That shouldn’t look like that…

Now take a close look at your model.  You should be able to see all on the topo lines you drew, with a surface in between them, but you will likely see some areas that look… off.  Two key things to look for are places where you see to many lines (Red), or topo lines disappear for no reason (Blue).

topoproblems

As you can see with the shadows turned on, missing lines come from an area where SketchUp arbitrarily decided to skip sections of topo lines and take a shortcut.  The extra lines happen when SketchUp forgets how to draw a flat surface in a small area, so it freaks out and starts throwing lines across the area.  Go to the View drop down and turn on “Hidden Geometry”.  You can now see all of the polygons created by the software.

Double click the topography you created, and you’ll notice that the hidden lines now highlight when you mouse over them as you are in component editing mode.  the first topo fixes you are going to do are the quick and easy ones.  You might notice an area where a topo line or two are covered, like this:

needtoturnhidden

Now you are going to use the tool on the far right in the “sandbox” toolbar- “Flip Edge”.  This takes a diagonal line and rotates it to the opposite diagonal.  Here is an animation of flipping the lines, one by one, to correct the topography.

turnani

Note: For some reason I cannot get the GIF to play, even though it does in the wordpress blog text window, so I don’t know what is going on.  Check it out here.

Once you go around and fix all of the areas you can using this method, you will have the more complicated work to do of manually adding and removing polygons, but that will wait for next week.





Tuesday Tutorial: Building your base

20 08 2009

When you start your 3D model, you start with a cad base, just like you have for your plan graphics and for your Bid Set.  However, there are some tweaks that need to be made, and things you have to pay more attention to then normal.  In this tutorial I’m going to take you up to the point of bringing your model into SketchUp.  Next time we will get into tweaking topo and base modeling, and then after that, the fine modeling level, and adding plants.  But for now, we are focusing on the base, as without a good base, the modeling process will be much more difficult and time consuming.

Note: When you use CAD, use Polyline, for the love of god.  It is only in EXTREMELY rare cases that line works BETTER than poly line, and this is NOT one of them.

Step 1: Clean up your normal CAD base

To start, open your normal CAD base and save a copy of it in a new folder.  Once you have your copied CAD file, there are a few fairly simple things you can do to clean up your base.  The first is to carefully go through and create new layers based on the different materials you will have in your model, and you should name them based on what material they will be to reduce confusion.

cadlayers

A key point here is to make sure you have a unique layer for each of your materials, as the layers will carry over to SketchUp.  Once you put everything on the correct layer, make sure all your corners are square and your polys close.  You are going to have to close them eventually anyway to make them into surfaces in SketchUp, so you might as well do that now in the friendly confines of CAD.

linebase

Step 2: Topography- think like a computer, one triangle at a time

Now its time to add topography.  There are two main issues you are going to have to address with topo: thinking about how the computer will create the surface, and thinking about making what goes on those surfaces.  This is the basic topo for the fictional park design I am working with.

topolinebasic

The first thing to realize is how SketchUp (or any other program) will create your topography.  As a quick intro/refresher- all 3d models are comprised of polygons- more precisely, triangles.  That is because 3 points makes a plane, and computers, as amazing as they are, cannot make true curves.  They can make thousands of tiny angles, but not an actual curve (there are ways to fake it more, but its still really a post production fake).  That is why when you make a curve in CAD and then zoom in it is suddenly angled lines that are kind of in the wrong spot.  Getting back to the point- your computer will make the smooth, rolling landscape by making a bunch of triangles and “softening” the edges, but it is still made up of triangles. These triangles are made by drawing lines from intersections on one line to intersections on another line.  For example, with these lines:

topoexample

This is (hopefully) where the triangles would be made:

topoexample2

This means 2 main things for your drawing.  First, and easiest: the further the camera is, the less detail that you will see, so by simplifying the lines that the triangles are made from, you reduce the number of triangles, and decrease rendering times.  So, when you are getting to areas that are far out of your model, and don’t need a lot of detail because they will act simply as context, you can often get away with making those smooth curves into sharp angles, and get nearly the same result for far less CPU power.

The second, and more tricky point: if you have a flat terrace, on your topography map that is shown by an area with a bigger spread in its contours, because we interpolate a steep slope suddenly being flat as essentially a level slope, even if the contours imply a 5% slope, because the slope is likely steep beyond the last “steep” contour.  However, the computer only sees what you put in, and cannot interpolate unevenly spaced contours, so it will have a slope that is steep from one contour the next, and then makes a hard turn to a 5% slope, and then a steep drop again.  Because of this, you have to edit your contour layer, and break one of the cardinal rules of topography- you are going to have a contour line split and then rejoin.  to create a truly level area you are going to have to surround the entire level area with equal contour lines, forcing the computer to see the contour lines your mind is reading between the lines on the map.

Level area the way its drawn on a map:

terracebadblank

The way that would be interpreted by a computer:

terracebad

The “fix” by breaking the rules:

terracefixedblank

The new triangle map (level area skipped for clarity):

terracefixed

Step 3: What is that path going to look like exactly?

The last big complication to 3d modeling on topography is getting your paths and areas that overlap topographic changes to look right on the slope.  The problem is, again, the fact that the surface of the ground will be made of triangles.  this becomes an issue because your triangles may not line up with the edges of materials.  There are a few things you can do to address this.  First, you could simplify your topography so that a minimal amount of it intersects site amenities, as I was able to do in my thesis project, as there are not many large grading changes other then wetlands.

university

At times, however, that is not a viable option, and you need to be able to control what your slope looks like on a given face.  To do that you need to ensure that a given section of slope is laterally level.  The first step for that is to edit the contours so that the slope crosses the area in a straight line perpendicular to the slope.  This is what my topography looks like after I add my loops for level ground:

curvingtopopath

You want to change all of the sloping areas here:

curvingtopopathhighlight

To look like this:

curvingtopopathfixed

Once you do this, you wait for further fixing until you are in SketchUp, and that will come next week…

Step 4: Elevate to a New Level

The next thing you have to do is elevate your topography and items to the levels they will be at.  In AutoCAD, you want to open the properties window for each contour line individually.  Once there, you will see an option for “elevation”.  This is where you input the height of a given contour line.  Go through each contour line, and object and assign them to the appropriate elevation.  I personally like to also change them to a set color, like magenta, so I know what I have moved.  Then once everything is adjusted you can select all and move the color back to “by layer”.

elevation

magenta

magenta3d

Once you have everything moved to the correct elevation, you can change the color back, and finish the fix for uneven paths.

Now its time to import into SketchUp and let the real fun begin…





Tuesday Tutorial Primer: So you want a 3D model…

11 08 2009

The first thing you have to figure out when you want to make a 3D model for a rendering is if you really want a 3D model rendering.  This decision is similar to the question of if you want a hand rendered plan or a Photoshop plan.  There are of course aesthetic reasons involved.  Both can work extremely well, and look amazing, but by understanding your client, your design, and the look you will get from either method of rendering you can make the choice of which is best for you on a given project.

There are more then just aesthetic differences to be looked at, however.  Ask yourself what the purpose of the model is.  Is it for images of a few conceptual designs, or for a final design proposal?  If you are looking to do more of a conceptual design presentation, then you may want to go with a hand drawing, as they are going to be relatively quick and easy, where a 3D model, even using software like SketchUp will be somewhat drawn out and difficult, as a lot of time will be spent re-making the same things multiple times in NEARLY the same way.  One of the reasons to go with a 3D model, in any level of design, is because of the ability to animate the design to increase the understanding of the place by allowing the client to “walk through” the site.  Also, if you think you may want multiple angles of a space, or think the client may want to see a few additional places later on, you may want to make a 3D model, as you can always add additional camera and have a new rendering for the cost of running one of your CPUs for a few hours overnight.  In the final design phase it may be far more worthwhile to create a 3D model as if a little extra time and the right software some pretty amazing images can be created.

Conceptual Hand:

sketchdemo

Conceptual SketchUp:

memorial

Final full 3D Rendering:

3DDemo

The third issue is purely a question of time: is it worthwhile to invest the time into a 3D model, or does it make more sense to do hand sketches.  One of the ways computer models are different from computer plans is that a computer plan is extremely easy to edit, giving itself a big advantage over hand graphics which must always be made from scratch, and many 3D models are difficult if not effectively impossible to edit.  Another time factor is the issue of the current hardware you have in terms of computers.  Go back and check out my post of computer hardware, and see if you have a suitable system for 3D modeling.  It can be done on almost any machine (I can run high end stuff on my 633 Mhz Celeron with 256 or RAM and a 32 Meg TNT graphics card if I’m feeling silly), but some machines will not only render slowly, but will slow down the creation process as well (See: Machine I just mentioned) to the point where you lose any hope of cost effectiveness.

There are some special uses that 3D models suit themselves to much more as well.  If you want to create a new element in an existing location, you can re-sketch the whole scene, or you could create the new element in 3D, and place it into the photo of the existing site.  This can be an extremely effective rendering, and isn’t something that has to take a lot of time to do.  The model for the following image is a fairly simple SketchUp model with careful lighting and camera placement to create the right angle and shadows, but was a very quick project for the quality of the end product.

bluff

In the coming weeks I will take you from a CAD plan to a conceptual model, and then off to higher end software such as Maya and 3D Studio Max for better textures, lighting, and reflections, using my new website as the structure.  For the time being, pick a small simple design you have, and, if you do not already have them, put the design into CAD software.  This CAD basemap will be the basis of our design, so close all your lines, and keep some decent layer organization.  Next week I will address the CAD file you will import into Sketchup, and in the weeks after is when the real fun will begin.