Some Common Digital-Model Myths Debunked
I’ve heard a lot of rumors and advice over the years about how to make models and artwork. Some of it is very good, but some of it is either completely wrong or the advice has been passed from one person to another in a “I heard from it from a guy who heard it from a guy who knew a guy who had a cousin who worked at ILM” type of way. I thought it would be fun to try to debunk some common myths.
1) You need a lot of polygons for your model to look good.
I’m sure everyone has heard this statement at some point: the best models have hundreds of thousands of polygons because that’s the only way to get good detail. If your model doesn’t have tons of polygons then you’re not doing it right. Go back to your modeller and sub-divide that model a few times to raise the polygon count. After all, nothing tells the world you’re a professional artist making high-quality items more than making a model with a million polygons.
Before I begin, let me explain a little: if your model has a lot of little parts and details, then, yes, you may need thousands of polygons. However, my point is when modellers use thousands of polygons to create curved surfaces when they can make the same model with 1/10 the polygon count. For example, if your model is an airliner (which is basically a rounded cylinder with wings), how many polygons do you need?
Here’s an example. Look at this image and guess how many polygons this model has:
But, you’re right- it’s almost impossible to tell how many polygons are in a model once the image is rendered. (The answer is 2,460 polygons with a 512×512 pixel photo-realistic image as a texture map.)
Second, most software programs have a built-in smoothing angle, which means you can use less polygons and still have a smooth model. Here’s another example:
Here’s a quick quiz: when you make a cube, how many polygons should it have? If you guessed 6, you’re right. If you guessed 12, you’re using a software program that creates triangles, so you’re right also. If you guessed anything else, you’re using a software program with a sub-division option set way too high.
I see this happen all the time with various models and I’m not quite sure of the cause. Do modellers really think they’re “adding detail” by subdividing a cube?
You may not be able to see a difference, but the software has to load the model into memory and calculate all the polygons in the scene. This probably doesn’t matter if your scene only has one sphere or cube in it, but what if your scene has 3 or 4 of these spheres (or boxes) plus texture maps plus additional models, which have their own texture maps? Why bog down your system if you don’t need to?
As an example, I downloaded some Star Trek ships that were converted for use in Poser. However, one of the ships had over 850,000 polygons because every single detail was modelled, including the ship name and number! The artist was proud that he didn’t use any texture maps, but this made the model almost unusable since Poser became very sluggish when moving the camera.
Many software programs can use displacement maps to add a significant level of detail without using extra polygons. In fact, some artists are making super-detailed human faces from a model with less than 200 polygons!
2) You need huge texture maps for your model to look good.
And after you’ve made your model with thousands of polygons, it’s time to give it textures. As everyone knows, the best models have huge textures maps, and like most things in life, bigger is better. Why drink a 12 ounce can of beer when you can have a “yard” of beer?
A lot of people think artists need 4096×4096 (or larger) maps to make their models look good. In fact, I’ve seen some artists make 6000×6000 pixel texture maps that were basically a “sand” texture. And I’ve seen other artists include a 8192×8192 pixel texture map in their products! (As a side note, earlier versions of Poser can’t load images larger than 4096×4096 and many users don’t like being told to upgrade to the latest version.)
The first issue with using large image maps is that it requires the software program to load the image into memory before and during rendering. This could bog down a system with low memory.
But the main issue is: why are these large texture maps even needed? Let’s take a step back and think about what the typical user is doing with these models. I would guess that a vast majority of them are making images and animations with the models for their own use or for posting in online galleries. If this is the case, then their images need to be small enough to post to these galleries. And if the image is going to be small, then the model (and the textures) will be even smaller.
Here’s an example of my own:
This is an 800×800 pixel image of my Tabby character (click the images for a larger version). She’s wearing clothes, so only her head and hands are showing (her legs are covered by a “stockings” clothing model). Her head measures roughly 80×80 pixels out of the entire image, yet there are many people who still insist that textures for human figures should be 6000×6000 pixels or more.
Additionally, even though you may design your texture maps to be large and highly detailed, how does your texture look after the artist exports their render to a compressed jpg image? And then how does it look when a website further compresses the image to display in their gallery?
But, wait, you say, what if someone like Pixar buys my model and uses it in their movie and the movie is shown in IMAX movie screens? That’s why I need to use 8192×8192 pixel texture maps. First, I hate to be the bearer of bad news, but Pixar isn’t going to buy your V4 “sexy lingerie” outfit from your small marketplace site to use in their next movie. For one, they don’t even use models like that, and more importantly, they make their own models. That’s right- most large studios like Pixar have teams of modellers and texture artists working on their projects so they don’t need to buy models from online marketplace sites.
Second, why are artists designing texture maps based on the minuscule, extremely low chance that your model will be used by Pixar? Why not design texture maps that are small and efficient and that don’t cause people’s computers to bog down?
Third, Pixar (and most other animation studios) have learned how to use shaders, specular maps, ambient maps, ambient occlusion, and displacement maps to give their models a higher level of detail.
Now, I’m not saying you shouldn’t ever use large texture maps, but you should think about whether it’s really necessary. Are you really making large textures that have a lot detailing or are you making large textures because you think this is a way to show off?
3) What’s the best modelling program?
This is an easy one. The best modelling program ever created is called “John’s Model Maker”. It’s $15,000 and can make models instantly. And coincidentally, I’m selling at my site. I made it myself, I get a cut of the sales, so that’s why I think it’s the best thing ever made.
In all seriousness, there is no “best” program. Like anything else in life, what does “the best” even mean? How is something the best? Is someone saying it’s the best because it’s their favorite?
Do you want a modelling program that’s easy to use, but may not create efficient models? Or do you want a program that requires a little more learning, but makes optimized models? Or do want a modeller with integrated animation or character design tools?
And what’s the end purpose of your models? Do you want to export them for use in other programs such as Poser or Vue? If so, you need to make sure the model is “clean” (no 2 sided polygons, no polygons with more than 4 sides, etc). But if you’re only using your model in that modelling/ animation software, then this isn’t as much of a concern.
If you ask this question in the forums, everyone will tell you their favorite software, but how many people will list the features and benefits? In fact, how many people will even ask you what you’re trying to do?
Personally, I use Lightwave to make my models, but a lot of people love 3D Studio Max. But there are also plenty of free and low-cost programs out there, like Blender and SketchUp. Each software has its own interface and learning curve, but they should all have plenty of online tutorials (or books) to help you get started. You could also go to the companies’ websites, find a 30 day free trial, and try out the program to see if you like the features.
4) This image took me 50 hours to render!
After all, nothing tells the world you’re a professional artist more than waiting two days for your image to finish rendering.
Usually, when an image takes that long to render, either the image size is too large (for example, you’re rendering a 5000×5000 pixel image) or your render settings are too high.
Usually, a setting of 75% to 80% max quality is more than enough for an image that will be posted on a website. Plus, there’s no need to render an image at 5000×5000 pixels when you’re going to reduce it to 800×800- remember, no website is going to accept a 5000 pixel render. But even if a site did accept it, why do you want your viewers scrolling all over the image, when they’ll see less than a 1024×768 section at a time? Don’t you want your viewers to appreciate the entire image?
See my previous blog about making images for the correct size: Designing Artwork for Online Galleries: Image Size and Your Audience.
5) Can you tell me how to make conforming clothing in Poser?
Sure, no problem. Just follow this simple chart:
In all seriousness, making conforming clothing in Poser is probably one of the biggest challenges in the digital model field. It’s one thing to make a car model with rigged wheels or a figure in Studio Max with a rigged skeleton, but Poser clothing requires modelling, joint editing, and more. After all, you’re making a model that’s designed to go on top of another model and follow the bending and posing of the model without showing any breaks or seams.
Here’s the short explanation:
The obvious first step: decide what kind of clothing you want to make. I suggest starting with something easy, but if you want to make sci-fi fantasy armor, then go for it.
1) You’ll first need to make the clothing model in a modelling program. And, no, you can’t make a model in Poser since it’s not a modelling program.
– See above for the “best” modelling program to use.
– You’ll need to import your target human figure and build the clothing around it.
– You do have experience making clothing or organic models, right?
– When you’re done modelling, you’ll need to export your model to obj format for use in Poser.
– Since you built your clothing around a Poser human figure, your model should be the correct size for Poser, but you may need to check the scale.
– Don’t forget that all obj files for use in Poser should go under the Runtime\Geometries folder.
Note: you’ll need to cut your models into “groups”, such as the hip, abdomen, chest, legs, arms, and so on. The step-by-step process to do this is beyond the scope of this article, but you can either assign the group information in your modelling program as you make your model or use another software program such as the AutoGroup Editor.
3) Create the Poser figure file (a cr2 file) with the joint information.
– Import the obj file into Poser and either rig it using Poser’s Setup Room or using a “phi” file.
– Or you could copy the base figure’s cr2 file and change the reference to use your obj file.
Note: the process of creating a figure file is also beyond the scope of this article.
4) Test in Poser.
– Load the human figure into the scene.
– Load your clothing into the scene and Conform it.
– Pose the human figure in various poses to make sure your clothing model doesn’t break or that any parts of the human model can be seen through the clothing (this is called “poke through”).
– If the model does break or if there’s poke through, you’ll need to adjust the joint centers or falloff zones.
– If there are any “spiky” polygons, you’ll need to use the Joint Editor to adjust the innerMatSphere so it includes the “spiky”.
– When everything works properly, save the model to the Figures library… making sure not to save the human figure with the clothing.
Note: the process of adjusting joints and falloff zones in Poser is beyond the scope of this article.
5) UV map and make textures.
– You do know how to UV map clothing, right?
– You do have a UV mapping program, right?
– You do know how to make clothing textures, such as fabric, leather, or plastic, right?
– You do have an image-editting program, right?
Note: the process of UV mapping and creating textures is way beyond the scope of this article.
6) Test in Poser again.
– Load the model back into Poser and colorize it: either attach the texture maps or use shaders.
– Pose the human figure in various poses to make sure the textures look good and don’t stretch or distort.
– When everything works properly, save the now-textured model to the Figures library… making sure not to save the human figure with the clothing.
7) Add morphs to your clothing.
– This step is actually pretty easy since you can use a software program like Morphing Clothes to copy morphs from the human figure to your clothing.
8) Test in Poser again.
– Load the human figure and your clothing model back into Poser.
– Adjust the morph dials on your clothing model to make sure it doesn’t break or that any parts of the human model can be seen through the clothing.
– Though if parts of the human figure can be seen when you adjust a morph dial on your clothing, simply adjust the corresponding morph dial on the human figure.
– If you made any changes to your clothing model, you can save it back to the Figures library, otherwise you’re done.
And, there you have it- you’ve made a conforming clothing model! See, it’s pretty easy after all.