Start 2 finish

Not limited to only drawing, show us all your forms of art. From sculpting to photos, GIRAFFES ARE BANNED!
Post Reply
User avatar
kaos
Noble Warrior
Posts: 4089
Joined: Mon Jun 14, 2004 3:09 pm
Location: Atlanta, Ga

Start 2 finish

Post by kaos » Mon Aug 23, 2010 6:24 am

I started a new model just a few hours ago.
and I thought this might be a good opportunity to kinda show some of the process, for anyone interested.
So I'm going to try and post up each stage in the models development from the beginning to the rendered video.

First, inspiration...
I was watching Labyrinth(David Bowie) today.
and I liked the armored goblins they had in the film.
They looked pretty badass.

Second, research...
I basically Google image searched "Goblins" and got a whole lot of various results.
went back to the film to figure out what exactly it was that appealed to me.
it was pretty much that they were armored little people.
and you never see little people in armor...unless they're portraying Tolkien-esque dwarfs, and these guys clearly weren't. They were pretty unique looking.

Next, building the mesh...
usually at this stage you're meant to start sketching and then the mesh follows afterward.
but I'm going to go a bit out of order, as really all I know for certain at this point is what body type I want, and after later on I'll figure out the design I want to apply to it.

So to start with, I thought I'd build out the base mesh.
Image

This is roughly 1h:30m of time, made from scratch.
I'll likely just take the hands from another model and attach them here.
as for the head, I'll need to wait on the design I come up with first before I get going on it.

Next is design, but I'll do that next post.(in a few hours)
A theme of my video is "Rabbits." So I'll be thinking up something with that theme in mind.
So I've got rabbits, goblins, and armor to work with...

To be continued.
User avatar
Mik
Born under a bad sign
Posts: 3394
Joined: Mon Jun 14, 2004 8:44 am
Location: Beyond your borders
Contact:

Re: Start 2 finish

Post by Mik » Mon Aug 23, 2010 9:33 am

Image


Not sure if that's what your going for, only think I could think of for goblins in armour :P
User avatar
kaos
Noble Warrior
Posts: 4089
Joined: Mon Jun 14, 2004 3:09 pm
Location: Atlanta, Ga

Re: Start 2 finish

Post by kaos » Mon Aug 23, 2010 1:11 pm

I started with the helmet first. the helmets likely to be the goblins most defining feature sense it will be ontop of an oversized head. The viewer is going to need to learn most of what he needs to know immediately from looking at the head.
So the head is going to need to have the themes covered.

I looked at several kinds of helmets from knight helmets to motorcycle helmets, and i even looked at the general shape of a rabbits head, and even considered a rabbit skull mask kind of thing.

Image

I wanted my armor to feel top heavy, to give the goblin an excuse to really hunch over during the animation.
so I started thinking about American football gear, and how top heavy that tends to look.

Image

The thing to remember is to not get hung up on these initial designs. for several reasons...
1.Translating 2d to 3d isnt usually a flawless 1:1 process, so things might have to change in 3d to make it work.
2. its concept art. that means its more an idea than a strict blueprint.
3. In the 3d space I might see some new design opportunities that I may want take advantage of.
So int he end, the 3d and the concept might end up looking completely different from one another.
User avatar
kaos
Noble Warrior
Posts: 4089
Joined: Mon Jun 14, 2004 3:09 pm
Location: Atlanta, Ga

Re: Start 2 finish

Post by kaos » Mon Aug 23, 2010 3:47 pm

Here I've started blocking out the design.
Not really committing to anything right away, just trying to get a feel for it in a 3d space.
you can see a couple of design tweaks have already happened.

Image

next, I'll probably start detailing the model a little bit, still making minor changes as I go, and get it prepared for sculpting.
User avatar
kaos
Noble Warrior
Posts: 4089
Joined: Mon Jun 14, 2004 3:09 pm
Location: Atlanta, Ga

Re: Start 2 finish

Post by kaos » Mon Aug 23, 2010 11:37 pm

I'm still working on it...

but I thought a video might be a good progress update.

Click here to view the YouTube video. Click here again to hide it.
User avatar
kaos
Noble Warrior
Posts: 4089
Joined: Mon Jun 14, 2004 3:09 pm
Location: Atlanta, Ga

Re: Start 2 finish

Post by kaos » Thu Aug 26, 2010 12:31 pm

ok...
so I'm done with the details, everything's unwrapped now, (though not very neatly)
and I'm ready for the sculpting stage.

Image

but first...
Here's the model with a simple diffuse map.

Image Image

And here again with a "The Great Dictator" skin.

Image Image

and a turnaround video...
Click here to view the YouTube video. Click here again to hide it.


now some sleep
User avatar
Chewi
Anti-Hero
Posts: 3521
Joined: Sat Jun 12, 2004 3:51 pm
Location: Edinburgh, Scotland

Re: Start 2 finish

Post by Chewi » Thu Aug 26, 2010 10:32 pm

Hey, sorry I didn't reply to this sooner. I wanted to take the time to read it properly. I appreciate seeing this broken down. I've played with modellers but never really got much past step 1. This is filling in a lot of blanks.

In the last post, where does the first image come from? Does it give you a bunch of flat images to paint on? I know that's kinda how it worked for Quake 2 skins. Or do you actually paint onto the model more directly now?

Also, is a lot of it just block colour that's brought to life by the lighting? For example, you can see that his helmet has more than one shade of green but did you just apply a single shade and let lighting do the rest? This kinda feels like a dumb question but I'm really not sure.
User avatar
kaos
Noble Warrior
Posts: 4089
Joined: Mon Jun 14, 2004 3:09 pm
Location: Atlanta, Ga

Re: Start 2 finish

Post by kaos » Fri Aug 27, 2010 10:47 am

The first image is the UV MAP. you have to create it.
to do that you used to have to select all the faces you wanted and cast a projection on those faces.
then select adjacent group of faces and do the same, and then stitch the parts together to make a seamless map image that you can draw across. but Now you can pretty much just select where the seems are and have the computer handle the projection for you.
no matter what though, the UV map is absolutely necessary if you're going to do any textureing at all.

Theirs a few ways to do it, and the layout is pretty much up to you.

Heres a map of the earth you've probably seen...

http://www.mncourts.gov/documents/2/ima ... _globe.gif

This is pretty much exactly the same idea of what a UV map is and how it works.
here are two more in a different style. still the same planet, just cut and flattened out a different way.

http://www.cs.ucla.edu/~kohler/z/icsilogo/isf-n9.png

http://www.genekeyes.com/CAHILL-LMW/201-3.jpg

So yeah... Quake 2...and Quake 1 (I made a couple skins for that game.)
and every game since then actually. (actually that's not true, but I'll mention it later.)

After your UV map is created, your 'skins' are basically make a new image that uses your UV MAP as a reference.
In my image I have the UVmap on top as the highest layer. Its only visible while I'm working on the image in photoshop, purely for my own reference. Once I'm done I usually delete the UV layer

UV MAP
Image

Texture(skin) with UV MAP on top
Image

The final image that gets used.
Image



Can you draw directly on the model now?
absolutely. but even so you still need a UV Map, and you're still creating the map above.
you're just doing it a different way is all.
It's not necessarily "easier" but it takes pain out of matching up the seams of a model so that they aren't obvious.

I'd do it more often myself If I could find a program was robust and user friendly as photoshop, but did it all in a 3D space. Theirs lots a programs out there claiming to be just that. but so far none of them lives up to that claim.


So with the color...

Image

Yeah most of the value in it comes from the lighting.
I do have a couple of shades in my texture, but the lighting is giving me even more by taking the geometry of the model into account, and including soft shadows as well as hard ones.

Below is the same model from the same angle. the one on the right is affected by light, the one on the left ignores the light completely, allowing you to see the texture for what it really is.

Image

here is a shot of exactly what the light is doing...
Image



and just cuzz...
here is whats called Ambient Occlusion. (AO)
I havent used this on any images before now.

Image
Its more shading than lighting...but its looks really nice on most things.

You'll see this more in film than in gaming. its pretty hard on the machines.
However, for gaming, what you can do is make an "AO bake" which basically maps out the shading and outputs it to a texture...then you can combine that with your flat colors.
Its kind of a cheap hack of a trick, but that's pretty much the nature of doing just about anything for gaming. That's why theirs so many kinds of texture maps...so the machine doesn't have to spend much power if at all, trying to do this stuff on the fly.

in film, you'll see this one combined with the regular lighting, and then that combined again with texture maps.
basically you're not likely to make a conscious notice of it until its not there.

AO is probably what made UP look as awesome as it did.
Image
User avatar
kaos
Noble Warrior
Posts: 4089
Joined: Mon Jun 14, 2004 3:09 pm
Location: Atlanta, Ga

Re: Start 2 finish

Post by kaos » Fri Aug 27, 2010 10:22 pm

sculpting might take a while...
I'm going to have to sculpt most of the parts separate from one another.

heres a look at what sculpting does...


Image
This is the original mesh of the upper body portion of the character.
it clocks in at 844 polygons

Image
This is after some subdivision and a quick bit of sculpting on the core area...
it clocks in at 856,576 polygons

856,576 on half a character will never do for a character model in a realtime videogame.
So that's where Normal mapping comes in handy. because the general shape of each model is the same (and the UVmaps are nearly identical) I can capture the hi-rez model information as a texture, and apply that texture to the more practical lo-rez model.(normal map)
The game engine will then start reflecting lights based on the information according to the normalmap rather than the actual low rez-model.
Thats going to make the 844 polygon model seem just as detailed as the 856k model to the untrained eye.

But because I can actually afford more polygons than I have, I'm probably going to use a heavier version of the model for my low-rez that clocks in at about 3k polygons.
why? because the the 3k model is going to be a lot closer in shape to the 856k model, and it will be even harder for anyone to tell that I'm using a normal map at all.
Post Reply