Back to building a 3D dinosaur - research

Our subject in the Field Museum, Chicago.

After a brief hiatus I've managed to post the latest in my occasional series on building a 3D dinosaur over at  Paleo Illustrata. This post outlines the research involved before preparing to model a Triceratops in 3D; some of this will be old hat to may of you but hopefully it will provide pointers to anyone wanting to have a go at creating ancient life reconstructions on the computer. This subject is already being covered by some fine artists and inspiration can be gained by swinging by Angie Rodrigues' brilliant blog where she's modelling an Olorotitan which is setting the standard (gulp!) by which all other 3D modellers might be judged, and also have a gander at thoracosaurus.blogspot.com where Evan Boucher created an animated Thoracosaurus neocesariensis for his masters thesis and is a accomplished piece of reconstruction complete with the methodology.

What differentiates all of these blogs is that there are as many ways of constructing a workflow, methodology and artistic experience and creativity to create 3D reconstructions as there are artists, so hopefully I will be adding to rather than reproducing the work of other workers. As ever, comments very welcome!

Making a Dinosaur in Carrara (Spline Technique)

So after having requests from a few people asking how I make my 3D Dinosaurs, I've finally made the time to do a quick series of tutorials. To be honest overall these will probably be of only direct use to Carrara users, but for you general readers this will give you a bit of insight into how I roughly make my 3D graphics (I don't use a traditional format afterall).

I am using, again, Carrara. At moment I'm on the latest version, number 8 (which they have finally fixed the majority of bugs on... after a mere 9 months!), but my tips for today are good for any version of the program (and even its ancestor software Raydream).

Where I am a real rebel is I will be showing you how to make a Dinosaur today with a Spline editor. Most 3D people make organic creatures with Vertex or Metaball methods. So if you're looking for insight into those realms I am not the guy for you.


So what is a spline object?

A spline is an "object" created by drawing several cross sections along a linear path, and filling in the spaces between the cross sections with "mass". While this might sound complicated it is actually rather simple.

Using this simple cylinder above as an illustration, the cylinder is formed by a single circle cross section which the computer stretches along the line to create the remaining mass. If you were to insert additional cross sections and change them from simple circles you can make your arm (as an arm is basically a slightly distorted cylinder).

If this still doesn't make sense, hopefully it'll make more sense as I jump into how I make my Dinosaurs.

So Carrara people, you're going to have to create spline objects to follow my technique. The necessary icon is show above when starting in the overall modelling window of a new project.

TIP: Try to drag the spline tool towards the middle of the modelling grid so you're object isn't huge or way off the modelling plane.

Once you create a spline object you should pop into here, the spline editor. Welcome to my workshop of wonders...

If you want to model in here you'll have to learn and master the use of the extrusion path (those purple lines you hopefully can see on the grids above... sorry the picture was shrunk in the upload).

Taking a closer look at one of the extrusion paths (there are two out of the three possible planes. Here I'm looking at the Y path... you will also in the long run want to get to know the Z plane path as well), the key thing are the extrusion points. Those are the black dots you can see on the purple line.

Whenever you create a new Spline object you will start with a basic extrusion path. This has a cross section point (red arrow) and a extrusion point (blue arrow). You are not restricted too, nor will you typically use, just one of each. You can have as many of either as you want. Though the more points you add the more difficult it will be to manage certain attributes on your object (but too few will make your life difficult with the other attributes... there is always a cost to these things).


What is the difference between a cross section point and an extrusion point?

The main difference is that a cross section point will more define your objects make up and look, while an extrusion point is a means of subtly influencing and tweaking the object. Both are necessary, but cross section points in the end are the most important.

So let's start with cross section points, as 1. their the most important and 2. you can't start really creating an object without cross sections.

Again a new Spline object starts with a single cross section point at the back of the extrusion path.

Tip: Sadly there is no display difference between a cross section point and an extrusion point. They both are displayed as black dots. You'll just have to try and remember which are which. If you loss track (don't worry we all do) swap to the director's camera point of view which will let you easily and quickly see which are which.

Selecting one of the shape primitive options from the top icon bar, draw an initial shape on the cross section.

If you stick with the initial format of a new spline, you're object will be determined from its back end. While in some cases this might be handy for some types of models, in my experience it is not so good for organic parts.

We're going to want to customize your initial setup.

Grabbing the add point tool, shown above, you can click on either extrusion path (side or bottom) to add a point. Remember to press the ALT key if you want a cross section point. Just left clicking with the add point will give you a standard extrusion point, NOT a cross section!

For my technique you'll want to add a cross section roughly in the middle of the extrusion path. Than as close to the original cross sections end of the extrusion path you'll want to add a extrusion point.

That should give you an object like this. Note how I now have 2 cross sections (both circles, but this will not be the case for long) out of 4 points. Those other two points, while not defining my objects overall shape, will be important to close the object with a solid end point.

Next I personally like to get rid of the cross section at the end of the path. The reason being that cross section have a greater effect on the objects shape, and in the ends I like to have a more easy and subtle means of controlling my object.

So select the remove point tool from the top and click on the end cross section to remove it from the path.

This now leaves us with a perfect starting point for ANY body part of a Dinosaur you might want.

We have three points (for now). A single cross section in the middle (marked by the red arrow) and two extrusion paths (blue arrows).

Again so you understand what is happening the cross section (red) is defining the extrusion points (blue), and at moment as I haven't modified any attributes of the extrusion path or added any more cross sections I'm getting a perfect cylinder.

Let's start changing this into something more Dinosaur like.

First we're going to want to change our cross sections. I personal like to start with a single customized cross section (as opposed to a generic circle of square), and work off this on all new cross sections.

Tip: Remember that if you add a new cross section when there is another predefined cross section on the path, that your new cross section will be a copy of that original cross section's shape. You can use this to your advantage with some quick advance planning.

To customize a cross section shape I personally start off with a primitive shape (typically a circle). I will than ungroup the shape to allow me access to its controlling points. To do this select the shape by clicking on it, than either by going into EDIT and choosing Ungroup or simply hitting Control U.

You will suddenly have a number of points around your object and their vector handles. You can move the points and vector handles any way you'd like around the cross section plane. So any possible shape you can imagine is possible... With enough time and play.

The shape points represent connecting points of the perimeter of the shape. The vector handles control the curvature (or lack there of) of the lines that connect the points.

So to easier see how we're changing a cross section we'll want to switch viewpoints. Selecting our view point option pick either Front or Current Selection to get a clear view of a cross section.

TIP: Try to do key cross sections first, as the more you add the harder it is to see them from the front, as Carrara stacks them all in this view. Meaning that it will get more difficult to tell layers apart especially when you're making a very linear object like a Dinosaur's body part.

Okay so here is your cross section up close and personal. Again this is the best when to manipulate a cross section. At least at the start of modelling.

To more easily customize our circle we're going to want to add points. For this select the add point tool again, and add some points to your circle where ever you are going to want to pull, distort, or reshape the cross section's shape.

Pulling the points around the cross section plane, and modifying the curve with the vector points you can start to make Dinosaurian looking cross sections.

You will have to play a little bit with this part to get the hang of it. I could do a tutorial section on it, but in my experience the only way to really understand this is to do it yourself. So enjoy, and if you hit any real snags let me know in the comment section and I'll try to help you out.

TIP: When finished reworking a particular cross section go into Section and select Centre. This will centre your cross section shape on the cross section plane, and make it easier to find and line up other cross sections with. You can also regroup a Cross Section by hitting Control-G to easily drag the shape around the cross section plane (as opposed to have to select EVERY point while an object is ungrouped).

So once we have modified our cross sections the way we want, how do we modify the overall extrusion path?

To do this we need to pick an extrusion envelope which gives us a means of controlling the way cross sections and/or extrusion points connect to one another.

Click on Geometry and select Extrusion Envelope. You have several options. Most of the time you'll want the Free mode, but to initially start off with say a symmetrically body part like the top profile of a head you can pick symmetry in plane.

In any envelope your points will suddenly appear above and below your extrusion path alone both extrusion axis. The points will all be connected by blue lines. This is simply a visual means of seeing how your objects points are connecting to each other.

You can grab any of these points and move them along the extrusion plane. Depending on the type of point and the envelope you have selected different things will happen.

Moving a Cross Sections point here will cause every non Cross Section point connected to it up until the next Cross Section to move with it in proportion to how much you move that one Cross Section. If you are in a symmetrical envelope the other point(s) of that particular section will move accordingly.

Again testing out and playing with this part of splines will clear up any questions you have.

Lastly you can curve and smooth out the connections between points. To do this you'll need to select the vector tool as shown above.

With this you can give any point along the extrusion path vector handles that will allow you to curve the connection lines and thus curve your object.

Applying multiple cross sections along the path, modifying their shapes individually, and finally editing their connections along the extrusion path is exactly how I create and modify my Dinosaur parts.

This just requires multiple rounds of all the steps I just outlined for you in this lesson. If you found this useful or interesting let me know in the comment section below, and I will do more if people want more.

Till then happy 3Ding, and good luck with your own three dimensional prehistoric creatures!

potential of 3D; skeletal reconstructions

The graphic skeletal reconstructions done by artists such as Gregory Paul, Scott Hartmann, Archosaurian and Dinomaniac has always inspired me in a way that life restorations and scenic representations - as beutiful as they are - can't achieve. Creating a 3D reconstruction from these however is often frustrating. It reveals the gaps in my knowledge and often even the gaps in everyone's knowledge. So inspired to create a cleanly graphic skeletal reconstruction in 3D that offers another dimension's worth of information but without proporting to be more than what it is... a reconstruction. My goal was a clearly stylized reconstruction that 1) looks good and 2) presents the information available (but not more) in as clear a way as possible.
As the PrehistoricTimes theme is Stegosaurus and a good friend of mine's daughter loves exactly that dinosaur, I thought - I'll give it a go. 3 days later I present you with the results.

I already see loads of things I want to tweak, but I want to share this exploration and get some feedback. There are two stylistic themes, one based on the graphic skeletal reconstructions I linked earlier - and I think my favorite. Very graphic and bold. But the second has some merit as well... volume is more readable. Like? Don't like? Improvements?

Here's a close-up of the skull, mouth open. The teeth fit right into each other in the shearing motion proposed in literature. This promises to be a great way of communicating the differences and relationships between clades, and I can't wait to try my hand at animating this guy, then revealing the underlying skeleton. Of course, it needs to be rigged first....


note: put up some thoughts about the npr nature of this at my blog.

me, 3D and Plateosaurus...

Hello Art Evolved! David here... I'm new to the community and new to paleoart. I thought I'd introduce myself by explaining what I see as the potential of this community and how I hope to contribute. Thanks to the admins: Peter, Glendon and Craig for welcoming me here. I know I don't have much to show yet.

I'm excited to be here because I hope participating scientists will speak up and say what's not realistic, what's speculative coolness and what's outrageous. And other artists too. Don't hesitate to suggest things or make requests! Personally, I'm not interested as much in achieving consensus as I am in visualizing alternative possibilities. For example, here's my go at plateosaurus for the challenge... I'm actually narrowing in on panphagia protos, but more on that later.

It's a wip... of course. And I'm constantly deciding things that I have insufficient knowledge of. ie.: lips. Craig's been great informing me about lips in theropods (I figured if they're such close relatives of sauropods, they'd make for good reference) and how to recognize their presence from skulls. Thanks! But I'm still not decided, so I did two versions:

(note: that's an animated gif... it should be showing the mouth in 3 poses)
What do you all think (especially you lurking experts)? If anyone has further alternatives to try out, let me know and (time allowing) I'll work 'em in.

More about me:
I'm a 3D artist specializing in non-photorealistic rendering... or npr. I see lots of potential for paleoart, as 3D has lots to offer - such as solid volumes. A render will create consistent forms and an organism can be volume-checked from different angles and eventually animated. But photoreal renders are very expensive in the sense that you have to invest in level-of-detail across the board - not just where you're interested - and there are considerable skills involved (modeling, texturing, lighting, rigging, animation) that can overwhelm an individual artist. The results also lay undo claim to being "real" - which can hardly be the goal of a scientifically respectful approach to paleology. While I love the experiential approach to many documentaries, I hope to make a hand-drawn alternative viable. An early test (low lod) can be seen here.

If you're interested in NPR, check out my blog at drip.de. I use Wordpress there and some of the formatting here is different. Tips how to create a fold, for example, are very welcome. Some commercial work can be seen at brainpets.com

Making a 3D Pterosaur Flap

Despite not being overly happy with my piece for the Pterosaur gallery, it did result my in learning a lot about posing a non linear 3D structure. Or in other words, it was really hard to make the wings flap!

Here is the finished piece, for reference.

So you'll notice that each Eudimorphodon is in a different and unique pose from the others.

They didn't start this way!

`

This is what and each of the three looking like in the beginning. Flattened and rather boring. I built them like this so I had a nice default model to start with, and thus not have to rebuild each individual Ptersaur from scratch.

The simpler solution to posing three different Pterosaurs would have been resculpted this model three times. Now I say simpler, in the sense of 3D knowledge. All you have to able to do is create objects in the program, as opposed to knowing the tools and options the program lets you use on these objects once they are made.

The resculpting approach is a lot of work, and would leave me with only the three poses. I'd have to repeat the process again for each and every new pose I might want in the future.

Making this resculpting option even more unattractive for a Pterosaur are its wings. Being made of both the arms and wing membrane, sculpting and modelling these two to match was a pain in the neck in the flattened pose (getting their front curvatures exactly the same in particular). I don't even want to think about having to worry about the bend in the wing too!

Lucky for me there is pretty powerful weapon in my 3D arsenal that was going to make things easier. A 3D "skeleton" rig...

`

Before I jump into the complex problem that was a Pterosaur wing, I'll give you the quick 101 on rigging (I personally have only just graduated to the 200 or 300 level... I'm aiming to enter my post grad level by this time next year :P).

Let's say I build this nice Dinosaur leg... funny enough I actually did, come to think of it ;)... The leg itself and the individual toes are all solid pieces. Apart from where the back end of the toes meet the ankle there no inherent joints in this model. Meaning as is there is no way for me to pose anything other than which way the toes will be pointing from the ankle.

Here is rendered, but it does not look very life like.

`

I could manually resculpted the pieces of the leg to simulate joints. Do not mistaken me though. This is not to say my model actually gains joints out of the process. Rather all I've done is go into the legs model and put a kink in it were I think a joint should be.

`

As this is not a real joint, many things happen. If you look at where the knee and ankle joint bend they just just don't look right. Unless you spend a lot of time with this sculpting a leg is not going to look like it is bending properly.

`

Another problem is that when I make my bends I can easily lose the length proportions I established on the unposed version. Meaning that if I'd put in proportions I'd measured from references (which on this model to be fair I just eyeballed... from a skeleton mind you, but I didn't precisely measure) my model can quickly become inaccurate.

Lastly my detailing doesn't move with the resculpting. So all the individually modelled toe scales don't follow or reorient with the leg and toes when I repose them. Meaning I have to go in and manually reposition each and every single one of them back onto the leg and toes...

`

Which adds up to a lot of wasted time, when you consider there is an "easier" way. Well easier for linear things like legs anyways.

`

That is of course skeletal rigging. What it does is allow you to define specific points within a 3D object that now act as points of articulation and rotation. In short it creates a real joint.

`

Each of the blue diamonds you see in this screen shot are called "bones". It is better to think of them as bone ends though. The diamonds becomes the point of rotation, and are where you model will bend at once the rig is "linked" to the objects. The lines connecting the diamonds function like real bones in that they are in movable (beyond the diamond points of rotation) and are solid.

`

Bones come with their own costs though sadly.

`

The first and foremost is that you can no longer modify or resculpt any 3D objects connected to the skeleton. Meaning you have to make sure the model is exactly the way you want it before rigging it.

The other big setback is that rigging a model eats up a lot of computing power and memory. So not only does it take longer to do things within the program, but your model tends to triple in file size.

Despite these inconveniences they make posing animal models SO much easier in the long run!

Once applied to a model you simply have to grab one of the diamonds and you can instantly and conveniently bend your model at the desire points.

Bones also simulate the manner in which you body moves. If I swing hip bone then everything connected below it move... Though if I just move the ankle then only the toes should move with it.

After rigging and posing we get something much more life like!

Skeletal rigs take care of all the problems I mentioned before.

The first and foremost is the ease of bending the model. Resculpting each piece takes forever! Plus if you don't get it right you have to start all over again... With the skeleton you can create a new pose within a few seconds!

`

Another huge difference, which wasn't obvious to me till I made the change over, the limb proportions are strictly maintained. When manually changing a model it is easy for you to get the length of limb segments wrong when their angled. Just compare this rigged leg to my resculpted one from before you'll see how wrong I was when trying to rescultpt the ankle.

`

Lastly all my detailing are locked onto and part of the whole rig. So the leg and toes move their detailing claws and scales move with them.

`

However if you look closely at the killer claw on this particular render you'll notice the one glitch I'm fighting to still overcome.

`

Bones and their points of rotation are fantastic on single linear objects. However when they interact with single small or non linear objects they have a tendency to stretch them in none realistic manners. It makes sense in the computer world, but takes some retooling of your logic as it is completely at odds with what you are used to in the material world.

`

My Pterosaur become the greatest challenge to my mode of thinking 3D vs. real world. When you think about a Pterosaurs wing membrane it is about as non linear an object as they come!

`

A Pterosaurs wing would have been an elastic object, and thus would have been effected and moved by many conflicting parts of the body (unlike the nice linear leg, the knee bone to the ankle bone connected to the toe bone etc.). Sure the majority of the wing should rotate up and down with the arm. However what about the bits that connect to the side and the leg?

`

Above is the first rig setup I tried on my Pterosaur, wishfully thinking that the wing membrane would behave like a perfect elastic if I merely defined the arms natural range of motion. Sadly I was about to learn in the virtual world you have to define a lot more then just the "real" points of rotation in a creature.

`

Not defining any "bones" in the membrane meant the computer could just bend and stretch it wherever it was easiest to do so. Resulting in this bizarre tenting effect.

`

Lesson learned. I was going to have to stop thinking of the wing membrane as real life elastic substance, and more as an extension of the wing through which the skeleton was going to have to expand and define.

`

I had my fears that this was going to get complicated (I hate it when I'm right, but we'll get to that in a bit). Just so I would knew, I tried a minimal fix to the problem.

`

I tried a single extension straight out from each major joint (except you'll not the top of the wing flap... which is a preview to how I was going to have to solve this) to see if it would define the bending and stretching seams for me.

`

The reasoning behind my logic is that each of those lines connecting the diamonds is an unbendable unmovable (except through rotation) section of the 3D object. If I bent the mid wing joint my hope had been that the tenting of the first rig would be force onto my now defined unbendable seam. In other words the tenting wouldn't look weird, because it'd be on the point where all the bending should be.

I was half right it turns out. Yes the wing bent perfectly at the midwing seam both on the arm and by bone extension all the way out to the edge of the membrane at the middle of the wing. As the computer didn't have any definition for the surrounding membrane however it found it easier to bend the wing bend just outside where I'd rigged!

Well the solution was "simple" enough, in logic, not in action. I was going to have to completely define and rig the whole surface of the wing membrane. Resulting in a rig that didn't so much resemble a vertebrate animal, but rather a weird exoskeletoned creature of some kind. If you were to think of all my 3D "bones" as a real skeleton that is.

The trouble there was I'd only been worried about the tenting and distortion with the wing in relation to the arm moving up and down. I hadn't paid any attention to the membrane's interaction and manipulation where it linked to the legs and body...

`

I was getting terrible distortion.

`

So once again I had to go in and add a dozen branching off membrane "bones" to help define and control the distortion along the legs.

In the end I had a pretty good "simple" rig that allowed me pretty good fluid control on my Pterosaur. I say simple in its use. I only have 5 major points (outside of posing the toes) to worry about.

It wasn't perfect (as none of my 3D skeletoning jobs are... remember I'm only a beginner to intermediate in their use so far). I couldn't bend the legs too much or the tenting would come back along the back edge of the wing. I also found that if I curled the inside toes to much a really weird tug line would appear all the way from the toe up to the chest...

`

So I had to be a little restrained in my poses.

`

Despite all this, and the mediocre piece it produced, I did learn a TON about my program's 3D rigging!

As a very quick aside, the inside joke of the photo these Pterosaurs are flying over is that they are fishing over top of some petrified trees. This is a photo of a famous petrified forest here in New Zealand.

There is a long fallen fossil log running between the far left Pterosaurs, a circular tree stump just above the bottom Pterosaur's left wing tip, and another circular stump in between the top two Pterosaurs in the upper middle.

Making of Karoo Sunset

As you've no doubt noticed ART Evolved has started to be consumed by excitement over the upcoming Pterosaur Gallery. For really good reason too. Pterosaurs have always been a huge icon of prehistory. So be sure to get started on your entry for the gallery today!


However I'd like to return to the Synapsids one more time, before it is nothing but flying reptiles around here ;p


Following Zach and his examination the creation of his own entry, I wanted to similarly share the process behind my own Synapsid piece. Where Zach approaches his with some regret (I personally think he is being too hard on himself, but I know the feeling. We are often our own worst critics!) I come at my own much more pleased.



The piece in question is this, Karoo Sunset. I aimed to have a picture that looked like a snap shot of the life in the day of. Unlike usual though, I actually ended up with something resembling what I'd intended.
`

After reading Peter Ward's book Gorgon (my brief review here) I felt inspired by the late Permian and the creatures from then we find in South Africa (the Karoo region in particular). Among the few well known and famous animals of this time and place were the Gorgonopsids, so naturally I wanted to recreate them.
`

I had in my mind a scene with a pack of these animals settling at sunset (though there isn't any real evidence I've read about indicating they were social). In that initial imagined state I imagined the piece as a lot more close up and tight then it ended up.
`

However despite it "opening up" all the key elements ended up in the final piece. An animal nawing away at the day's kill, youngsters tumbling around in play, several starting to doze off, and a larger adult showing off its freightening teeth in a big yawn.
`

Though the scene worked out well, I certainly didn't have a plan that lead me to this end product. In fact to be honest I mostly fluked my way there...



Of course the first step was to build a Gorgonopsid. As I work almost exclusively in 3D CG, I'm quite literal when I say build.
`

I was thinking about ART Evolved while building my Gorgon's head (the first thing I tend to build on any animal), and so documented nearly every major modelling step I took to create the skull and jaw. You'll find it all compiled into this little slideshow.
`

This was in total about 45min to an hour of work. It is kind of depressing that it all compresses into 30 seconds so easily...
`

Sadly Gorgonopsids being Synapsids have fairly boring skulls from a construction point of view. They don't have the fun holes and openings of reptilian skulls like Dinosaurs. So if people find this animation neat I'll record the next Dinosaur I build so you can see how a more complicated build comes together.

This one had only one minor 3Ding drama, that being the eye placement. You can see me correct this around 21 seconds.


Anyways with the beast built now I needed to texture it...

This ended up being the most difficult phase in the end.



My first problem was I couldn't decide on what I wanted it to look like. Was it going to be scaly with just some tuffs of fur, or was it going to have a big woolly coat.
`

In the end due to my embarassing attempts at scales with tuff of fur, I opted for just a hair covering.
`

With just a hair covering, creating the baseline model texture was stupid easy. Just this simple brown with black random lines, to simulate a base coat of fur, done on a much larger scale.


However this basline was going to need some 3D fur to make it looking convincing. Enter the HUGE problem. Fur...

Now as readers of my other blogs know, I've been doing a lot of feathered theropods lately, and these have really brought some of my 3D Dinos up to a new level. However you'll also find that despite my love of them in the the final product, feathers (and in this case fur, which is just a slightly different application of the same process) are the bane of my existence!
`

This was among my first attempts at applying hair to my Gorgon... A very cool effect (one I'll be stealing for any woolly animal galleries we do here in the future ;p) but certainly not a realistic looking Gorgon!

Why the crazy non-fur looking fur?
`

I'm using a feature of Carrara called Surface Replication. In it I select any smaller objects (in this case the different individually modelled hairs you see floating above the Gorgon) that I want copied along the surface of a larger object (in this case the various parts of the Gorgon). This feature in theory is incredibly handy, as you can make up to 10 000 copies of the smaller shapes with this utility. Individually placing and modelling that many hairs would take me months (and I just won't do it!).
`

However sadly it isn't as easy as you'd hope. There are a number of factors making it difficult. The size of the hair in comparison to the body ended up being a critical issue, and one I'd hit in a really hard way putting my scene together... I'll get back to that soon.

The thing that bugs me the most about this application is that whoever programmed this feature didn't think about the orientation controls. In theory the fur on my beast should align to the master object (again floating about the Gorgon), however somewhere in the math of this process all the X, Y, and Z attributes swap. Meaning that how I align my fur in the "real" world isn't how they align on the creature.
`

Which is why he appears so shaggy in these early attempts. Fur (or feathers) that stick out make your critter look unkept and unhealthy. So it'd be nice to have intuitive control on these things to slick them back...
`
So after playing with the size to body ratio, and the orientation of my fur...
`

I finally got this final version (well okay close to final. I didn't save a shot like this of the final final model).

I went for mostly subtle sized hair, but a larger thicker "mane" around the neck. In this pic it doesn't look so good (as it is also one version before my final one), but in the final piece the manes on the adults worked.

Next I came to where I was going to stick my critters for the scene. The Karoo 250 million years ago was a semi arid desert. Once it'd been a lovely productive forest, but with the number of converging geologic factors caused by the formation of Pangea, the world was slowly being turned into a giant desert...
`

This meant that I didn't just need a desert shot, but due to my desire for a sunset, a desert in sunset. As I didn't have any photos that matched this description, it meant I was going to have to construct my own location in 3D.

`
Though it was a lot more work then my usual compositing (which is work in and of itself). This would take me a week. However the control this gave me over lighting every element contributed to the awesomeness of the end product.

Building the desert itself ended up being rather straight forward. I went for a mesa valley, containing an increasingly rare lake.
`
Where the terrian was trickiest (other then scaling the landscape to match my Gorgons... but that's in a moment) was lighting it. If you compare this early test render to the last, you'll see my conundrum. Sunsets generate a beautiful range of yellows, oranges, and reds. The question is which of these did I want my scene dominated by.
`

The other complication in this was how many different lights was I going to need?

In 3D scenes the lighting is completely up to the modeller, and there isn't just one type of light that fixes all.
`

I'll probably at some point do a full tutorial on 3D lighting (as I've been learning a lot about it this year), but here is roughly how I lite the scene and solved the color problem.

`

In the end the Karoo ended up with 5 lights. The first one is obvious, and one in the real world you'd expect. There is a great feature in Carrara that not only generates the 3D sun you see in the scene, but allows you to make it a light source that mimics properties of real sunlight (such as the colour, brightness, and even clouds... all of these changing on how you position the sun in the sky... so in this case it helped set a baseline sunset lighting).

`

However if you think about this position of the sun to my subjects, they'd all be backlite and thus nothing more then silouettes. The other issue I had was with shadow casting.

`

To solve the Shadow issue (and a bit of my backlighting issue) I created a general "bulb" light directly above the sun (just above the top frame in the final render) that cast a bright yellow light. Bulb lights work just like a uncovered light bulb, they cast light in every direction.

This first bulb was to lighten up the whole scene, and thus help increase the contrast of the shadows. Due to its better angle I made this my central shadow casting light so all the shadows in this scene aren't strictly scientific from a physics point of view :P.

`

This however made the scene too bright. To darken it back down , and yet better light the objects (if that makes sense) from viewers point of view I created a dark red bulb that I placed directly behind the camera. In essence where the picture was taken from.

`

This lit the scene perfectly, but the Gorgons were still very dark. To fix this I created a lighter red light that was set to only light the creatures (and their fur! Which I missed in first couple passes at the scene). This is a fun benefit of 3D light, unlike real light, you can make it effect only things of your choosing!

`

My last light was for a subtle, but needed effect, and that was the glisten of the uneaten meat on the carcass. This was a nearly white "spot" light, as the pure white mixed with my other scene lights nicely. Spotlights can be thought of as like either a real spot light, or metaphorically like a flashlight. It shot a beam straight ahead of it whereever you point it, but that's it. Only things in that line get lite. In this case just the meat objects on the skeleton.

`

All the individual elements were ready. It was finally time to put them together.

When I hit yet another fur related problem! Which you can see here.

Fur and other surface replicators have another annoying catch. Once you've created them if you change the size of the recipiant object (in this case I had to make the Gorgon smaller, so I shrunk its body parts) the fur doesn't change size to match!

`

I'd modelled the desert on the scale of 200 units long by 200 units wide. My Gorgon was 40 units long. Meaning it was quite large in what should have been a large desertscape. Attempting to shrink my Gorgon caused this fun fur effect...

`

At least my nice lighting obscurred how dumb it is!

`

After rescaling the landscape to an impressive 3000 by 3000 units (anymore then that and my lighting won't work... though this week I have a new means to light even a million by million object!) I had to readjust the lighting a bit. Due to the shift in the sun (to get it back over the now taller horizon) the colour shifted a bit, and I couldn't quite get it back. Oh well.

`

To increase the feel of realism I decided to make a swarm of flies and shrubs scattered aross the scene. These were accomplished like fur (the flies a slight varation, but not worth going into so close to the end).

The last thing I had to do was import Gorgons (sadly one by one) position, pose, and light them. This took a long time just due to the computational requirments of so many 3D objects.

`

It all came to together as this. Hopefully as you've seen with the errors I documented (and the dozens of others I neglected to document) that though this piece is really awesome, I didn't directly set out knowing I'd get it this way.

`

However this first immersive lesson in lighting has me looking at this factor a lot lately, and I will probably be able to set out for similar effects in the future.

`

What I'm most proud of is that the multi layered lighting has accidentally mimicked the look of a painting. If you can't enlargen this version here is an enlargened version for a closer up look. This is the closest I've ever come to emulating my childhood hero Charles Knight... Now if only I could actually be as good as him :P