My sister asked me to design the logo for her website. Her vision: a woman in warrior pose holding a lotus in her upturned palm. My immediate thought: why not use Blender? After all, I had some success playing with existing logos earlier.

An enduring logo has to work in print, be quickly recognizable, and turn out well in just two colors. That means anything done in 3D must translate well to 2D. But because Blender is such a great tool with so many useful plug-ins, it acutally turned out to be an ideal starting point for this rather complex design.
The combination of MakeHuman and auto-rigging gave me the ability to fine-tune a female humanoid form to my (and Lisa’s) liking, then quickly rig it with an armature and pose it into exactly the right position. There’s a great white-paper on makehuman and autorigging here. Finally, rendering the female as shadeless white on a white background with the render trace setting gives the figure you see in the video above.

Rigging

Posed

For the lotus, I simply modeled two types of petals and used SpinDup to create the inner and outer rings of petals. I found that seeing the back layer of petals made the object a bit too complex so I simply sliced it in half with a plane before rendering. Then I played with the angle and positioning to get a shot that looked good when rendered as shadeless white with a trace outline.

Bisected Lotus

Finally, I took both stills into Freehand MX and used the autotool to convert them to vector curves. I added the bottom leaves of the lotus manually and continued to tweak and simplify the bezier curves to taste. The end result is an object I can fill with any color or gradient, scale up for print runs without pixilation, and composite against still images to create the website banner. There’s no doubt that the flexibility of a 3D object and the ability to reposition both poses and camera angles at will came in extremely handy in designing this 2D logo.

I started by modelling a cat in Plasticine (click to enlarge):

So I could get a “feel” for the different geometric shapes that compose the cat. I used cylinders for the different parts of the legs, a cone for the head, a large cylinder for the body, and semi-spheres for feet. This helped me understand how the cat is put together, so that I could compose my own cat out of geometric primitives in Blender.

I started by assembling half a cat using the same primitives I used in clay (click to enlarge):

And then mirrored these components across to create a full cat (click to enlarge):

I then duplicated these same primitives into some new layers, and started playing with static particle effects to create a layer of fur. Here are the settings I used for the body:

And the preview of all the different primitives, each with their own fur settings for direction, length, etc. to attempt to simulate how the different regions of a cat’s fur behave:

Next I created an armature. I studied feline skeletal anatomy to figure out the rough proportions and which hinges bend (and how). I also used this guide in naming the bones of my armature (tibia/fibula, femur, metatarsals, etc.). By parenting the bones correctly to each other and using the IK Solver for the feet, I was able to create an armature that can be posed in ways that are analogous to how a cat really moves. (click image below to enlarge)

I assembled this armature inside the primitives, then parented the various primitives to their corresponding bones. I did this for both the skin primitives layer and the fur primitives layer (since, after all, they need to move together). This approach seems much easier to me than trying to creat a single, uniform mesh and then select all the correct vertices to parent to the corresponding bones.

Having assigned everything correctly and played with various poses to ensure the armature was behaving in a reasonable way, the next step was to turn the disparate mesh primitives into a single, unified skin. This way, rather than looking like a cat composed of geometric shapes, or a cat wearing plate mail armor, the vertices interact and pull on eachother just like a real unified convering of skin. Mostly this was a process of trial and error, using the knife tool to cut away and remove overlapping regions, then “stitch” the vertices of two meshes together using the vertex merge operation. This meant lining up the right number of vertices between objects, positioning them in relation to eacht other, and then finally joining the two meshes into one so they could be stitched. This was probably the longest part of the process, and as you can see from the movie above I still have some work to do on the front shoulders to get them to behave more like a real cat’s shoulders. This will be especially important for the “stalking” action, where the cat’s soulders often protrude while walking due to their trying to keep their torso low to the ground.

Finally, I created walk cycle actions in the action editor. I created a “walk” (left two; right two) and a trot (front left, back right; back left, front right) action, taking care to use a uniform distance between each step (2 BU). Then I wanted to simulate–as I have seen cats often do in real life–a transition between the two. I pasted the last pose of the walk cycle at the beginning of my new “walk-trot” action, and the first pose of the trot cycle at the end of this new action. The difference was only in the position of the front feet–they were flipped. So, I added in the same poses for the back feet as a traditional “walk”, but for the front feet I knew I would have to accomplish an odd number of steps in the same time span. I used 60 frames as my standard for each of the three actions, so while each step happened in 15 frames in “walk” and “trot”, and could happen still for the back legs of “walk-trot”, the front legs had to take a step every 10 frames instead. The trick, of course, is that I wanted the transition to have the same constant speed as the previous two actions (2BU per step). This meant if the cat was covering 2BU in 30 frames on the back legs, the front legs would have to cover 1.333 (repeating) BU in 20 frames in order to keep up. And that’s precisely what I did. Then I put them together in a sequence (click to enlarge):

Unfortunately, the “follow path” action property seems to only allow one action, and does not allow switching between different actions. So, instead, I had to perform the distance translation for the cat’s armature (with everything parented to follow along, of course) manually. However, since the rate of the three actions was constant, it was an easy equation–4BU per 60, or 12 BU for the whole movie. The result is what you see, above. Unfortunately, the uniform ground plane doesn’t show off the footfalls, which due to the simple calculations above do not slip around. However, the camera angle also doesn’t show the well-modeled back legs, either. Mostly, what you see is that I have plenty more to do, including vertex painting for the face, modeling and installing eyes, nose, and mouth, continuing to refine the fur (possibly using the fiber generator to supplement what static particles can do), and adding lots of other actions I can composite together in the NLA editor (like “twitch tail”, “wiggle ears”, and “sniff around”) to assemble very complex behaviors just like a real cat. I’m looking forward to learning much more as I continue to improve this model, and hope to ultimately use it in a short movie.

Thanks to Legaz’s great Blender/QTVR Tutorial, I was able to make my first QTVR movie using Blender. I wanted to create a useful reference for future QTVR movies in Blender, so I created a scene with a camera inside a cube and reference numbers in front of each wall. The camera rotates from frames 1-6 to face walls 1-6, thereby making it easy to render the necessary images for MakeCubic to create the VR movie. You can download the blender library file and reference movie here or here. Then simply import the Camera object from the .blend file (Objects->Camera) into your landscape or detailed interior scene, render frames 1-6, and feed them to MakeCubic. Works great!

Growing up, small electric motors held tremendous promise. Add a few plastic gears, axles, rubberband timing belts, and you could rig up your own invention. Out of reverence for this small token of possibility, I decided to model and render a somewhat glamorized shiny electric motor. I already have plans for building virtual gizmos with it which may or may not involve Stafano Saleri’s BMG. I used static particles for the copper wire inside the insulation as well as copper-and-green colorbands and colored clouds textures applied to the specularity color of the casing metal to give my best impersonation of what these little motors really look like. Enjoy!

I have recently become interested in atmospheric effects, perhaps partly due to the recent inclement weather. This animation demonstrates the use of rain effects similar to those in feeblemind’s tutorial, volumetric fog, rain dripping from rooftops, using the water plugin to simulate rain on an uneven surface (tile), rain in puddles versus rain in grass (generated using RipSting’s fiber generator), and stopping rain at a boundary to create a dry interior. All in all I am pleased with the “atmosphere” and look forward to detailing the inside of the building, adding trees and other landscape, and eventually using this scene as part of an animated story.

First, I modeled a pagoda using traditional mesh techniques. I wanted to create a kind of “lush anime” effect, rather than photographic realism, so I kept the texturing to a minimum and focused mostly on solid forms. I put each level of the pagoda at a different layer to make it easier to detail the insides later:

Next, I modelled the roof tiles individually by duplcating, joining, skewing, and tapering a simple teardrop-shaped plane tile. Then I extruded the tiles downward ever so slightly so that when you approach them at the level they appear to have some slight volume (versus a paper-thin plane). Once finished applying a little texture and some lighting, I was reasonably happy with the building. For terrain, I used fractal subdivsion on a plane, pulled it around a bit, and applied a fractal pattern with graduated colorbands between brown and green to give the appearance of uneven patches of grass and earth. Finally, I added a scattering of mesh fibers to simulate delicate spring grass. Overall, I am quite pleased with the effect of the pagoda on a summer day:

(click to enlarge)

To create the illusion of ambient occlusion, I turned on the mist effect and tuned it so that it obscured the “hills” in the background, giving greater depth to the horizon. Then I created concentric spheres and applied the cloud texture to modify the alpha channel, effectively simulating some realistic fog. I did the same for a hemisphere that encompasses the entire scene, which when set against a grey-to-black gradient backdrop created a storm cloud effect.

Then, as I said, I followed a process similar to feeblemind’s tutorial to create a plane that emits raindrops. I made them shadeless; essentially solid white drops composed from five vertices and some smoothing that fall down in a random pattern. Then I rigged squares around the edges of each layer of the rooftop, and played with the emission properties a bit to get a nice dripping effect. Finally, I added a translucent plane of murky “water” that intersects the ground plane at its lowest points (just like real water, which pools at the lowerst point) to create puddles. I played a lot with the water effects plugin on this plane and also a translucent copy of the tiles which I superimposed just above (along the Z-axis) the real tiles. This way, I can literally turn on or off the rain effects by hiding objects in certain layers. Finally, I played with the gamma and saturation settings on the render to get an effect I liked:

Overall, I’m happy with the end effect and learned a lot about how to create the illusion of atmosphere in a scene using lighting, animated plugins, color correction, and other tricks of the trade.

]]> http://www.robertpeake.com/archives/141-DavidAllen-Logo.html/feed 2 http://www.robertpeake.com/archives/142-First-Blender-Animation-Still-Life.html http://www.robertpeake.com/archives/142-First-Blender-Animation-Still-Life.html#comments Sun, 27 Feb 2005 09:00:00 +0000 Robert Peake http://www.robertpeake.com/?p=142

Click here for Quicktime movie

My first ever attempt at a 3D animation using Blender came out pretty well. I decided I wanted to do a simple still life, and wanted to explore the power of Blender’s recent integration with Yafray. I had tried my hand at a raytracing program for Mac at least a decade ago, and had always been fascinated by this particular juncture between art and technology.

Details of my adventure follow:

My first attempt at mesh modelling an apple, before I discovered Catmull-Clark subsurfs and smoothing, looked pretty bad:

But thanks to innumerable quality tutorials and the Blender 2.3 Guide, I soon got the hang of basic mesh modelling and lighting:

From here, I started exploring the possibilities of lighting effects, looking at nighttime lighting:

And daylight:

Finally settling on a kind of dawn/morning effect:

Next I modelled a lamp:

And played with lamp lights and spotlights as well as material translucency to acheive a more believable effect:

Finally, I added a variety of image maps to the various models in my scene to give them the illusion of realistic texture and color, including this page in the book:

I rendered the animation in three segments totalling about 18 hours of render time on my Powerbook G4 with a 1.25GHz PowerPC processor. Only later did I discover I probably could have acheived a similar effect using the very fast internal Blender rendering engine using environment maps for the mirror effect. Still, the end result is a decent first attempt at a simple fly-through animation.

Technical note: I rendered the second and third phases long after the first phase, and somehow didn’t get the lights just right. So, rather than re-render the first 300 frames I decided to render the last 30 of those frames in that set, then use a “dissolve” effect in iMovie to seamlessly fade into the second half of the animation. Most people probably wouldn’t have noticed the lighting change at frame 301, as it’s fairly subtle, but I wanted to try my hand at a little post-processing, so I added fade-in and fade-out as well.