Come Clarity

Since The End of Things is officially complete, please see the blog for my next project: Come Clarity.

Can't have a light, Without Dark to Stick it In.

The film is complete as of October 9th. Until I am ready to post it, I will update this blog a few more times with information regarding the techniques that I used to produce it.

Today I will discuss the creation of the tower’s light beam – the most central effect in The End of Things.

Blocking

There are several steps that I must take before I create the light beam in the scene. The first step is to block it out with a transparent poly-cone.

Blocking the light beam.

This particular scene is built to scale. The camera must truck in slowly and I do not want the tower to be moving at all. Thus the beam has to start far far away. This will cause some problems later when creating the shadows (though I did not know it at the time).

Note the gaps between the ocean waves; it is unnecessary to fill the entire scene. If placed in correct relation to the camera, only four waves are needed to create the illusion of an entire ocean.

Light Setup

The shot’s lighting setup.

In this step I hide the poly-cone and begin to light the scene. I start with the strongest light source (1) – the light beam which emits from to the top of the tower, where its eye should be. Then I place a wide-angled spot light (2) high above the scene to simulate the light coming from the sky or moon. I use a spot light instead of a direction light, because I like to have the illumination taper off towards the edge of the frame.

The third light (3) is an additional spot light that shines on the wave around the boat. This forms the circle that the light beam casts on the ocean’s surface.

Right above Hatchiman, is another spot light (4). This one generates a shadow for the character and the gondola and casts it on the water. Realistically, the shadow would be cast from light #1. However, I thought that the image would be more interesting if the shadow was cast towards the camera.

The two final lights are a directional light that is used as a rim light for the tower (5) and a point light (6) that is hidden behind one of the waves. This one subtly lights the tower from the bottom, creating an interesting gradient that I could later manipulate during compositing.

Tower’s Eye

The tower’s eye is a simple point light with default settings (color white, intensity 1.0). I added a light glow and modified the settings (see image below). The eye is then rendered on a separate pass with a black background.

The tower’s eye and its settings.

The Beam

The light beam’s main shape is a spot light with a fog added to it. Since the light source is quite far away, I’ve extended the decay regions all the way past Hatchiman. I didn’t however, turn decay on. I prefer to extend the decay instead of scaling the light, since scaling seems to cause errors when I map a texture to the light.

The light beam’s main shape.

For this particular light, I mapped a ramp to the color so that the edges taper off in a smoother fashion. The intensity is controlled by an intensity curve, simulating a slight decay.

Like the tower’s eye, this light is rendered on a separate pass with a black background.

Note: With the exception of particles, the entire film is rendered using Maya Software Render.

Fog

Originally, I tried to create all of the light effects on the same light. This caused a lot of problems and produced some undesired artifacts when rendering the animation. Thus I decided to split the light beam into several lights, render each in its own pass and composite them all later in After Effects.

Fog Light and it’s texture settings.

The fog light is identical to the light beam in its settings, except that the fog density is mapped with a volume noise.

Dust

The dust from the light beam is actually a volume noise texture, mapped into the density of the fog. Originally, I tried to use a granite texture (like in this tutorial), and although it looked nice in a still image, it did not animate well.

Dust Light and it’s texture settings.

Using volume noise with small amplitude and an animated time attribute produced a very nice result. It looks more like glitter, which is it what it actually is, but I still refer to it as dust.

Shadows

The light that emits the beautiful shadows in the fog is placed right in front of Hatchiman, and not in the distance where light beam begins. This is because the distant light seems to conjure up flickering artifacts when the shot animates.

The shadows!

In After Effects, these shadows are masked using the original beam shape and thus they appear to be coming from it.

Depth

I projected a black to white ramp on a cone, and scaled and positioned it to cover the beam. Later in After Effects, I used this cone in order to give the light beam a depth of field effect.

A simple cone with a ramp mapped to it.

Animation

Below is the animated light beam in after effects. I’ve placed each pass on top of each other using the Add Transfer Mode.

Click to play QuickTime.

As you can see there is still some flickering in the shadows projecting from the bow of the boat. I tried to raise the resolution on the shadow map and increase some of the quality settings, but was unable to find a satisfactory fix. Thus I ended up lowering their opacity in hopes that they will not draw too much attention from the viewer.

Compositing

Click to play QuickTime.

The colors of the beam are tinted to a golden color, in order to compliment the dark blues and purples of the ocean. This was done using After Effects’ CC Toner effect, which can be found under the color correction effect menu. I’ve also added Compound Blur and a CC Pulse Ripple to create a subtle intensity shift in the light, making it feel more organic. Finally, I tied it all off with a Fast Blur set to 1.0.

Below, you can see the final composited scene:

Click to play QuickTime.

A Little Contact

All of the 3D seems to be rendered (tfu tfu), and I am currently compositing the film. Thanks to the feedback of my fellow classmates, I have decided to recomposite all of the shots, and particularly to fix the colors so that the elements read better while enhancing the desired atmosphere.

Below is a contact sheet with different color experiments. I picked #4, but I'm open to suggestions.

Contact Sheet

Foaming at the Mouth

Summer is almost over and my last year of graduate school is about to begin.

I have spent most of the summer recreating the materials that I have lost during the crash; this did not take as long as I estimated. However, I did run into several technical problems, mostly concerning the creation of the foam between the moving gondola and the ocean. I will delve into the details of this problem and the journey to its solution in the following post.

nParticles

As stated before in a previous post, foam development was abandoned due to time constrains and the inability to utilize Maya’s nParticles on the hardware that was available to me. Thanks to the extra time that I was allotted over the summer, I had more hours to sit and research different techniques in order to produce this element. Although, a minor aspect, that most viewers won’t even notice, the foam plays an essential part in giving the illusion that the gondola is interacting with the water.

I originally attempted to use particle sprites instead of particle clouds, but couldn’t figure out a way to control the opacity and generation of the sprites. It seemed that I would have to learn scripting in order to do so, which was something that I wanted to avoid.

Fluids

I spent several days studying Maya’s impressive fluids system; I was able to come up with a satisfactory foam effect.

Foam created with Maya Fluids, click to play QuickTime.

The problem was that once the boat moved and the waves were running, the foam seemed to be “swallowed” by the water.

Foam being swallowed by the water, click to play QuickTime.

Constraining the fluid box to the boat, or trying to animate its translation and rotation separately was too difficult, as I had to adjust almost every frame and then check it in the render view. This took up about two weeks of my time and it still did not look satisfactory.

A positive result of this experiment is that I now knew the basics of Maya’s fluid system, and also, I was able to create usable foam for when the breaking waves hit the ocean.

Foam test for the breaking wave.
(See the last clip in this post for a final composite between the foam and the waves).
Click to play QuickTime.

Water splash, created by live action ink.
(Similar effects will be used to create the underwater and abstract sequences).
Click to play QuickTime.

While it worked for the breaking waves, it still did not work properly for the boat. I needed to find a way to create the foam, so that it would be automated and I wouldn’t have to spend too much time adjusting it in each scene.

Textures

The next attempt was to try and build a polygon mesh and apply an animated texture to it.
I created two sets of animated textures, using Particle Illusion, one for the bow of the boat and one for the wake.

Foam Texture, click to play QuickTime.

Wake Texture, click to play QuickTime.

This technique was faster to render than the previous fluid method, but still resulted in a similar problem of being “swallowed” by the ocean. Even though the render was faster, it still proved to be a cumbersome process that would take up too much time to adjust in each of the shots. More so, the animated texture files took up a lot of gigabytes.

Polygons with textured foam. The wake is projected on the Ocean Surface. Click to play QuickTime.

At this point I was ready to give up (not the first time). I left the project for a few days to work on some other things, and soon a new idea came to mind...

...Came Full Circle

I would generate an intersecting toon line between the gondola and the ocean, convert it into a mesh and emit the foam particles from that.

Toon line generated between the gondola and the ocean, click to play QuickTime.

This is what I tried originally with the cloud particles, which proved to be too resource heavy. I would have to bite the bullet and learn some scripting, in order to control the sprite particles, which were faster to render.

Fortunately, I came across Mike Rhone’s Explosion Rig Tutorial, which proved be very valuable as well as educational. Whatever the tutorial didn’t clear up, Mr. Rhone was very happy to help with via email.

Click to play QuickTime.

After reviewing the tutorial several times, and making sure that I understood the coding (learning through practice is easiest for me). I adjusted the parameters of the particle rig and attached them to the toon line that intersected between the gondola and the ocean.

Click to enlarge image.

As you can see in the test below, the foam is automatically generated and killed between the gondola and the ocean. This allows me to import this rig into any scene without having to do any major adjustments each time.

Notice how the foam automatically stays around the boat, click to play QuickTime.

The foam is ready to go, and I am able to focus on the more creative aspects of the project once again. Below, is a clip of the first fully composited shot. I’ve included the wake texture as a projection underneath the boat to complete the effect.

Below is the completed shot, composited and in Adobe After Effects:

Completed first shot with effects. Please forgive the poor compression and colors.
Click to play QuickTime.

A Note About Rendering Sprites

I would like to note that Maya can only render sprites through its Hardware Render. I have encountered a couple of problems using this render, and I would like to share the solutions that I have found in order to overcome them.

Black Frames with Hardware Render
If you are receiving a black or blank frame with the hardware render, try to change the camera clipping. I found that reducing it usually solves the problem.

Slow Rendering with Hardware Render
Generally, Hardware Render is very fast. I have encountered, however, that sometimes it’s even slower than software. At first this seemed very random, as a low resolution scene would take three hours to output, while the same scene at hi-res took a few minutes.

What I discovered, is that if you cache the particles, Hardware Render will work very fast. If they are un-cached, that is most likely the cause of the slow render.
Here is a quick tutorial about particle caching.

Semester Conclusion

It is the end of the semester, and I must type up a conclusion for the directed study courses that involved this project. One dealt with the style and look of the picture, while the other dealt with Stereoscopy.

Looking at this blog, I think it premature to arrive at a conclusion, as a lot still seems to be missing.

This is not exactly the case, however.

Two days ago I was very close to finishing this project, the files were all set to render over night. When I arrived at the lab the next morning, there were errors on every single computer that I set to render. This is normal and I was ready for such an event to happen. What I was not ready for, was a corruption of the files. It appears that the combination of Render Layers and References can cause serious problems to the Maya files (http://forums.cgsociety.org/showthread.php?t=316471). Thus most of my files have been corrupted and been rendered useless.

At first this did not seem like a major setback. I had older versions on a portable hard drive and it would take me about a day to setup the files again in a proper fashion without using render layers. As I turned to get my hard drive, its cable was caught in the wheel of my chair and dropped to the floor. This of course was its demise. I have backups of all the files on another computer, but as of now they are at least two weeks old. Thus I am unable at the moment to conclude this research with all of latest materials.

I will go ahead and write about some of the intentions and discoveries that I have made all the same, and how I think this will reflect in the continuation of this project (I am determined to finish it very soon) and my following project – my thesis.

On Learning Maya

At the beginning of the semester I had no knowledge of the Maya software package. I have worked with 3D Studio Max before, but opted to switch as many people around me were familiar with Maya, and so could help me if I ran into difficulties.
In the past six months I have learned to model, unwrap, texture, light use nParticles, Toon and PaintFX in Maya as well as a little MEL scripting.

I have also discovered that the working process in Maya is %20 creation and %80 percent problem solving. While I do not mind the challenge of an occasional problem, this ratio can get tedious under a tight deadline.

When working analog, every change one wants to make should be considered harshly, as it will usually cost a massive amount of time to execute. But in Maya, things are so easy to change and tweak that I tend to lose track of the big picture and misplace myself in detail.

This is something that I have to watch out for should I continue using this software. In my further projects, I will cut a new animatic once a week, so that I can always keep an eye on the entire process.

Working Pipeline

Since my entire rendering and compositing pipeline was based on Render Layers (which could be an incredible tool if properly implemented), and since they seem to be the cause of the file corruption. I have to find another way of setting up the project. The old school way of setting up a different file for each pass will probably do the trick. I will also import all references and cache all of the animation before render is to begin.

Perhaps this particular film was not the best for testing out a style. The Ocean itself was such an undertaking – creating the moving waves, textures, foam and splashes -- that it took up most of my time. I had less time then, to work on the character and how it will blend into the environment. In that sense, the crash could have been a blessing, as I can now take a few more weeks to focus on this aspect of the film.

Over the next few weeks I plan on rebuilding the corrupt scenes in a more organized fashion and hopefully come to a more positive conclusion before thesis work begins at the end of June.

Stereoscopy

The Stereo cameras in Maya 2009 are another one of the software’s blessings. They are easy to work with and allow the artist to focus on the shot more than on the technicality of achieving the stereo effect.

When I first began to learn about Stereoscopy at the start of the academic year, I was convinced that every film should be in stereo. I am less convinced of it now, as it adds another layer of work to an already challenging process. However, I do think that that extra layer can be used on some films, particularly animation to add another dimension or simulate a point of view which would be harder on a flat image.

Setting up Stereo cameras in Maya is quite easy, and once I have my files rebuilt, I will cut a new stereo-animatic for this film. I believe that the depth of the media can add to the feeling of separation that the character has with its environment.

Animatic - May 2nd, 2009

With all the hard work that went into rigging the waves, animating them still proved to be an "ant-work" process. Thus I decided to cut out the first shot of the film because of its length and complexity, and substitute it with several shorter shots.
This way, I could avoid animating the waves for a minute and a half straight.

During this time I added a couple of more shots to the film. Being a few weeks from the final deadline this was a risk, but it ended up making my job easier and making the film stronger.

Click to Play Quicktime

Smudged Edges

Several posts ago, I wrote about the smudged edges that I wanted to have on top of the geometry. I was partially successful in applying this theory to practice.

These smudges were created by assigning a toon outline to a surface shaded object and then attaching a paintFX brush to the toonline. The brush is then set to SMEAR or BLUR, depending on the desired effect.

It all works well until things are set in motion:

Click to Play Quicktime

As you can see, the toon line and paint brush causes everything to jitter when the elements are in motion.

Click to Play Quicktime

With minimal movement, things look better. Unfortunately the character will be riding the waves and moving frantically up and down. Due to lack of time, this technique will have to be dropped and the smudging will be done in post.

Foam

Water foam has been a tricky issue in this project and in my opinion hasn’t been solved yet.
I did a few tests after studying Maya2009’s nParticles. They are quite amazing in their collision capabilities, but the ultimately they are too slow and impractical to work with, considering the machines that I have available to me.

Below is a test of my result:

Click to play Quicktime

If time permits, I will try to spend a few days with Maya fluids to see if I can discover a more efficient way to create foam. All of these elements will be composited together to look more natural of course.

Building the Water

About six years ago, I wanted to animate an ocean for my film – More Human than Human. I have never attempted such a task and my instructor –-Raz Oved- - advised me that I should change my idea. Never the less I insisted, and Raz suggested that I would try making it in 3D. His exact words, if I remember correctly were: “There will be a lot of blood and tears in this process, but perhaps an ocean will come out of that”.
Indeed he was right, and after that experience I vowed not to make a film with an ocean ever again. I broke that vow a year later with In the Silence, and I am a breaking it once again with The End of Things.

Hatchiman will forever be linked to the ocean

This concept image represents the geometric look that I wanted for the waves.There are two types of waves here, the breaking waves are the ones that curl, and the surface waves are the ones that act as the water surface.

Theoretically, I would create a polygon box with many subdivisions, stretch it and then create some sort of rig to control the wave’s movement.

The Breaking Wave Rig

I experimented with several different rigs, before I found one that was efficient to animate with and gave me the results that I desired. The first rig consisted of an IK spline that controlled the back end of the wave. I could pull its segments up and down with the help of the clusters. The front curl was controlled with forward IKs. By selecting all the purple joints in sequence, I could curl them by rotating the Y Axis.

Click to play Quicktime

Next I tried to simplify and built a simple line of joints. The idea was to quickly animate the wave and not rely on too much automation. It was a simple rig, but gave stiff animation results.

Click to play Quicktime

Here is a similar rig to the last, with a polygon box smooth-binded to it. Not quite there yet…

Click to play Quicktime

These three attempts brought me to the conclusion that a spline IK would be useful for controlling the back of the wave as it results in a smooth motion. Forward IKs would be useful to control the curl. There was still an issue of the geometry curling into itself though.

To solve this, I created two separate rigs -- one for the back and one for the front. The back was controlled with a spline IK, and the front was simple joints. I wrote an expression where each joint in the chain automatically rotates and scales down an increment in relation to the joint preceding it. I created two attributes out of the expressions: WAVE controls the curling, while CURL controls the scaling.

Once the two rigs were complete, I connected them to each other via another joint and then smooth binded additional joints on top of the spline IK. I attached locators to each of these joints for quicker control and I assigned the WAVE and CURL attribute to the front locator. Allowing fast access to the curling attributes.


Later, I quintupled the wave and parented it to a nurb curve that allows me to control all five at once. Below is the first successful test animation.

Click to play Quicktime


The Surface Wave Rig

After creating the breaking waves, the surface waves proved to be an easy task. I simply created an identical rig to the back rig of the breaking wave, only with more joints.
I placed numbers as the controls for each joint so that I would be able to track the high points and low points of the wave.

Below is a test of the surface waves and the breaking waves together:

Click to play Quicktime

I would like to thank my good friend Thomas Huang for all the time and knowledge that he has given me with rigging these waves. Without his help, I would probably still be rigging today.

Elements Complete

A long silence from me usually means that things are getting done. Below are images of the three main elements -- modeled, textured and rigged:

Hatchiman

All models (including this one) were created in 3D Studio Max 2009. I have been using this package for many years now and I find it very comfortable for modeling and unwrapping. From here on, the rest of the work will be completed in Maya 2009, as I am learning that it is more versatile than Max, in particularity with animation and effects.

Blendshapes for facial expressions. There are only eight expressions (four for each side of the face) as not many expressions are required for this piece.

A rig was completed in a matter of three days thanks to Anzovin Studio's The Setup Machine. I have added the floating joints on each side and the center in order to help control the coat.
Again, this rig is a simple one and it is designed to accomplish only the animation required in my specific shots.

Gondola

Please note that the textures on all of the models are quite basic at the moment. I am using a simple Surface shader with a painted texture. In the following weeks I will begin to work on the rendering style and I am prepared to build upon these shaders.

The Gondola has a separate controller for the boat and the oar. It is rigged so that the lantern will swing and maintain it's arc when the boat it tipped. This was accomplished with the help of Maya's constraint system.

The Tower

The tower has two forms. On your left it is closed proper, and on your right it opens it's massive jaw in order to suck in Hatchiman. I still need to add the beam of light that will be projected from it's eye.

Blendshapes are used to control each of the tower's teeth individually. The top slider of each tooth controls it's sharpness, while the bottom controls it's position. After constructing this system I have discovered the sharpening blendshape does not make much of a difference and I will probably not use it at all.
I have decided to use blendshapes in order to maintain the tower's "flowy" effect.

Quick Paint Test

This is a quick painting that I did over a render of the model, this is a step towards the rendering style that I am going for. I would like to smudge the model's edges into the background.

First Animatic

The purpose of this project is to act as a style development for my thesis while also being a self contained little film. Thus the visuals will be more important then the story itself. In this animatic, I have attempted to present the seven simple shots that make up this piece as well as laying out a temporary soundtrack.

The animatic comes in two forms, standard and stereoscopic. The latter requires red and blue 3d glasses to view. The final film will utilize polarized projection, which requires colorless 3d glasses similar to the ones used in Sony's BeoWulf. This type of projection allows the filmmaker to keep the quality of the color in the film, but is a little more difficult to work with.

Standard Animatic

Length 1:29, 7mbs

Stereoscopic Animatic (Anaglyph)

Length 1:29, 7mbs

Hatchiman

I have used the character of Hatchiman in several films and stories before. He first appeared to me in a dream, or rather it was myself, standing in a yellow field with green sky, wearing the stitched leather garb.

First drawing of Hatchiman, the morning after the dream (2003)

I am now attempting to use Hatchiman in this project, as I find that he tends to lead me through different paths of self discovery. Thus below, you will find a brief history of his development and growth:

Following the dream, I was intrigued with the character's look and began writing stories with the purpose of figuring what the character was about. I now believe that the stories and comics that I drew were too introspective and thus felt like cliches. It was not until 2003, when I began working on a short animated film called More Human than Human that the character began to take form. In this short , a starving Hatchiman washes off the edge of the earth and is rescued by a Windor (dragon-like creature), who carries him to her nest.

More Human than Human, 2003, 4 minutes

Sensing Hatchiman's hunger, the Windor takes off to find him food. Unfortunately, Hatchiman's impatience gets the best of him. He lights a small fire in the center of the nest, and begins to boil the Windor's children. When the mother returns, Hatchiman burns her and pushes her into the sea. The film ends with a back shot of Hatchiman taking off his mask, revealing long black locks of hair. Coupled with his waify physic, this brings up the question of the character's gender.

While I am not sure how successful this film turned out to be, it never the less established for me that the character is selfish, impatient, ruled by his desires, and also, perhaps gender-ambiguous.


Soon after, I was to begin working on my thesis project at Camera Obsucra School of Arts in Tel-Aviv.

I wanted to discover more about the character, and so I began to write his childhood story, which asked the some of following questions; where and how he was raised, the society and the world that he lived in, his ambitions, struggles and goals. All of these traits were to be portrayed in my fourth and final year film, entitled Shadows Fall. I developed this project intensely over the summer of 2003 with the help and support of my instructors, Tami Bernstein, Amit Shalev and Raz Oved.

The cast of Shadows Fall

By the beginning of the first semester of the last year, I had two small books, each with over one-hundred pages of text, drawings and storyboards that would be used as the building blocks for the film. I also had a 10 minute detailed animatic to accompany these books. My instructors were very excited about my progress and were eager to see what I would produce. Unfortunately, the project seemed like it would be to grand to finish properly over the course of nine months and so I opted to scrap it. I proposed a different project starring Hatchiman, call Shadow Bazaar, but it was met with some hostility. I am not sure if it was simply a backlash to all of the anticipation brought on by the pre-production of Shadows Fall, or if the idea of the new film was inherently lacking (I think it was).

Thus I began work on another story with a different character that somehow led me back to Hatchiman!? This was the story of In the Silence. My thesis film for Camera Obsucra.

In the Silence, 2004, 7 minutes

In this movie Hatchiman began to establish himself as a wanderer, and a character that is only concerned with his own survival. I realized that most of his stories began or ended in the ocean. Even in the current tale that I am trying to tell (The End of Things) he is sailing at sea.

Character designs for In the Silence (2004), including sculptures for both Hatchiman and the Spiker

I left Israel in 2004 and with that I also left Hatchiman behind. Yet recently I have began to wonder about his timeline....

...how do his stories link up together to create an entire lifetime? Has he grown and matured with me over the years?

While The End of Things is not a film that will explore the character's personality, I am finding that it is helping me to discover new things about him, and thus about myself. Hatchiman perhaps, is an alter-ego of mine, a self-critique and a tool to solve (and cope with) my own personal dilemmas.

Recent stylistic designs for Hatchiman (2008)