Hi I'm Hamish. This is a documentation of my wonderful adventures as I make my way from a flailing animation student, to a powerful and successful art ninja (I hope.)
I'll be posting my work from Animation College NZ fairly regularly (at least that's the plan) as well as my own personal work. So hold onto your socks or they may get blown off. If you like what you see, good for you. Hopefully there'll be more soon enough.

By the way here's some shameless self advertising.
Deviantart
Portfolio (basically the same stuff but whatevs)
Youtube

Wednesday 2 December 2015

Rationale

To start this project I had to come brain storm to come up with my animation idea.
 
My final idea was a Marionette puppet Spanish soap opera recreation of the ‘I am your father’ scene from Star Wars. I chose this idea for the same reason Pixar created Toy Story, because toys are much easier to animate and render than humans.

Once I had my idea down I had to go about creating concepts for my character and scene, as well as a storyboard. I designed my character with 3D in mind, considering basic shapes and how I would model him.
 
 
 
I also designed a second character made of the same pieces as the main character to save time. Then I made turnarounds of each character so that I could put them into Maya as modeling reference.
 
With the turnaround imported into the front and side views of Maya I was able to model over the image using x-ray view. I built the characters out of the basic shapes I mentioned earlier. 
    
I had two attempts at modeling. The first worked well enough but the mesh was messy and didn’t smooth very well. The second time I planned how I would build the mesh more carefully and it came out much cleaner. I also made sure to keep the character low-poly so that I could smooth later and keep Maya running faster. Low-poly is also much easier to UV map than high-poly.
  
I modeled the body, head,and hair as single pieces rather than halves that I could mirror because stitching halves together can be messy and the pieces were relatively simple shapes that could be modeled using just world symmetry on my modeling tools. I did however model all of the limbs on only one side and mirrored them. I found that mirroring objects inverts normals on the mac version of Maya which meant that I had to flip the normals back and delete their history. As this was my second attempt at modeling I had already discovered that not deleting history can lead to some horrific crashes and glitches so I made sure to keep my project clean.
 


I made a sheet of colour concepts for my character and then picked my favorite so that I could paint my textures.
 
I had to UV-unwrap my objects so that I could texture them. I used multiple techniques such as plannar mapping, cylindrical mapping, and automatic mapping. Each has it’s own use for different types of objects. I did a lot of stitching to make the maps easy to paint on.
Next I painted/photoshopped my textures. I made some out of heavily edited pre-existing textures and painted other parts by hand.
 

When I went to un-smooth my meshes to start rigging I discovered that un-smoothing ruins the textures if they were created on a smoothed model, so I had to go back to some previous versions to retrieve unsmoothed meshes and fix some textures for low-poly.
Once I fixed the textures I started placing joints to build the skeleton. I didn’t plan the structure very much to start with and my first attempt didn’t work out very well. I also had a lot of crashes due to some history on certain meshes that I had left.
   
Once I’d figured out how joints and constraints worked by playing around, following Keat’s tutorials, and watching some Digital Tutors tutorials, I started my second attempt. I had a plan in my head for the structure and it worked well. My original idea was to keep the puppets fairly rigid with just toy-like joints but I decided to make some of the pieces flexible by adding more joints inside them to give a better performance.
    
I was careful to place joints in the centre of where I wanted parts to pivot by using snapping and I ordered my hierarchy as I’d been instructed along with consistent naming conventions. I managed to keep everything very clean and had no issues this time.
    
I made all of my controls using curves and constrained and parented them in the hierarchy. I also used set driven keys for the mouth and for the fingers so that I had a simple hand control and so that the mouth had limited movement. I used IK on the legs and arms with pole vectors on the knees and elbows so that I could control the limbs more naturally.
 
Next I made a fairly simple test animation to make sure I had all of the control over the rig that I wanted. It was very easy to animate with so I was happy with the rig.
 
I then repeated all of the texturing and rigging steps for the Grandma character and found that it went a lot faster than with the first character.

Next it was time to set up my scene. I had a sudden idea of how to change the scene so I did some new concepts for the stage and a new storyboard. The new concept was about a 1 minute long video rather than 10 or so seconds.
 
I started to model the room and desk, and I also modeled a bunch of objects to litter around like a bottle, candle stand, and some books.
I also had to UV-map and texture these objects, I also tried bump mapping, transparency maps, and specular maps, and played around with all kinds of material types and properties.
 
 
I started doing rendering tests with different rendering engines and settings. I settled on the MentalRay renderer and built a physical sun and sky in the render settings. I really liked the look of this render even though the render times were immense.
 
Once the scene was more or less built I started laying out my shots according to the new storyboard. I made 7 different cameras for the 7 shots in the plan and placed my characters into the scene. I rendered out single frames of the shots and put them together in After effects to get my initial timing.

I began to block out the major movements and figured out where the words would be placed. I recorded myself saying the lines I had planned and put together some simple music in Garageband. Then I imported the audio file into my Maya timeline so that I could match up the movement with the sound.
I roughed out all of the lip synching and started to clean up the character and camera animations. I did various playblasts along the way, mostly of single shots. In order to save on rendering time I usually had a shot rendering while I was working on another shot so I didn’t have a playblast of all of the shots to watch together until the final renders were mostly finished.
 
I found that the renders were taking even longer than I expected even though I was rendering on two computers so I had to render some of the shots using the Maya software render engine to save time. I optimized my scene for this renderer by changing some materials and adding more lights in an attempt to more closely match the look of the MentalRay renders.
 
I rendered my shots a jpeg-sequences and brought them into after effects to edit them together. I could have used Premiere to edit but I had a lot of effects I wanted to include so I chose AE. I did some color correction, mainly on the Maya Software renders to make them match the other shots. I also added some atmospheric lighting using solids, masks, and various built in effects.
Then I did a final colour grade using Magic Bullet and added chromatic-aberration to give an old VHS type feel. I added simple titles, credits and subtitles to the video and then it was finished.
 

Gettin it Done

Kaaaaay. So I haven't posted for freaking aaaages again. At this point school is actually over for the year and my animation is finally finished. Yaaay.

But here's what I've done since the last post...

So I came up with a new idea of how to set up and animate the scene. I decided that it'd be cool to create a sort of old attic where the marionette theater was sitting on a desk with some other junk. So first I did a super quick concept of the new scene along with some visual research.
Then I also did a quick storyboard V2 where I added some nice shots to show off the room.
Then I started modeling the room.
I kept the room fairly simple. I just used planes for the walls and floor and I made a desk out of boxes. I imported a couch that I had made for the previous version of the stage.
 I modeled a new stage to look like a simple cardboard toy stage. I built it out of planes and circles that I merged together, then I edited the meshes and extruded into 3D shapes.
I added a roof and back walls to close off the room. This way light could bounce more accurately around the room. I also added a window which is the main light source for the scene and some curtains which were edited from the stage curtains.
 
Next I started modeling various things to place on the desk using the visual research as reference.
 
  
Then I decided to do rendering tests to see what rendering engine to use. To start with I used a previous version of my stage to do this. I switched around settings and eventually settled on the MentalRay renderer with a physical sun and sky. MentalRay takes much longer to render but it looks far better than the Maya software render.

Then I UV-mapped the models and made textures in photoshop. I also experimented with generated textures using noise and fabric nodes in Maya as well as Bump maps, transparency maps, and specular/reflection maps. I did rendering tests as well to see how my materials were working.
 
 
 Next I began to import these models into my scene and did more render tests.
I continued to play around with material settings inside of the scene. Instantly the physical sun and sky settings in MentalRay lit my scene fairly well. I adjusted the sun direction to come through the window at a better angle and added an ambient light to fill the scene a little better as well as an area light to bring out the details behind the stage.

At this point the scene was essentially finished so I moved on to animation.
First I made 7 different cameras for the 7 shots I had planned and I framed each of them according to the storyboard. I also placed my characters in the scene and rendered out a first playblast of the whole thing to start getting my timing/pacing right for the shots.
Once I was happy with that timing I started blocking out the character animation.
Then I did my voice recording. I wrote my lines and used Google translate to get them in Spanish. My older brother helped me out with the recording by hooking me up with some equipment and doing the recordings while I said the lines. I took the audio files and matched them up to the first playblast to make sure I was happy with the timing. I imported the playblast with the sound into Garageband and used some loops to make some simple background music and timed it all to the video. I then exported all of that audio as a single file to use for lip syncing in Maya.
To lip sync I just had to import the audio file to my timeline and then match the mouth movements to the sound. Since I was just using puppets the lip syncing was pretty easy, just moving the jaws up and down. With a more complex rig lip syncing would be much more difficult.
Once I had the lip syncing all done it was fairly easy to animate the characters in time with the talking. I also animated the cameras in time with the shots so that they moved at the right times. Using the graph editor I was able to clean up the movements of the characters and fine tune things a little more.
While I worked one one shot I always had another shot rendering in the background. I was rendering on two seperate computers to speed it up because the render times were outrageous.

After rendering the first shot I discovered a lot of noise in the render. After some playing around I found that I had 3 main issues. My shadow quality was too low under my lights settings, my anti-aliasing quality was too low, and my final gather quality in indirect lighting was too low. Raising these values fixed the noise issues but made my render times even higher. I decided to raise them until there was little enough noise in the shot that I could de-noise it later in after effects. That way my render times stayed a little lower.

From there I only had to edit the shots together and do some post work. I rendered the shots out as jpeg sequences and each time one finished I imported it to after effects and synced it with the audio. After having a lot of failed renders due to crashes, animation errors, computers being unplugged half way through, lighting errors, and many other human errors, I had more than run out of time and had to settle with rendering out some shots with the Maya Software renderer. I figured I should still try to match the lighting and look of the shots as closely as I could. So I added new lights, changed some material properties and tweaked some render settings and I managed something that looked halfway decent.

I brought the rest of my shots into After effects and got to work. First I had to colour-correct the Maya software shots to match the MentalRay shots. Then I de-noised some shots and checked that eveyrthing was nicely synced and the timing was good. I did some small effects like light rays coming through the window using solids tracked to the footage and light effects.
Then I used a plugin by Red Giant called magic bullet on an adjustment layer to colour grade the whole short. Inside of that plugin I added multiple coloration effects including chromatic-aberration to give an old VHS look to the video.
Once that was all done I just had to add my credits and I was done. I rendered it out and put the video through Adobe Media Encoder to compress the video to a reasonable file size.

And there it is.
All finished.
Crazy late but finished.
I need to work on time management.
And realizing how much I can achieve. I aim way too high.
But oh well. It's done. Enjoy.
~Sayonara Blogger-kun <3