posing my character

I made a plinth out of old save files of this project felt it was pretty apt to have all the failed attempts below my victorious functioning model. It was nice to finish the project with the how we started it modelling our own version of the demo rig.

Finishing off the demo character

I had make the final touches to the character, a few issues from previous mistakes cropped up during this process. When trying to connect the eyelids to the head the constraint would not work for some reason. I realised that there were some issues with the hierarchy and the head controller was not parented to the neck so the controllers movement was not quite right. Solving this allowed me to apply the parent constraint to the eyelids.

Due to impatience I had been scrubbing through some of the tutorial videos which meant I had actually missed some key pieces of information. One thing I missed was that the legs contain two different IK handles. It was apparent when I couldn't put a pole vector on the knee locator. It was a simple fix I just deleted the leg IK's and made sure they were put back in the correct region of the hierarchy however if I'd watched the video properly at the time I would have not wasted time now trying to figure out and solve what went wrong.

The main task was putting in the custom controllers however at this stage I had already done them on my worm model so it was fairly painless process doing so just quite laborious. I had to fix some of the weight paints as the foot controls brought up some issues surrounding it.

I was pleased with the overall model as it showed my learning process through maya and was an embodiment of that however it was far from perfect. There are a lot of things that could be improved with it most of which lie in the rigging process.

Limitation: Animation meets live action

Computer generated animation gave birth to a new generation of live action film films and unlocked the possibilities for writers and directors that were once closed to them. Animation has given artists and auteurs alike the ability to create imagery that could not exist in the real world and tell incredible stories through this limitless scope animation provides.

Live action film has used animation for a long time to help tell stories for example the great Ray Harryhausen created many stop motion special effects for block buster films. Unfortunately these kind of special effects quickly became dated with emerging camera technologies and film makers started looking to computers for the next generation of special effects.

In 1972 the westworld became the first feature length film to feature computer generated animation. This was the start of a revolution in film making. Most blockbuster films these days will contain some for of CG animation and many feature it heavily. The highest grossing film of 2015 was "jurassic world" , a film without CG animation could not be created. All of the dinosaurs where created using 3D animation. The high poly models and complex texturing make animated characters look like they belong in the real world. The technology is always evolving as computer become more powerful the limitations of 3D animations is a list growing ever smaller.

A problem this has created is film makers have taken CGI as a tool and seriously overused it. Its much easier and often cheaper to do something in CG that it would be as a prop.  The over glossy approaching of green screen everything such as the films of Michael bay lack any kind of substance and almost feel like an excuse for a technical excise. Actors often work better with physical sets rather than working with green screening and I've found films that try to have as many physical sets as possible feel a lot more genuine. A great example of this is interstellar which used used CGI a lot more conservatively than it could of. Nolan pushed for as many physical sets as possible for actors to interact with and the result was one of the best sci-fi films of a generation.

Sound editing

James volunteered to do a large portion of the sound since I did a bigger bulk of the post production. We went over the animation before hand so we could discuss what sound we thought should go into the final composition. He sourced a large amount of the sounds from sound libraries/internet, however some of the sounds we needed to make our selves. Most of the sounds we needed to make were character sounds of moans and shrieks. we spent a few hours in the sound booth getting these sounds we both voiced our own characters and directed each other sounds. I felt I was a little more directive when it came to the recording I knew exactly what I wanted for the sounds and the emotions they should portray however I found it difficult to try and make/direct James to create these sounds.
James put the rest of these sounds into the sequence and sorted out the levels. Once he'd finished that I went over all of the sounds for a second edit I added a few background sounds, changed the levels and replaced sounds I thought did not work. I felt it was important that I was involved in the editing process of the sound even if it was right at the end because sound can completely changing the tone and meaning of the animation.
I felt we should have planned the sound at a much earlier stage in the production, it was a bit of an after thought and I think it was obvious when watching the final animation that it was. This is something to bare in mind for future productions as sound design is very important. I found trying to not take control over the sound design difficult or overly critique however it was important I didn't as its a joint creative process. This has been a learning curve for me to relinquish creative control of some aspects as I am not used to collaborating.

titles and credits

I didn't just want the credits on a black background or a still because I thought this was would look unpolished. I didn't want to invest lots of time into it but it needed to match the aesthetic to the rest of the production. I rendered single images with alphas of apples and of a tree trunk. In after effects I key-framed the apples and added some motion blur to make a short loop. I was very happy with the result and think it will add that extra bit of refinement to out animation. 

I was unsure what fonts to use for the titles and credit but I knew I wanted ones that were somewhat light hearted. A lot of film uses fonts that are quite conservative however animated shorts are normally select a little more creatively. I used a font for the opening titles that looked somewhat hand drawn because it was quite slender in design so didn't distract too much from the opening shot but still had an fun feel to it. For the credits I scrolled through a list of them in premiere, I knew I wanted something that went with the first set of titles but didn't know what. My graphic design/typography knowledge is fairly rudimentary so most of of the decision making was mainly done via taste. 

Limitation: Production stages->learning curves

With 2D animation once the animatic is made and the characters are designed people can start animating right away however with 3D animation there are a lot of preproduction processes to do before production can commence.
Before animators can start production characters must be modelled, rigged and textured before they are ready to go into production. If theres lots to be prepared there will have to be a long wait before animation can actually start. This can be very costly for a studio. There are many specific skills that a 3D animator could hone over a long amount of time which would force them to be stuck in a very specialised area. Whilst in other fields of animation you can focus more so on becoming a better animator (focusing just on the production) to be a 3D animator you need to have a wider range of technical skills to get to the animation stage. This could prove to be a problem for the independent animator who may be working alone there are many more tasks they'd have to be able to perform. These specialists may also have these skill sets in different softwares which can become problematic for animators seeking employment with studios who use different softwares. Specialising at a skill in a type of software may close opportunities to some jobs/studios that use another kind of software.

Limitation: computing power

3D animation requires a lot of processing power for different processes. The reason CG animation did not appear till the 70's and an animated feature not till the 90's (toy story) was there were no computers powerful enough before then to handle 3D animation. 

For feature films where the levels of detail of the animations is very high and there's a lot of footage that needs to be create the studios have to have what are known as render farms. Render farms are many many computers all hooked up made for the soul purpose to render animation footage. They will have incredibly good components to run them to keep render times down. 

The problem of using/needing very powerful computers is they come at a price, people who don't work for huge studios like pixar or dreamworks may not have these resources available to them. 2D animation can be done with analog components or relatively cheap software that can be run on relatively basic hardware. The "buy in" to have 3D capable equipment may turn potential animators and studios away from using 3D and turn to other types of animation.  

The avaiblity of technology has also dictated the quality of the animation animators have been able to produce. Although early 3D animation has some excellent examples there were severe limitations on what they could produce because of the computing power available to them. A good example is the polycounts available from john lasseter's first animation compared to hyperrealistic character from unchartered 4

Limitation: lighting

Lighting is a very important part of film and animation. It can set the mood of scene and change an environment from dull to spectacular very quickly. Good lighting can be achieved in 2D animation however in 3D Path tracing or Ray tracing light can create astonishingly realistic looking lighting on scenes. This realistic light can make animated landscapes a lot more immersive to the audience. It works by having paths of light which can bounce of objects. Objects depending on there material will reflect/absorb light differently which means the rays of light can bounce an illuminate parts of the scene.

http://nerdist.com/why-does-disneys-3d-animation-look-so-real/ This was an excellent source on how the lighting was done for big hero six. The films lighting looked spectacular and created some very realistic looking animation even on very stylised characters.

This kind of lighting that path tracing achieves is a level of realistic that 2D animation can not reach without an insane attention to detail in every shot where better results could probably be created in 3D much quicker. This would give animators more time to construct the environment and create the world they're trying to present to the audience.

Limitation: uncanny valley

The uncanny valley is a trait with realistic animation which is described through a graph. How  relatable a character is goes up proportionally to how realistic the character is until an event horizon which the character become too real an slightly off looking. This is the pit of the uncanny valley and creates some creepy monstrosities. This limitation has emerged with the development of 3D animation. Animators being able to create more and more realistic characters has enabled this unfortunate trait however improving technology should help animators dig there way out of the valley in the near future.

Shot from the polar express a film that has all 100 minutes at the bottom of this valley. The biggest problem faced by uncanny characters is dead eyes that seems to be something that realistic 3D characters can't solve.
 
This is a nice graph showing the uncanny valley with some examples.

Final editing

Once all the scenes were polished in after effects we rendered them out and put them into premiere. We lined it up with our original animatic however some of the scenes ran slightly over. Our animation ran 5 seconds over the minute limit without titles so we knew we had to cut some footage out. After a few watches of the footage there was plenty of fat to trim from it we cut it down to the wanted sizes relatively quickly and I feel the animation will be better for it.
When editing one thing I noticed which could be a problem (which is also something I often struggle with) was some of the shots may be too quick making the animation harder to follow for viewers. This is something I need to get feedback on the final product and can bare this in mind and try and learn from it in future projects.

After effects rendering

 knew that the normal H.264 compresses the video and can alter the quality of it which made me think having footage rendered twice with the same codec may alter its final quality. I researched this online and the general consensus on this was that I should render everything in lossless until the final export from premiere the lossless format is uncompressed and does not alter the quality at all. A draw back is the lossless files are rather large but its worth it to not have our colour palettes ruined .

post production fixes

There were quite a few post productions fixes to be done throughout the animation mainly was cleaning up issues to do with lighting, masking out errors or just adding something that should have been there. We split these fixes up so we both did some of the post production work. Here are some of the examples:

In one scene the tree above the spider was way too dark so had to be lightened. It couldn't be left because the previous shot is very similar but with the correct lighting so this one needed to be corrected. I took a still from the first frame of the previous shot and James gave me a copy of the shot with no spider. I had to carefully mask out the spider pretty much on a frame by frame basis which was very time consuming but I was not aware of any other ways to try fix it. I tried a couple of methods to fix it but this was the only successful one.

In this scene there is supposed to be a string of web. We thought the best solution would be to add it in post and this was definitely the right decision. It was a pretty simple fix I added is using a shape layer with a few points on it. This allowed me to key frame the movement from frame to frame. 

In this scene we'd storyboarded the eyes to close over the camera, again we thought it would be easiest to make this effect in post. I first tried using just mask expansions to create the effect but it was to jarring. I then changed the mask shapes and also the opacity of the comp with them closing which created a much nicer effect. I added multiple eye shapes to make it more obvious that it was the spiders eyes you were looking though. 

post production tests

Once I had the first pieces of footage rendered, I did some post production tests to create a "post production bible" for the project. This would allow us to apply the same fixes to any scene that needed it. The first thing I did with this scene is fix a glitch that came from the render where the shadow in the bottom corners flickered. I took a still from the image sequence and masked it so it would stop flickering.
Next I realised that the two character should be in shadow because there should be the tree top above which was removed so the camera could be there. I used a duplicated layer with a feathered mask in the region I wanted the shadow, I tried changing the levels however this did not create a nice effect. I found using the saturation effect to desaturate the region and reduced the lightness to created the best shadow.
The next step was colour correction I tried lots of different effects to try and get a nice result however the best were the auto level with most of the original image showing this helped cool of the palette. The colour off set was good to help make the hue slightly greener when under the tree. The levels effect was great for overall small tweaks when the others were not enough, especially using separate levels for RGB.
The final thing I needed to add was depth of field, I found using depth of field in maya difficult and I knew it increased render times. I use a similar method to the other effects I created a new masked layer to blur out the ground. Since I'd already tried many different blur effects in previous composite tests I used the same one which was the gaussian blur.


This is a breakdown of the of post effects added to this scene.

Playblast edit 2

After we'd fixed all the issues brought up from the first play blast edit we made another one this time I made sure to get feedback from matt and some classmates. This was incredibly beneficial and was some of the most useful feedback we'd gotten over the course of the entire project. There were a few issues that I wouldn't have noticed or had been looking at too long that they went over my head. 

The problems that were pointed out which I fixed were: Making the anticipation movement better on the worms jump, added more of a rebound to the spider after it falls on its web, some camera shots were off centre. The biggest issue was camera movement throughout the animation.
Although before this animation I thought I needed to use more in my animation a lot of the movements were unnecessary and made it hard to follow. It became clear that framing the shots was the most important part and that if the movement could help the story then it was worth putting in other wise better to leave a static camera.
I spent a bit of extra time fixing the animations matt suggested could do with tweaking and it was definitely worth while.

light linking

Since we were using directional lighting with shadows the tree canopy became an issue. Light couldn't reach the two character on the branch so the scene was nicely lit but the characters were incredibly dark. I tried adding things like point and spot lights but these didn't light the characters very nicely and effected the good lighting we had on the rest of the environment. I solution I was shown was something called light linking. This was an attribute menu which  controlled what lights had influences on. I duplicated the directional lighting and made one work on just the environment and the other work only on the characters. This meant the character light could pass through the tree and light the characters nicely whilst the background kept the same lighting as before. This was a great solution and the lighting created was really nice especially with scenes similar to the below shot. 

compositing tests

 I wanted to test some compositing in backgrounds before I moved onto the post production phase of the project. The main issue I had with this was a thin black outline round the transparent regions of the images. This was caused not by the render settings in maya but how after effects interpolated the footage. The footage must be set to premultiply when bringing in anything with an alpha this set the black outline to alpha. Made all the compositing a lot cleaner. I also experimented with comping trees into the background. When they are just put in its really jarring, to fix this I added a blur to the trees I tested a lot of different blur effects but I found the gaussian blur created the nicest effect. 

playblasting and big edits

Once I'd reached the stage in a sequence where I felt all the animation was complete I learnt that running a playblast of it was very important similar to that of checking your work with the render queue in after effects. I found just playing it back wasn't good enough and that it was easy to miss some mistakes in the motion where as playblasts show all the proper timing without dropping any frames.
Once James and I had animated all the sequences I got us both to make playblasts of all the animations. I then put them together in premiere according to the animatic. This would give us an idea of how they all ran together and the smooth playback of it all showed us a lot of errors that needed fixing.

This was an incredibly useful exercise the main thing it showed us that we would have not noticed otherwise was bad transitions between shots. 

Problems we discovered were some movement issues with the spider in scene 17 and the grass needed to be fixed as there was only a temporary fix at this point. The scenery boundaries were very obvious and needed to be extended, the apple needed to fall faster, the transition between 3rd and 1st person needed to be edited. There were some issues of camera transitions between a few scenes as well. We will both fix the errors and come back to the problem.

Learning to use another rig

This was more challenging than I thought it probably didn't help that the character had 8 legs to animate. Once I was accustomed to animating my character I then of course had to get used to animating my partners. The biggest thing that required practice was the leg IK's I found them quite hard to select and you need to move them in conjunction with the locators for the knees. I kept a lot of the motion a bit simpler for the spider such as jumping and lurching back. This saved a lot of effort trying to learn a walk cycle that wasn't really needed. I was happy with the animation that I've been doing with the spider I feel it works well in conjunction with the worm however plenty of practice is needed.

import problems

Once the background was complete we were both ready to start animating the characters in conjunction with each other however we ran into some problems. The background was made of many different objects and was therefore was impossible to scale, and both mine and James's character had issues when scaled. 

The easiest problem to fix was the background, I selected all the objects and combined them so they were one object. I made sure everything else was deleted so it would be a clean import. This created a perfectly scalable and rotatable background.

James sorted out the scaling on his character however there where some issues with it rotating 180 °  and my character could not scale however it could rotate with ease. A simple solution was reached where everything was scaled around my character and the orientation was set around James's spider. 

Starting to animate my worm

I wanted to start animating with what would be the hardest movement because I knew once I'd sorted that the rest would be a lot easier. The motion I started with was my worms walk cycle I knew once I'd figured out one cycle it could be duplicated. Looking at reference material I knew the secondary animation of the segments was really important when the worm was moving. I struggled for a while to get the timing of the secondary motion right but once I'd made it work it seemed a lot more obvious how to get that movement.

At this stage I didn't have importable scenery or the secondary character but I wanted to test character interaction and camera movement. I used a box to fill in for James's character and I was surprised at how much I could add to it by giving it some simple movements. When I do the actual animation I'll redo the worms movement to see if I can improve upon it and animate the spider along side in a similar way. 

First tests with cameras, lighting and rendering + final crit

After I'd finished texturing everything I moved onto a test render of the environment. This would be a test of a moving camera, lighting a scene and rendering it. I thought this would be a simple enough task however my very rudimentary understanding of all these things made sure there were plenty of roadblocks during this technical test. The first thing I tried working with was the camera. I was trying to created large camera sweeps that showed the entire environment and I tried to animate it the same way as a character and the results were quite jerky. A much better way that I should have done and will for the final scene is to put a camera on a curve which can use it as a track. The next issue I was having was with lighting. The problem was with using directional lighting and I could not get it to light the scene correctly. I now know the problem was with both render settings and the settings of the lighting but at the time I just opted for lighting the scene with spot lights. This was the first time I was batch rendering with this project so I did a bit of research into what would work best, a few sources said tif files were good to render with as they remain uncompressed so theres no loss of quality when working with them.

This is the final outcome of the tests, the colours are really off this is mainly due to the lighting set up and also the colour correction in the render settings in maya. I was happy with this as a test and it was done just in time for the final crit which meant I could get feedback on it. Generally the feedback was very positive on the work, the title was not well received which was fair it had been put together quickly and badly. 

texturing the background

James modelled the background and I was tasked with texturing it. We'd already done some pre production tests on this so a lot of the design side of this had already been planned out and been well received. I wanted an environment that was really pretty but also that was quite cartoon like. The first thing I worked on was the leafs for the tree tops. I used the same square brush method I had previously with a very similar palette which was set for the whole project. I was really happy with how it looked however I wanted to add another element so I decided to use the same effect but using a different shape. I made a triangular shaped brush in photoshop then used it in the same way and it created a nice texture that was a bit different to the others. 

I used a similar method with the tree trunks but I drew black lines over the top, which actually worked out a lot better when the texture was on the models. The only problem I found with this is it didn't match some of the uv maps 100% so the patterns go a bit funky however it sort of fits the general aesthetic. For textures such as the ground and the bushes plus other texture where the shapes where a bit trickier I used a one size fits all style solution to texture them. I used the same texture as above but I made sure the entire jpeg was covered so the object would definitely be completely covered.

 

Finished tree with both trunk and bush textures.

texturing my character

I went for the more pastely colours from my different palettes from the pre production phase as that is what feedback suggested was best in the crit. I found that it was good to paint outside the lines on the UV as when I didn't it created white gaps on the character.  I was ready to test other patterns on the worm but when I used this first one it came out exactly how I wanted it to in relation to the original designs. 

Once I knew the man colour pattern worked I proceeded to added the spots I had to try a few different sizes on the map as it was hard to get them to the correct size when the texture was actually on the model. 

I had some problems which must have cropped up during the mirroring process but there was a dodgy vert in the mouth that was messing up the uv map of the mouth causing some unpleasant effects. I figured deleting it was the best solution and it didn't seem to cause any negative side effects.

using camera's

Before this project I had not paid much attention to cameras in general. I knew framing was very important but never really thought much about camera movements. We had a brief seminar from matt about camera movement in film, why camera movement was important. The main thing I never realised was that in film there is almost always camera movement its just generally really subtle. Camera movements also help to set the tone of the scene such as a dolly zoom which gives a high suspense vertigo feel to a scene instantly. I have a lot of practice to do with this as I am very novice with camera work and my shot framing could be a lot better as well.

animating research

These are links to some of the videos I used for inspiration and for reference. I wanted the movements of my worm to be ridiculous (lots of secondary jiggling motion) but also move like an inch worm. 

The character Heimlich from bugs life was the first thing I thought about when thinking of the motion of the character. His motion is something everyone from my generation will remember from there childhood. His motion its incredibly jiggly and cumbersome which make the character look very silly. It looks like a lot of effort for this character to move and I wanted to get the same thing across with my worm character.

I found this animation really useful for reference on secondary motion but also how my character would make certain movements like standing up on its back segments. There was also a lot of good exaggeration in this animation which really added to the comedy which I found could be good if applied to my animation.

I also looked at real life inch worms since thats actually what my character is based from. From the early concept stage of this project I loved the movement of inchworms and thought they moved in such an interesting yet ridiculous way I wanted my character to do something similar. I watched this and a few other videos of inchworms and caterpillars moving around to get reference for my "walk cycle" my worm was going to do.

Ribbon spline problems

Once my model was all bound and the influence weights were fixed I realised I'd forgotten to create a master controller. When the controller was set up I realised that something was really one when I moved the controller the entire middle of the worm went crazy.

 I was completely lost on what the problems was I tried to resolve it myself I thought it could possibly be do with my hierarchies however I was way off with that. I emailed matt about it were he managed to fix the problem.

The problem was the clusters were attached to the controllers but also parented or parent constrained to the worked coordinates of the controller. This caused accumulative manipulation, this is when a controller is moved the object is moved twice by two forces causing the accumulative motion. Matt used a custom expression which stopped this by tying the clusters to the controllers so they were only effected locally. This completely solved the issue and allowed me to animate my character. The problem was probably down to an error when creating the ribbon spline I should have done more research into making one to get a better understanding this may have prevented this issue.

painting weights problems

Painting weights on the demo character was fairly painless however I ran into some issues during this process on my worm. This was down to not fully understanding what the process does. When painting my weights I found that new influences kept cropping up automatically from unwanted joints. I did not realise that an influence of one meant that part was 100% influenced by that joint. This means if a section is 100% influenced it can not be influenced by another joint. With this knowledge I could fix my influence weights by making sure each segment had 100% painted for there respective joints to stop unwanted influences, this completely fixed the problem.

making controllers and skeleton worm

This was a simpler task in someways the creating a skeleton for a biped however I did have to factor in the ribbon spline. The most important movement was the worms body, all the humour of the character was going to derive from its stupid movement. The other two important movements where the eyes and the head. I created a separate skeleton for the head, eyes and arms of the worm. I found orienting/positioning the central controllers quite difficult, this is because I still wasn't quite sure how the ribbon spline worked even though I'd successfully made one. So I was unsure what I needed to position the controllers around (joints or clusters). I ended up orienting them round the clusters however I'm still unsure if this was the right decision but it seemed to work.

It was important for some scenes that the arms could pop out of the worms body and the mouth was able to move. I was unsure how I was going to do this my original plan was just through rigging however I was told a much better and easier way was to use blend shapes. I had to make a different blend shape for each movement I wanted to achieve, there were five movements. The results created from this were perfect and created some great movement. Creating a good blend for the thumb and arm was difficult however I found it was a lot easier to do it by snap verts in un-smoothed view. I used set driven keys for the blend shapes and put them onto relevant controllers for when I wanted to animate them. The set driven keys allowed a much greater range of in betweens in the motion of the blend shapes. 

modelling my worm

Once we'd received feedback from the interim crit we were able to model our characters. I struggled at first on how to approach making this character as it was very different to the demo character we'd be building. I tried making it by deforming a cylinder and this produced a horrible result this was because I was trying to cut corners thinking I'd speed up the process. By doing this I ended up waisting more time. Matt showed me a much better way of starting which I then used to start it from scratch. It was far more similar to the way we built the demo character, I realised then that this is the best way to start a good character to build the best geometry possible.

I created one segment then I duplicated it across a few times. I deformed it on each iteration as I wanted the worm to be asymmetrical to give it a goofier feel.

Producing the face was a little more difficult because of the shape it was hard to make sure it had good geometry whilst also looking good when smoothed. The face was the most cartoony part of the character design I was worried it would not translate very well however I was pleasantly surprised with the result which I thought was very successful.

I finished the modelling process by mirror the character. This should have been the easiest stage however I had a few problems. The first was that I forgot to delete my history which stopped it working the first few trys. After that was resolved I had a few issues with verts that merged which should have. This was an easy fix by just creating some more in the right places.