Wednesday, 27 February 2013

Modelling the Environment, Setting up the Render Camera and Animating the Alien

For the modelling process, I used the front orthographic view in Maya. I projected an image plane of my interior concept art and modelled the environment, working in layers relating to the photoshop document, which was then later taken in to After Effects and given a slow zooming motion. The photoshop layers seemed to make sense in terms of organisation, so that in Maya I could hide certain layers when accessing geometry for UV mapping, making the process less time consuming. You can see how they layers were organised bellow:

Consider the HD format of the recorded video footage for the whole project, I then went into the render settings and changed the width and height to 1920 by 1080 pixels. As my original concept painting was created in an A3 format, I had to then set up a new orthographic front view titled 'render cam', enable a resolution gate and slightly move and scale the interior geometry until I was happy with the composition.

Although the scaling of the scene deviates slightly from the original A3 image plane, the composition strongly resembles the concept design for the shot.

I tried exporting my rigged alien as an FBX file, but could import the geometry with the rig still attached, meaning my weight painting would be lost. As there was no animation involved in the interior, it proved easier to export this scene and take it over to my rigged character scene. This did mean I had to set up the render camera again, which wasn't too difficult as I was working in the orthographic straight on view, meaning no angles had to be considered in relation to the green screen footage.

I decided to include and animate only one alien, to the right of the shot where the composition feels more empty around the central cocoon. I feel that when animating in Maya with the automatic easing animation curves, it is easy for movement to be slow and laboured. I wanted to try a dynamic animation, meaning some of the alien's movements are very fast and sudden:

There is a sense of anticipation at the start of the sequence where the alien very slowly turns to scan its surroundings. Suddenly it sharply snaps its head to a full turn as if startled by something. This adds to the sensation I discussed earlier, of the feeling they are being curiously pulled in, yet being intrusive adding to the suspense of the shot. Note that when the alien jolts the opposite direction, there is a sort of kinetic chain principle working up from the ground. First the hips drive leaving the upper body slightly behind to catch up. One could argue that the head would instinctively lead if startled, yet I like the idea that the alien is very attuned and dialled in to its surroundings, continuing to to scan the original area where a threat may lurk, whilst preparing to snap around simultaneously to catch out any flanking manoeuvres. With the fist controls, I added expression in the hands. The hands often signify when the alien is tense, for example when the creature inhales and lifts its shoulders, the fists clench. Perhaps the animation could have been improved if I had have included some more follow through action in perhaps the arms with the sharp aggressive turns. Also, with the first quick full body turn (kinetic chain movement), there is a slight jerk in the middle of the animation which I could not seem to solve in the graph editor. Another slight floor is when alien turns back to face the actor in the cocoon. The creature slightly over rotates and then readjusts. This could have passed as the follow through principle, but looks jerky and unrealistic. I am also aware that the right thumb seems to have broken and moved moved into an irregular position. However, I am hoping this will not be too noticeable in the low-lit distant final shot.

Saturday, 23 February 2013

Modelling the 'Termite Mound' Ship

With this design being a very chaotic organic one, I felt I could experiment with using only the front image plane from my concept drawings, and leaving room for some creative freedom within Maya. The  design was fairly simple, so after separating all of the spikes within the design from the main UV, I again placed in the coloured grid to make sure the texture was flowing nicely around my model:

I made a few very subtle tweaks to the UV map, but generally the parts in view looked very tidy. This lead me on to the texturing stage. Again I used Nagel's custom skin brushes to get the tessellated bee hive effect. On top of this I did a bold what I like to think of as a 'circuit board' effect. The bright purple pattern follows through some of scales created from Nagel's brushes, as well as breaking off into smaller patterns. This gives the impression that some form of energy is flowing through the ship, giving this organic meets synthetic theme to the design. My original bump map consisted of a desaturated complete image, with the levels tweaked to add more contrast and depth in the texture:

I decided I wanted the purple pattern to maintain the bump texture of the layers beneath, giving the effect that it is glowing from beneath the skin of the ship, almost like a painted on effect. I am aware of self illuminating maps in Unity, which perhaps is something I could look into in Maya to give a greater glow effect to the pattern layer. For now, here revised bump map texture:

Wednesday, 20 February 2013

The Rough Cut

Before progressing further with the modelling of my 3D scenes and assets, I put together a rough edit with my collected footage:

This edit focuses on the timings of each shot and sound placement. The timings will be useful when animating the 3D elements to correspond with the length of the shots. This means I am not spending time animating what would be cropped out in the final composition. Also, the consideration of the sound composition at this early stage will mean I am not scrambling for sounds at the last minute, which  often tends to happen when focusing on the visual side of things and neglecting the equally as imported audio elements. For the final shot I used the same sounds from my after effects moving composition. The other sounds were a mix of online clips and recordings from the original shoot.

As you can see, the final green screen shot was fairly straight forward when it came to shooting. Our filming group had two redhead lights shining through dampeners to illuminate a green screen cloth. With another group member I then held some green screen silk over the actor hiding most of his body. In hindsight, there were a few different approaches I could have taken. I will likely be masking right up to the actors neck when compositing, meaning the green screen will only be relevant around the head. With this in mind, I could have potentially set up the small pop out green screen and grabbed a close-up shot of the actors head, to then scale down to size in After Effects. With our group setting up the large green screen cloth anyway, this wouldn't have saved any time, but is something to be aware of in the future.

I feel with the rough cut completed, I can can now begin animating as soon as possible.

I felt it would be interesting to have a look at the pre visualisation or 'previs' process at a professional industry level. The video below shows how  3D pre visualisation was used for 'X-Men: First Class'. A devoted team incorporate 3D animation tools such as motion capture to clan out the framing of shots, prior to filming. Perhaps it could have interesting to produce a 3D mock-up as a group to inform the framing our shots in advanced:

Rigging Issues

Instead of recording the entire process of rigging, I felt it would make more sense to reflect on some of the problems I faced trying to rig my bizzare insect creature.

Placing the joints and building the skeleton to fit my geometry was straight forward after following tutorial videos placed on our virtual learning environment from a previous module. The key issue I faced throughout the process was some of the strange angles I had included in my turnaround reference image. This wasn't too much of an issue for the spine controls, but for the forearms it caused some problems. Because the arms weren't stretched out horizontally with my model, I couldn't lock off the orient constraint to just the X axis. At first this only meant I had to tweak the wrist on all axes being careful that the translation made sense with the position of the shoulder joint. Where this effected my model however was with the forearm roll function. I planned on using a different multiply function for each arm, using all three inputs/outputs for each axis. However, translating each axis the forearm roll joint by 50% of the wrist joint caused the rig to break:

I decided to leave the forearm roll joint as again my model will be performing a simple stationary animation in the distance when it comes to final shot. However, next time I create a turnaround I will depict horizontal outstretched arms. Perhaps I could have bound the geometry to the skeleton prior to adding the forearm and hand controls, and outstretched the shoulders manually in Maya. After doing so I could then have added the controllers and only oriented along the x axis. This is something to consider in the future when modelling and rigging in Maya.

Weight Painting was relatively successful. I began by taking away the influence on the geometry from all the end joints apart from the toes which require influence when creating the reverse foot lock controls. I find that often the head and shoulder joints require a significant amount of attention. The head usually has influence from neck and shoulder joints which causes parts of the head geometry to warp and drag as the controls are moved. With the helmet, I simply needed full influence on the head geometry from the head joint. I also usually find that the shoulders need pinching under the arms as to not drag the chest and rib areas when the controls are rotated. I usually find that removing influence from around the arm pits adds some level of realism. One problem I experienced around the upper body was with the nose of the helmet as shown below:

I immediately assumed that some small level of influence from a joint lower down the chain was causing the vertex to drag down. However, after going through every single joint with the colour gradient mode set I simply couldn't locate the culprit (From the screenshot I am aware that working in high quality preview mode is bad practice when not focusing on texturing or lighting. When working at home I always work with the textures deactivated to make Maya run as smoothly as possible. However I grabbed this shot whilst experimenting with my UV textures on the fast running college computers).

The most frustrating issue I faced was weight painting the feet. Because of the stubby length and height of the feet, the joints felt squashed up and most of the time caused the feet to shrink when tweaking the custom attributes. It was hard to remove influence from the ball joint of each foot for the toe tap and peel heel controls. It meant that the peel heel came out looking like a limited 'stand tip' control without the ball joint firmly planted on the ground:

As you can see the joints become squashed together even causing the geometry to appear as if it's shrinking. The toe tap was equally unsuccessful:

Again the alien isn't so much tapping its toe as rocking back completely on its heel. Fortunately, with the creature remaining stationary in my final shot, I will not need to worry so much about the peel heel or toe tap controls. If the custom attributes had been successful I could possibly have applied them very subtly to add extra detail and realism to the animation. Fortunately, I managed to apply the fist controls, animating each finger separately. Hopefully I will be able to get some expression in the hands with my animation.

The final point I want to raise wasn't an enormous issue, but would have been if I had wanted my alien to walk. I would have had to better consider the use of inverse kinematics or 'IK handles'. For the legs I added one IK handle each running top of the leg down to the ankle joint. I wasn't aware that adding multiple IK handles was a possibility until a classmate pointed it out to me later. At this stage I was already satisfied to some degree with the rig and didn't want to loose time correcting something that will have little to no effect on the final scene. The single IK handle in the triple jointed leg rig means that the geometry founds in on itself when the root control is lowered:

After completing the rig, I locked all of the controls with the limit information in the attributes editor. This means the rig cannot be broken. I then grouped everything together for housekeeping. Bellow is a quick posed model of my alien:

Although I am working outside of my desired area of specialisation, I feel that I am constantly improving my understanding of the workflow involved in the full game development process. For example, through my rigging practice I am now aware of the problems angled geometry can cause when adding adding constraints.

Modelling the 'Bug Alien'

I began this particular process by producing a turnaround image. Based on my modelling skills, I kept the main body organic and simple as appose to layering up armour parts. I then included the 'mosquito' helmet design from my earlier development:

I felt this was good practice in the area of concept art, my main interest within the game industry. I used guides within photoshop to accurately line up features making the image mostly accurate and precise. One thing I did get wrong was the positioning of the eyes. The side view image shows the eyes on the side of the head suggesting split vision like a for example a bird. The eyes on the front view are facing forwards suggesting they are composed like a human's. When modelling I went for the side reference for the eyes as the alien is depicted standing sideways on in my final shot concept piece. One more problem is that the helmet could be seen as a way of avoiding the use of blend shapes. It is true that I animation is an area out of my comfort zone, although I will hopeful find a way to include blend shapes elsewhere (possibly within my organic environment).

As has happened to me previously, my first attempt attempt at modelling went wrong. After setting up my image planes I decided I wanted to start modelling from the head and work my way down the rest of the body:

Although the head itself wasn't too bad, I felt I had already accumulated too many faces at the base of the neck, meaning extruding down to create the rest of the body would have involved lots of painstaking vertex shifting to get accurate results. This is a lesson I should have perhaps learnt from my last module when modelling a spaceship. Again I began by working detail into a small area making extruding from that point very confusing due to the number of faces. Although I am aware some modellers like to began by modelling the head/face, I decided it made more sense for me to start by blocking out the body before confusing myself with edge loops and vertices:

From this approach I was still able to successfully model the helmet. I feel that my understanding of edge flow is gradually improving. My model consists of only edge loops and no stray vertices. There are some five sided shapes (usually avoided) around the calf muscles and the hips to bring out the exaggerated features from my designs.

I displayed this un-smoothed as it clearly shows the arguably unorthodox vertices. When smoothed however, these exaggerated features fit the geometry nicely. One area of the geometry I aren't too happy with is where the legs meet the groin. Below is the tidy looking un-smoothed results, compared with the stretched smoothed results:

As you can see, the edge flow around the groin is extremely ugly. There was some brief experimentation to try and solve the problem, but I couldn't seem to remove the harsh transition between the groin flowing into the inside of the leg. Realistically, my alien characters in the final shot will be stationary, and I feel that only if I wanted to include a walk cycle would the groin be a major problem. Also the creature will be in a low lit cave quite a distance away from the camera. This means the groin will most likely be in shadow. This groin edge flow however, is definitely something to keep in mind when modelling in the future.

I will also talk briefly on producing the UV texture for this model. This is one area I am beginning to feel comfortable in. I began by taking a front view planar projection of the entire model. Then using the 'Cut UV edges' tool, I separated features such as the arms and various parts of the legs away from the main geometry. This was most useful as the three-jointed legs overlaped making unfolding very awkard. I seperated the three separate sections of the leg, unfolded them separately and then sewed the correct UV edges back together. I always tried to hide the seams in places out of view where possible, as this is the place where the textures are most likely not going to be seamless. The using a special coloured grid texture, I tried to tidy up the UV map so that all the squares of the grid were un-stretched and flowing nicely around the geometry. I didn't spent too long on this as again, the creature will be in the distance and it is an organic model so arguably there is room for subtle stretching. As soon as the grid looked relatively tidy over the geometry I took a UV snapshot over to Photoshop:

In photoshop I was able to experiment with some custom brushes I had downloaded from an artist called Nagel. He created a great creature skin selection, of which one of the scales brushed was very effective. The tessellated nature of the scales meant that they reflecting the mathematical beehive structure informing some of my concept work. I added a simple bump map and applied the textures to a new custom 'Blinn' in the Hypershade menu in Maya. This shiny specular texture gave a wet slimy finish to my putrid insect-like alien. Bellow is the completed UV texture map: