Four: Animation and a New Laboratory

I've been working to improve the animations for my project. Feedback has it that where the most improvement is needed in when Four runs or walks. However one of the larger complications has been that in order to smooth out the way hands moved about, the control around the wrist was re-parented so that it moves with the base and not the torso.

This has the added advantage that she can more easily grasp things, and the hand stays in position better when Four is stationary, but it does mean that all the hand positions need readjusting.

  One update that might have slipped is an improvement to the scale of the rooms. Everything was a little large and off-scale so one of the challenges has been to compress a few things so the environment feels less cavernous. This may have a knock on effect in that Four would need to be smaller. Part of what I learned while researching for the project is that grubs can be much larger than the insect they grow into, likely because the fat and material they store up is also what keeps them alive during the energy-intensive metamorphosis project. It might be a simple case of not making the size differences as drastic but at the same time, Four isn't a low-level predator, and animals whose shoulders may reach the waist when on all fours can stretch taller than an adult human when they try standing on their hind legs.

Four: Adjusting Cameras, the Grub is Live

  For the past couple of days I have primarily been working on creating animations for the creature to move about the world. 

  More significantly, the camera's focal length has been changed to provide a more natural look. Looking back and forth the previous camera angle felt a little wide, exaggerating the distance between the foreground and the background which made the rooms look enormous. So far it is only the sequences before and with the grub and with the adult form that the camera angle has been adjusted. The camera angles for the juvenile form will need to be updated and that is on my list.

Most of the camera angles have remained the same save for the distance the camera is from the subject, save for one or two minor changes to improve flow and keep interest. Some of the angles had to be re-tilted as they pointed towards the floor but the improvement does feel significant.

Four: Back in the Game

  The Easter break knocked me out a bit but now I'm back in the game! The Juvenile has been roughly worked out so what is upcoming is working on the adult's animations. For all the work I made on the model, this one is the briefest on screen, but it's also perhaps the most tense sequence as it shows Four's final escape.

  At this point, most of the sequences are done, some basic polish is needed for the corridor escape but with its completion I plan to go back and improve the placeholder animations for the grub, as all that form really does is slide around its incubator like it is living on ice.

  To demonstrate her freedom the idea was that she would open the airlock to a blinding light, finally getting a chance to see the world outside of her laboratory prison. Initially a basic white, this was tinged to be more golden to better indicate it is sunlight.

  Four's model does receive light, however due to the shader used she doesn't have an ambient occlusion shader on her form, this and a lack of any external lighting is what is creating the pitch black shape of her profile.

  The idea for the final moment is she will be temporarily blinded from the light from outside, before leaving the base and running to freedom.

Four: Sequences Improving and Teasing Finer Details

As I'm getting into animating, I'm finding myself occasionally adding the smaller details. As can be seen in the final few seconds, her face does indeed move. I had initially planned to reserve this for second pass animating (to add finer details after the bulk movements are sequenced) but it certainly adds a certain something to the character's life at this early stage.

Four: Perfecting A Jog

  Animation sequences continue. Much of today was spent in an attempt to get Four's jog/run right. A bit of referencing on otter run cycles helped to correct things after some difficulty and this might be good enough to install into a previous segment.

  A pre-vis section will come soon  I am nearly done with the first pass of the juvenile sections to the piece.

Four: Office Exploration

  More progress done, this time the sequence where Four explores her owner's office as Amelia expresses doubts continuing a relationship with him. A bit of time was taken to get her rise to the table looking right.

The way she pushes the phone off the table, maybe there is a connotation there given the way she pushes it? She's not entirely sweets and strawberries especially as she gets older and the other scientists and the people upstairs start thinking of David less.

Four: Skin and Colour

  With the model's rig just about ready I moved on to working out the textures. When making the normal map I found that the some regions were divided by seams. These seams were blended over in Mudbox, a technique I also plan to use to clean up the divides between the texture meshes, giving much cleaner result that's faster than trying to fix the seams in Photoshop.

However the normal map appears to be a little faint, but it is there and on more extreme areas such as the throat, it is visible. I am considering going back into Mudbox to bring the shapes out it might just be I need some more time working out what could be defined by its shape and what would be defined by patterns. The possibility exists of also using a displacement map so any minor changes to the geometry do not require restructuring the corrective poses.

  With the above picture I tried combining a greyscale of the normal map with the texture map to bring out the features. I consider it a bit of a mixed success though, as it has also ended up making poor Four look like she has severe sunburn. Beyond that, some greenish tint to coincide with her environment and some wrinkles and she's almost good to go skinwise.

Don't mind her lack of eyes. For efficiency's sake I decided to create the texture map on a copy of Four's geometry that matches the geometry on the rigged model. All the eyes really need are a texture map with the pupil stretched across the top edge of the image file.

The teeth and claws could easily be a shader at this stage. I just need to revise how to give the transparency of the edges. The idea that I believe will work is to give a mock rim highlight. The texture files are so far somewhat large - 8192 pixels square which I will likely keep for extreme closeup shots. Smaller files can be easily created for more distant or full-body shots.

  In fact I will probably need to do this, such size extends to the normal and specular maps which will eat processing power during rendering. Understanding my shots will be important in working out the correct resolution required. the current size might be too large for what I'm doing save for extreme closeups.

Four: Corrective Blendshaping

  I've been quiet but that doesn't mean I have not been busy. Although a long couple of months did put me mostly out of commission for a day or two.

The transformation of the throat works a fair bit better now. Looks a little more convincing and I find it cool that you can see some of the muscle in action.

  At the moment I have a very slight problem that when adding a twist to the wrist, the value would set itself to a value greater than or less than 1 and it would hideously deform to look like the wrist was grotesquely swollen.

  The arrangement I have so far is the best I can do with the current wrist despite attempts like clearing any influences around the joint before working the vertices. Though oddly this is the only problem site so far as there was no problem working the twist into the shoulder.

Colo Colo: First Animations

  I was unable to update my progress on the weekend but after essentially re-skinning the model I feel confident I am at a state I can animate. So what I attempted this afternoon was an animation for Colo Colo running to start me off. Errors in the model have come up, and I feel I may have made the thighs too wide but progress is progress. Over the next couple of days I hope to build a set of action animations to give the creature some personality. The time spent correcting my mistakes has however eaten into time that could have been spent with other preparations.

  I am not dissatisfied however, as this entire project has been ambitious compared to what I have done previously, and I am not disappointed as I have always enjoyed the concept of a challenge. What I have produced so far I do not feel came out badly.

  Enough confessions, I also prepared a playblast. The difference between this and the rendered version is mainly from improving from what I had discovered using this initial animation sequence.

Colo Colo: Head, Feet and Chest

  Progress is developing nicely with Colo Colo, progress with the texture maps are coming along. For the most part I think the body is in a state I can consider it finished.

  The one issue so far is that there is a problem with the thigh's UV. One line down the very front of each thigh for the rear leg seems to be getting errors. When smoothed, this one line wraps around itself, creating an artifact in the form of a highly compressed duplicate of the rear leg's texture that fits between this vertical line of UVs and one set of UVs on its of the UV map, effectively projecting the rear leg's UV down one line of faces. This error however, does not show up in the projection in the viewport but that can be down to a lower-quality projection.

  One thing I have definitely concluded working on this is that at some point I'll need to go into Mudbox to fix the seams between the projections. I had attempted to mimic the colour of the legs between the two projections, but due to the way I had shaded the leg in the body texture map, the translation wasn't perfect. The above render is the closest I might be able to get for now.

  Errors aside, I have made a lot of progress on the head for Colo Colo. Texturing as something to resemble the concept art I had made, I think I'm fairly close. The other major thing I added was a mechanism to mimic the effect of the creature's heart beating or lungs cycling (chest motion essentially). At present it may be slightly too rapid, but I can control the rate as the chest expansion is tied to a driven key and an attribute in the shoulder control. The testing speed is based on how small creatures tend to have very fast heart-rates, as well as looking at some footage of lizards and myself in a mirror to get an idea of how extreme the inhalation and exhalation is during the breathing cycle.

  The beauty of controlling the rate like this is that with only a couple of tweaks, the breathing rate can be refined, or this rate is for when it exerts itself. Though changes in rate might involve either swapping out for a fresh duplicate with the differing or transitional rate, or input each cycle by copying and pasting.

  Or a MEL script.

Colo Colo: Rigging the face

  With a partially done texture in place, priority today was on installing the final controls which were for the eyes and eyebrows. This was achieved by keying the transformations to blendshapes set separate from the main model, the transformations on these blend shapes would be reflected on the main model. The main downside to this measure is there are three instances of Colo Colo in the scene, but these can be hidden both in the workspace and in the outliner.

  The eyelidss were a bit of trial and error to get the angle right, as they're not human eyes. One of the alleviations I found during testing was angling the eyes so they twisted slightly as they closed. Not what normally happens but it allowed me to somewhat preserve the positions of the corner joints. I also made the discovery that I might have added too few polygons and so there is quite a bit of stretching on the upper eyelid in particular, but this can be rectified by a tweak to the UV.

  The sub-surface scattering is so far coming out reasonably well, even on the current skin. This can especially be seen i nthe light-obscured areas of the body such as the right-hand thigh. The priority now is goin to be textures, so I can make Colo Colo look really nice

  When rendered out, I think the texture comes out fairly well. There's some artifacts but these can be fixed by some minor modification of the texture map inside Mubox.

Adaptation B: Ink Stain Practice

  Yesterday evening, before streaming I investigated using ink stains to help get some idea for shapes to make the Colo Colo with. The set of images above were made on coarser paper than the images below, which were painted on cartridge paper. I plan to use these either to create a composite image or to combine these with the blackouts I created in previous sessions.

  I like the more spider-like qualities these ink stains have, and some of them I can see a vague shape that could relate to Colo Colo (5 could be used for a mouth while 13 could be a leg or a paw). The use of ink-stains gave me less control yes but also a greater sense of expressiveness.

Narrative Project: Bringing A Branch Coral To Life

  Today was the commencement of one of the most important elements: The coral that would chase our fish through the canyon. Since the start of the production phase of the project the idea was the coral would grow in order to pursue the fish. For this reason, what we needed was something that could look like it was expanding during its pursuit. We agreed from the early stages to see what we could do about  making this possible with blend-shapes. And there was talk of interpolating various bend-shape scales.

  To do all this I ended up considering a few methods. The one I have currently settled on was suggested by Alan, where I create a blend-shape stage for each set of buds that would then be combined into one slider sequence. I have spent a healthy portion of the afternoon preparing all stages.

  To make this work, I have grouped each budding into a set starting from the top, I have marked buddings using NURBS circles and grouped each budding that forms simultaneously into their own groups, using the NURBS circles as identifiers. These groups were then added to their own render layer for easy toggling so that I can work my way down. I then set about the long task of compressing the whole coral down to the base.

  It is a painstaking process but currently I am a little under halfway through the process, which makes it worthwhile foresight to have created the anemone tendril yesterday. My hope is to have this structure complete by the end of the weekend and possibly rigged for flexibility. But rigging with a skeleton is on the 'maybe' pile as this might be able to work well enough on its own. I estimate that I will need perhaps twelve different models to create the blend sequence.

  I had a few other ideas that worked in a similar fashion. These involved rigging the coral to expand sequentially, however as can be seen in the image above, previous attempts had a few visualisation issues. Attempts to rectify this with proper timing proved rather time-consuming in order to hide the visual errors and I I progressed I discovered that the meshes were folding themselves inside out.. 

  I learned early on that when compressing a model, using the central square in the scale manipulator will quickly invert the model. So each compression to nothing was done by way of compressing the dimensions on each axis, which unlike the uniform manipulator, didn't invert the model.

Narrative Project: The Scales On The Fish

   My primary task this afternoon was to get a working system for having a set of alien scales grow from the body of Ravine's main fish characters. To do this it was suggested to me to use blend-shapes, which I spent the afternoon working on in order to get the transformation right

  The core idea was to create a layer that would go within the skin. This layer would fit to the body by being wrapped around using a lattice deformer. At the time a model was unavailable, so I made do with a sphere contorted into a rough shape to represent the fish's body. The final setup would need more precise positioning as the the idea is that the scales would grow from around the eyes and sweep towards the tail as they spread.

   Because I was using blendshapes, I could not simply use the duplicate tool, as that would only duplicate the geometry of each scale and not the blendshape.

  Another possible change to the final model is perhaps something less formally arranged. This is supposed to be a parasite, something alien to the fish so sharing the aesthetic of the environment it should be distorted and strange.

  The idea was that the spread would progress along the side of the body. But to give it a natural feel the scales should ideally look like they're growing independantly. This was achieved by giving variance as to when the scales would start and how quickly they would stop growing.

 We may be able to simplify the visual appearance but this is the UV map of the scale. Structurally however I think I have stripped what I can without losing perceptual quality. Though I do wonder if smaller, lower-polygon versions would be useful at any point.