The End: An Evaluation

And so we come to the end of my final year at Glamorgan University, and the end of my degree.

It’s been a wild ride, particularly with this final project; I ran into time-management issues early on, mainly as a result of taking too long fussing over details, and procrastinating on tasks I wasn’t quite sure how to handle (trees for my environments being one strong example). Some projects evolved; one collaborative project was intended to become a fully-fledged game demo but floundered upon the team’s introduction to the engine, the other gave me experience with trying to replicate something that my creative lead had a very specific mental image for, and the back-and-forth exchange of ideas and feedback that took place in that project was something that I found extremely invaluable. 

Overall though, I’m rather happy with what I produced in the end. It’s a rare thing that a project I complete lives up to the image I have in my head at conception, but this year, I think it’s actually been pretty close, with the Knight and the Garden in particular. I’m sure that in time I’ll look back on both of these projects and see them as hopelessly amateur, but right now I’m happy with them and that’s all I can really ask for; as an artist, I’ll always be looking to improve in every project, and I suppose knowing I’ll dislike this work when I look back on it later is (hopefully!) a sign that my work will have evolved further in the meantime.

If there’s one thing I’ve learned from this final project, it is this: take the plunge. If you’re unsure of how to do something, don’t wait around until the perfect solution arrives in your lap, test things. Experiment. See what works. I did this when experimenting on how to combine multiple textures for my landscapes, and it paid off. If I did these projects over again, I would have liked to do the same for my work with planes and alpha-based shrubs and trees – I hesitated with these, and it cost me. Some of my attempts at these were successful, others less so; I would like to take more time to examine how some succeed where others fail, and why.

I still feel like I have a lot to learn, though! It’s the Dunning-Kruger effect in action, maybe; every time I learn something new, I discover a little more about how much I have left to learn. Overall I’m pretty happy with how my final production work turned out this year, and I learned a lot, but I have a few weak spots that I want to work on improving soon; in my next project I want to use Maya, learn how to effectively polypaint in Zbrush, and possibly experiment with retopologising software. I think I’m going to be using the design for an existing IP for this one, so I can concentrate on the technical aspect rather than the design, but it’s a design I’m extremely excited about. I’ll say no more at the moment, but: watch this space for updates ;)

Peace!

-          Liz

 

I'm nearly done with the brother characters - just making some final tweaks to the face maps before I hand them over for a final check from the director! After that, they'll need to be skinned quickly and put into the poses needed for the cinematic. I'm pretty happy with how they've turned out, and when I look back it's amazing how the elder brother in particular has evolved over this process in terms of proportions. They're now at a point where I'm finally pretty happy with them. Children are notoriously difficult to draw and model, and these boys are a long way from realistic, but they are finally at a stage where I am reasonably happy with them. And as a learning experience, they've been invaluable.

Do they look like brothers to you?

Brothers update!

Working on refining this model's face map. Face textures always seem the most difficult to me - I'm working with only a simple phong shader here instead of using a renderer capable of sub-surface scattering, so it's all about striking a careful balance using the diffuse, gloss and specular maps. Here, I'm putting the final touches to the diffuse map.

Displaying the textures in my viewport is having the slightly unnerving side-effect of making the character look demonic, though - that's a side effect of the method I've used to give some depth to the eyes, but it doesn't half look strange!

She's just about finished! I'm really pretty happy with how she's turned out. I've just spent the last hour resizing all my textures for her - my diffuse and normal maps in particular were being worked at either 2048 or 4096 pixels, mainly because I subscribe to the view that you can always size something DOWN, it's always much harder increasing the resolution if you make it too small! However, I was rather suprised at how little detail was lost when I sized these down, particularly when it comes to gloss and specular maps. The lesson here is that those maps, in particular, do not need to be particularly large!

I'm still a little unsure about her hair shading - I'm using an anisotropic shader set at 90 degrees to create the effect, but I left the highlights quite soft and wide. Making them harsher had the effect of making the hair seem wet and glossy rather than "hair-like": shall have to do some further research on this, and try and improve the effect in future projects.

Unusual Normals

Believe it or not, I've never actually made a tree using alphas before. After a bit of research about how best to go about this, I decided to take a crack at it, build my tree and then try a Max script called SlideNormalThief. The idea with this script is that it will transfer the normals from one object to another, thus enabling oneself to even out the direction of normals across the entire surface of the tree canopy. The first screenshot is my tree with the reference object for the canopy overlaid on it.

Second screenshot is the vanilla plum tree, as it is with the standard normal.

Third screenshot is the plum tree with averaged normals.

Is it just me, or does the second screenshot look significantly better than the first?

After some consideration, I decided that actually, it looked better without altered normals. Experiment failed, but lessons learned. Anyway, here are my other tree trunks! They're awaiting their foliage.

Revenge of the crates and barrels!

Making more crates and barrels? Say is isn't so!

I've just started texturing my campsite. Some of the first targets - the supplies crates, and the flags. Since this is meant to be used in a snowy environment, I've used 3DS Max's Viewport Canvas to paint layers of snow onto some of the static objects, like the flag: but for repeatable objects, like the crates, I've had to avoid that. They need to be able to be placed anywhere!

I've just finished the first pass of the body textures for the brothers. It's not perfect, by any stretch of the imagination - there are specular issues and the occlusion is rather heavy-handed and needs to be taken down a notch (or several!) The front of the younger's jacket seems rather flat and "blocky" to me in the mesh, too, and needs a bit of relaxing. The next step with this is to get feedback from the director - this is the first time the clothes have been textures (and coloured in something more effective than quick block colours)! Before I go any further and make changes, I need to make sure the general direction I'm heading is approved by the director - these are his characters, after all!

Maybe she's born with it?

One thing that's really been relearned from today: it's worth paying attention to real-life makeup tips when modelling! The old adage about mascara making your eyes look bigger turns out to be true. I created the headsculpt of my knight and finished it to my own satisfaction a while ago. But upon blocking in eyelashes this afternoon? Suddenly her eyes looked ridiculously large! Luckily it was something I could fix with some use of the scale tool and some soft-selection of verts on the head and not something I had to dive back into Zbrush to fix, but it's another thing to bear in mind the next time I sculpt a head, particularly a female one.

Just another lesson I've learned this year when it comes to poly modelling heads, I suppose!

The Grid

Have been doing more texture work on the Japanese garden. I've got some problems with mismatching resolutions at the moment; I need to fix that. I'm pretty happy with the revision of the grid texture, though, which you can see in the second picture. It's beginning to come together!

The illusion of life

I'm working on some of the strange, peripheral parts of my knight model at the moment. This chain of screenshots shows the progression: after my last post, I resized my character's eyes slightly before creating an alpha map for her eyelashes.

Then, I did some experimenting over how to give her some eyes that had life in them. I'm aware that some game engines have very clever tricks with shaders to create an illusion of depth using a simple sphere for the eye, but after a search I could find an easy way to replicate such an effect with the simple scanline rendering I'm using for this. In the second screenshot, you can see me weighing up the advantages of making the mesh for the eyes more complex: the eyes at the bottom are simple spheres with a material and texture, but the eyes at the top are two spheres. The "eye" sphere uses the same map as the bottom eyes but has had the iris area flattened out, while the "gloss/lens" sphere is still perfectly spherical. It's got a simple thin-wall refraction material on it to simulate some level of depth. Looking at these tests (particularly when looking at an angle), I decided that adding the extra polys was worth it. There is something about the upper eyes that just look more lifelike.

The third screenshot shows my progress on creating a simple normal map for the inner mouth (it will eventually be added to th face map, where there is dead space I'm not using). I'm still in the process of creating this.

I've been building cliffs over the last few days!

I think there are only about 4 tris in that whole mesh. The rest is quads. I feel like this might be unnessecarily finicky of me.

Knight update!

I'm busy creating the alpha planes for the hair at the moment. This is my second attempt, and it's looking much more promising this time. There is still, of course, a lot of texture work still to do!

The first screenshot is is a nice opportunity to show how my edgeloops are placed, but really, I rather loved the co-incidental rainbow the wireframe colours were forming when combined with the normal map I had sitting on display. Sometimes even the most accidental, incidental things can be visually pleasing - at least to me!

Campsite

I've started my final project - building a campsite! It's mainly a project centred around small props - a project based around creating small environmental assets that may be the kind of thing I'd hope to be doing in emplyment should I get a job as a junior artist. Here's hoping!

This is the result of a few day's work: there are a few things (a sword, a basket, and the shields) which need further modelling, everything else is unwrapped and ready to be textured.

I think I might want to put more detail into the tent, though!

The changing faces of Casper

Children's faces are hard. Believe it or not, these three screenshots are all of the same head mesh, all of the same character. I'm fast learning that you really, really need to make sure that "perspective" is turned on when working in Zbrush, because it can warp your vision of what you're doing to insane degrees if you're doing finicky work like heads. After getting the first head back out of zbrush, I found I was hugely unhappy with the shape of it, he just looked too alien. But I'm fast learning that it pays to carefully use your layers and save your Zbrush tools, so luckily I have the ability to quickly alter the base shape of the head underneath all the detail work.

The third screenshow has an altered chin and side of face, mainly accomplished using the relax tool. I'm still not happy with it, though; it needs a lot more work, and some more collaboration with the person in charge of modelling the "older" version of this character.

Modelling children is a steep learning curve. But I feel like I'm learning a lot by doing so.

Younger Brother: Progress

I've finished the normal map for the younger brother! ....As you can see, I've yet to start on the specular or diffuse maps, though.

I am quite happy with the head sculpt, though. And I think it's interesting how your perspective on "what looks good" can evolve so drastically once you start having points of comparison. The first image in this post is the exact same model I posted when I had first finished the model (albeit with fixed smooth groups!). It looked perfectly fine to me at the time, and was built to the exact proportions of the model sheet/turnaround I had created. But looking back now, his eyes are so big that it makes him look like an alien! I fell for the assumption that to make someone younger, you make their eyes bigger - whereas after some more observation, it's really more about a softening of features.

I'm not entirely happy with the way the mesh is behaving around the lips, though. It's in danger of going into unintentional "duckface" territory, but I haven't decided how I'm going to go about fixing this yet - either by diving back into zbrush, by editing the normal map or by messing with the base mesh itself.

After being unhappy with the first sculpt of Caspers head coming out of zbrush (it was strangely wide, although I'm not sure how I didn't spot that in Zbrush!) I went back and resculpted Casper's head. This is the export of the lowest level of that sculpt, to have the high poly projected onto it. This version looks far more promising than my initial attempts, thankfully.

I am slightly worried that the character's hair is going to look "plastered on": I do, however, have to do some research regarding hair alphas for game use for my personal project, the knight. If I can get hair working nicely on that model, I may experiment on giving this character hair with alphas as well.

I've been building lanterns and props for the last few days. Becoming a little neurotic over sharp corners and chamfers. Where does one draw the line? I don't want things to have edges that are too sharp, but when the props are this small, you don't want to overdo it and clog up the scene with unneccesary polys and verts.

There's a mix of approaches here. The tallest lantern favours chamfering, but the lantern on the far left is fairly plain. As long as they all look good from a reasonable distance, then they're fulfilling their purpose.

Knight progress!

I've just started work texturing this in Zbrush. The armour is complicated, so I'm sculpting it piece by piece and assembling the normal map in photoshop as I go. So far I only have the upper shoulderplates, elbow pads and upper capes normal mapped - this is just me testing out how they look. I'm toying with the idea of using multiple shaders on this one object (one for metals, one for softer materials I can use a phong or blinn shader for) although both use the same diffuse/spec/normal map. I forsee some toying with shaders in my future, but I'm going to concentrate on finishing the normal map!

Just finished the base mesh for my knight character! There are a few things I'll need to go back to: I'm not going to create things that rely on snapping to vertices that might yet move at the moment! Hair, eyelashes and inner mouth are among the parts of the mesh I haven't created yet. Next step is to unwrap this all into two maps (one for head, one for body) and then take them into Zbrush for some high poly sculpting.

Garden update!

I've been building mountains! It took a suprisingly long time, and I'm sure there must be some shortcuts to doing this kind of thing that I missed. The garden suddenly looks a lot more complete now, though: it's nice to not have a gaping void where the garden's exit should be!

The start of a puzzle....

Beginning to assemble my garden assets into something vaguely resembling a scene! I have to say, it's immediately exciting to see these assets come together, even though I think all of them need a little more work at the moment!

I've finished building my fence! I can't say I'm done texturing it yet though, as I'm not yet convinced about the wood texture. I've designed this to be modular pieces: each part joins neatly onto the next to create a more complex fence system. The idea is to be able to create instances of these rather than copies, in order to reduce memory time. I have to say though, I'm quite happy with the Zbrush sculpt. I was unsure about Zbrush at first, but as time goes on, I'm beginning to like it more than I like Mudbox!

Arches!

My arches have been modelled and textured! The smaller one (at the back) is going to be cloned dozens of times to make a tunnel. I'm looking forward to setting this up and seeing how it looks! These tunnels were one of the most striking sights I saw in Japan when I vistited, so I'm excited to how my own version will come together.

These arches are usually highly varnished, so there's quite a light spec map on here combined with a high gloss setting. I've finally settled on this version but it might be subject to change if the design of the overall level ends up needing a "softer" feel. This is also a clean version - I considered dirtying these textures up (scratching some of the paint off the wood at points, muddying and weathering the posts going into the ground etc.) but I think I'm going to leave it for the moment until I see the level of realism other members of my team are going for with their work. These existing textures can always be dirtied up later :)

The block

Last year I never quite got the hang of projection mapping in 3DSMax, instead preferring other methods of normal map generating. With this block, I took the opportunity to rectify this a little. I built the low poly version first, unwrapped it, cloned it and split it into component parts, then cloned the parts and built them into high poly versions, adding structural detail and them putting them through Zbrush to add extra details. The method works pretty well, although it does carry the risk of me crashing my computer when reimporting the high-poly versions into Max ready for mapping. Looking at other reccomended workflows it seems I may have been working backwards slightly by building the low poly first - but then again, it made my unwrapping job very easy and it meant the low-poly model was optimised from the get-go. Swings and roundabouts I suppose, but I'd love to know what other modellers prefer as a workflow. Sculpt -> high poly -> low poly, high-poly -> sculpt -> low poly, or the way I made this?

Brothers: Initial Models

I've been working on the base meshes for the brothers, and I'm finally getting to the point where I'll be ready to unwrap it and take the parts into zbrush! There are a few things to sort, most notably areas around the collars and cuffs. I also need to recheck the smoothing groups.

The elder brother, Casper, actually has two heads - the first was modelled roughly and then taken into zbrush with the intention of building a beautiful model from the rough shape - BIG mistake! Firstly, I needed to take more time to make the head a good base to work from. Secondly, my edge loops were terrible, particularly around the mouth; instead of curving the loops for the mouth round in a tight circle and routing the edges radiating out from it around the chin and eyes, I've been looping it round the back; very inelegant. I redid the head and I feel it's much better for it. Hopefully it will also be a better candidate for sculpting in zbrush, too!

Preproduction: Finished!