JAY TARRANT
VISUAL DEVELOPMENT | CONCEPT ARTIST
VFX PATHWAY
UNIT 6 -
VFX - REALISTIC ANIMAL 3D MODEL
I feel like most people, if you gave them 100% free-will in a project, to create a realistic living-animal of their choice (with a few limitations of course) would choose their own pets.
Of course I would love to immortalise my cat into a 3D Modelling software for the rest of eternity?! If it turns out awful and I cannot accurately model a cat in Blender that looks like my cat, yes I may beat myself up and consider myself a laughing stock. Will it be a bit funny though if I accidentally make a really ugly cat, trying to recreate my own? Yeah, a little bit.
​
In terms of research, I have what may as well be an entire library of reference images for this portion of the project. Unfortunately, I am not going to be able to post a whole lot of them on this blog, as they are all .HEIC files, which for some reason no longer opens on WIX, or Photoshop, and changing all of them to .JPG or .PNG for the sake of this Blog does not seem like a good use of my time. But I can assure you, outside of online research and references, I have pictures of cats in every position you could imagine.
​
After that little disclaimer, I'd like to introduce you to Roxy, who is the unwilling participant in this experiment, and who will, hopefully, be forever immortal on an External Hard drive after this project has finished.
​
​
REFERENCES
I had thousands of living images of my cat for reference, a select few of literal THOUSANDS have been included in this blog, which was helpful when sculpting and taking into consideration how the body/skin moved, and also for texturing. Although it did cause me some difficulty, as not many of the pictures were high quality, or that "perfect pose" that I needed for reference, so as I was working I was swapping back between tons of photos to try and get an understanding of exactly where textures were, or little details in the face/body that makes every cat unique. I also collected lots of material from online, such as anatomy pictures, muscle groups + skeletons, because underneath all the individual features, the anatomy is exactly the same. Annoyingly, a lot of my personal reference pictures were close-ups, and I didn't have as many pictures of the body or specific features, such as fur pattern as I would have liked. I've learnt that although I have a living reference to look at, and was under the assumption that I knew exactly what Roxy looked like, I found myself constantly looking for pictures and asking myself "Is that right?" "I can't remember how her nose/eyes/ ears/fur looks"
BASE MESH
1. Modelling the initial 'block' for my base mesh, was really quite easy. Not only was I following a picture of a cat directly on blender, but also through Molly's recordings, it meant I could go back and re-watch as many times as I needed to make sure I had all of the vertices in roughly the right place.
​
I did have a few difficulties at the start. This included only selecting half of the mesh so when moving into (<) this view, I had a very distorted model, with only half of the base mesh having been moved around correctly. This was quickly solved (as most things were at this point) when I realised I had missed a sentence in the recording, saying to do it in 'wireframe' view, otherwise the mesh wouldn't move as one.
2. Although tedious, modelling from the block, to what eventually became my base mesh, was relatively easy. I took Molly's advice that 'a more detailed base mesh, makes sculpting easier' as seriously as I could. Once I had got the hang of moving everything around and set my references up properly, I found that I was actually making more details, not for the later ease of it (also for the later ease) but because I was enjoying sculpting the base mesh.
​
At this stage I was worried that my base mesh looked more that of a base mesh for a dog than a cat, but skewing away from my references images didn't seems to help. It was also at this stage that I was figuring out how to model the ears and tail (this wasn't included in the recording). Following the same procedure I had with the legs, it was a pretty straightforward process. However, at this point prior to sculpting, I was worried that I had done the ears incorrectly, as the very tips of the ears were as single vertices, which looked fine, but thinking back to the limited knowledge I had of sculpting, I was assuming it was going to cause me problems.
3. After creating the main silhouette of the base mesh above, I went in further to focus on some finer details, that I thought starting to lay out now would help during the sculpting process (it did, massively).
​
I added some slight indentations to the face, to outline where the nose and eyes would roughly sit. I also thickened up the tail at the base, rather than having a long stem like above and trying to sculpt on top of it. Working on the legs, I added some weight to them, as well as attempting to start outlining the skin/muscle in the front of the thigh, that overhangs onto the belly. I also dropped the belly down slightly in-between the legs, as the first version of my base mesh looked very very skinny (yes, cats are skinny) but I also knew there is skin (and a lot of fur) there.
One thing I WISH I did at this point, and I regret massively, is not doing anything further with the feet. The feet caused me so many problems, and looking back, if I had even slightly bulked them out on my base mesh, it would have saved me a lot of hassle and work when sculpting.
SCULPTING
Going back to what I was saying about the feet, this is a perfect example of why I wish I had modelled the feet slightly in the base mesh. You can see straight away that there is a lot of solid body in the entire of my beginning sculpt, and the feet are so solid and skinny, compared to the rest of the body. Where I had bulked up the rest of the body, It meant I was doing a lot more removing and cutting into the mesh, rather than building and adding detail, which is what I had to do with the feet.
​
Starting the sculpt, was a really enjoyable process, but I struggled to look through it being a 'building' process, so I was focusing on very small details, rather than the entire body at once, meaning my proportions and grouping was slightly off. Luckily, I did realise I needed to steer away from this.
​
One thing that was in my mind looking at the model at this stage, was that the head was too small (or body too big however you'd like to look at it) Luckily, as I sculpted, I managed to bring the head size up a bit, as the body shrunk slightly, so they evened out.
I think out of the entire sculpting process, I spent 75% of the time just trying to sculpt the face. Even with what I thought was a very solid base mesh, I still kept having to go back to the face and move things around, I also struggled because although I had some very good facial references. Fur and skin distorts the look of the face, and trying to find a good middle ground was a struggle.
​
The ears were way to big, and pointy, at this stage, I think this was also throwing me off when trying to model the rest of the face. If I was aiming to sculpt dog ears.... I'd say 10/10, but I wasn't. Luckily you'll see that I did fix this. I was also worried about sub-dividing my model too much, so the majority of my first struggles came with trying to sculpt on a model with not enough vertices, resulting in this very pixelated build.
​
I was too involved in multiple different reference images at this point, although looking at multiple images did help in the long run, I was attempting to combine multiple different face shapes at once, while also taking into consideration the fur on the face of multiple different cats, and it was creating a very cursed model.
This is the exact stage in the sculpting process that I was getting really worried. Separating the head, legs, body etc from each other, it looked okay. But as a whole it looked absolutely awful. It looked like I had modelled a dog that wanted to be a cat. The head was still too small, I still was struggling with the face, and overall it looked very weirdly lumpy.
​
At this point I was starting to get stuck in my head a little bit, I had my reference images, and although they were helping, my cat didn't look like the reference images (let alone looked like Roxy). I decided to show my model to some people in class (including Molly) to get some advice. Even though a lot of the feedback I got was very obvious criticism, such as the nose being too long, neck too skinny, bigger ribcage, it was all things my brain had completely skipped over.
​
I still hadn't started modelling the feet at this point, that was because I had realised what I had done to myself by not starting them in the base mesh, so I was heavily avoiding starting, as I knew how awful they were going to be to try and sculpt from literally nothing.
These 3 pictures kind of show the transformation my cat went through, from the advice from others, it look it from an animal that resembled a cat, to a better, but equally awful uncanny valley looking cat. It looks so much more like a cat, but the face was still slightly off, it looked human almost and it needed just a little bit more work to be able to call it a cat.
​
I took all the advice given and executed it. I made the neck much thicker. This did leave me with some confusion though, as from the top view, the neck looked very skinny, and from the bottom looked too thick. But after finding some pictures of cat, I realised that to a point this is just how cats look, it is just hidden by the thick fur that is on their necks.
​
I also thickened up the shoulders, as although cats are very slender, I was putting the weight in wrong places. This all came down to confusion on my part, having to separate a cat from their fur, but also think about the properties of the fur that I wouldn't be able to fully recreate in a particle system.
​
Although I still wasn't happy with the model above, moving the ears down and out made such a huge difference to the cats facial structure, and once I had successfully moved them, I dont know how I even looked at the first image and thought it was in any way correct.
Still no signs of feet :(
Due to the uncanny valley effect, and also internal nerves that I was running behind
(I wasn't) I asked Molly for some advice on fixing the cats face.
​
On the right you can see the help she gave me. I was previously under the impression (from my references and research) that a lot of the fur on a cats face made up the 'poof'.
​
Also due to technical difficulties on my half, I was using a sculpting tool wrong, so asking Molly in person meant that I knew how to access a tool that I didn't know about before, meaning that creating the nose was something I was struggling with for basically no reason.
​
After Molly's mini tutorial on cat faces with me, I did go back in and add some more details to the cat face, using specific references of Roxy, where (with any animal) her face differs slightly from the 'standard' cat.
This is my final cat sculpt, the progression from my starting cat to this is slightly dramatic. After Molly's help you'll notice that I went back in and added some further detail around the eyes, and to the shape of the skull, I also very slightly toned down the intensity of this cats cheeks, Roxy is a very skinny cat (she is old) so I feel like compared to the 'average' cat, her face is slightly skinnier, once fur gets added on, I felt like this could alter her face shape a bit making her look too 'poofy'. The 3/4 angle also gives a really good idea to the amount of weight I added to her neck, as well as some extra definition in the back of her skull. Although she still looks quite lumpy here, I have managed to capture a lot of details, which means that I will still keep a very detailed mesh when doing retopology.
feet?
​
Honestly, I could not have imagined how extraordinarily difficult the feet would be. This was definitely a learning point that has STUCK. I must have tried 5 or 6 times just to get a solid foot shape and every time it would just disfigure and turn into something like this (<).
​
I originally decided I was just going to have to box model them during retopology. There will be further complaints about that below. What I ended up doing, was, after realising that box modelling the feet was equally as difficult. I made a new object, and separated sculpted some very simple looking feet, and then separately did the retopology and joined it together with my model. They are some very simply modelled feet compared to the rest of my model. But i am hoping, with a bit of magic, and the help of particle systems, I will be able to make them look a little bit less simple....
​
Definitely a lesson learnt, Don't skip the feet while modelling the base mesh!
RETOPOLOGY
ATTEMPT 1
In theory, the retopology sounded really easy. Long... but easy. IT IS NOT.
​
I had SO MANY issues doing the retopology, in total, I think I restarted the retopology 3 times before finally finishing it on the fourth try. This set me back quite a bit on time, which ultimately affected my end result, but learning is learning. retopology is a vital skill I needed to learn and I knew this, so I didn't want to do a bad job and then later down the line not be able to do retopology properly. In this first attempt, I realized that I was making my retopology so small that it may as well have the same amount of vertices as my sculpt, but with this also being my first attempt at doing retopology, I was struggling to create nice joining points.
Here is a good example of me trying to cut corners (for no reason because there was no corners to cut). I had decided to join vertices together as triangles rather than squares. This was in theory a very easy fix, adding or removing the additional vertices.
​
Also I had made this version of my retopology so detailed, I had even made an indent around where the tear duct is located. This wasn't necessary at all at this stage, especially as almost all of this was going to be covered or partially covered by fur.
ATTEMPT 2
I took a weird approach to my second attempt, thinking I was going to 'hack' the system and make my life easier. I didn't, I made it much harder.
​
Instead of extruding squares and knocking off big chunks of the topology at once, I thought it would be smart idea to extrude vertices and work my way around a 'loop cut' I have made. This didn't work, as there was no symmetry or consistency in my man-made loop cuts, so all of my topology was very uneven.
​
The only advantage to my weird 'loopholes' was that I learnt pretty much exactly what I shouldn't be doing to try and make my life easier. Soo that's one plus to redoing my retopology 4 times.
ATTEMPT 3
FINAL TOPOLOGY
I thought my third attempt was going really well, until about the half way mark, when I realized my mirror modifier had disappeared all of a sudden. No problem, I just added it back again and continued on modelling. It wasn't until later on when I was trying to refine my topology when I realized what had happened. I had done a multitude of things at once. My mirror modifier had applied itself roughly 30 minutes before I had realized, meaning I had a large section of half of the model without any topology (the other section with this) so when I reapplied the mirror modifier, it fixed this large gap I had, but it also modified the entire portions of the cat that had already previously been mirrored. I didn't notice this until later on, when I had a lot of vertices on one side, but also I had started merging both of these mirror meshes to each other, at that point making it almost impossible to separate from one another easily or quickly.
After finally having some feet on my model, when it make to doing the topology, I decided to make the feet slightly smaller, as they looked extremely goofy, with how big they were on my cat.
Sculpted cat (right), and retopology cat (left)
After too many attempts, I had finally finished my retopology. I was slightly disappointed to find that I had lost so much detail. Molly had briefly mentioned making a normal map to preserve some of the detail, and although it would be covered by fur, I decided I really wanted to do this. I knew how to make a normal map already (from a personal project during summer) but I was disappointed to find out that although I seam mapped both cats the exact same way, the UV maps were completely different, meaning the baked texture from my sculpt, didn't line up with my new sculpted cat. If I get a chance before the deadline I want to research into this to see if I can figure it out, but because it isn't detrimental to my final model, I am leaving it as a work in progress (disappointingly) for the time being.
​
But. Just looking at the two different topologies above, there is a massive difference in the amount of vertices. In my original sculpt, you cant even see the majority of the vertices there's that many, they just merge into a big black smudge, whereas in my new retopology you can start to see the simpler layout.
​
I didn't do anything fancy with the seam lines, nor did I struggle with anything in particular. I followed along with Molly's tutorial and where we were told the best seam points were to unwrap our models. There were some transition points in my seam where the connection was slightly sketchy, having to join up multiple loop edges (where my retopology was slightly wonky). But I didn't have any issues with UV unwrapping this, apart from making my UV mesh a little weird looking in places.
Sculpted cat topology (right), and retopology cat (left)
< Wonky Seams
TEXTURING
After UV unwrapping my mesh, I started with a solid black base colour, as Roxy is mainly covered in black fur. I'm unsure what the line down the middle of my model is, I did briefly try to lessen it, but it doesn't seem to go away, it hasn't affected my texturing or fur yet so, unless needed I'm choosing to ignore it?
From the start I knew because I was modelling off my cat, I wasn't going to be able to completely follow the texturing process, unless I was to find multiple images of cats (that were high quality, and in the posing, lighting I needed) (spoiler: I didn't). This wasn't a problem, as I actually really enjoy texturing, and I was excited at the chance to teach myself a thing or two about creating my own textures. Also because this is a furred animal, there's a lot of room for error in my texturing, as a lot of the detail will be hidden by the fur.
My original plan for texturing, was to paint the base colours in blender, where I could paint directly onto my 3D model (this meant that all of the patches and details would be in the correct places), then transfer the file into Photoshop, and add details from there. I did partially follow this, but I also had other ideas later on that I wanted to try out, as I realised once I did make some small changes in Photoshop, it is much easier to texture paint directly onto my 3D model.
​
On the Left, that's my initial texture painting, just using black and white, that I did in blender. I wasn't too neat with it, as I knew I was going to be going over it in some sort of software, but I wanted to get Roxy's patches down, and make sure I had that exactly right, as ultimately, that's going to be the make or break for her as a 3D model.
I had an idea, to attempt the texture painting in blender, and see what happens, purely out of curiosity,, but also because I knew if I could do the texturing in Blender it would be easier. Following Molly's video, I found a well lit picture of a cat, and went about trying to texture paint.
​
In theory, I'm sure I could have got this to work (it did successfully work on her face) but because Roxy has very prominent black and white spots, it was difficult to texture her fur without creating a lot of colour difference. This is my plan anyways. Although she is a black cat, I do want some slightly colour differences in her fur, for discolouration and light distortion, but not as harsh as this has created.
I decided instead of using textures from cats I had found online, I was going to attempt to make my own textures to use. I had learnt this method for illustration in college, but decided to see if I could successfully transfer it to this use. I found another image of a well lit cat, that I liked the hair pattern on, and very importantly, the hair was flowing in a mostly singular direction. I then took this isnt photoshop, and clone stamped it into a single 2000x2000 grid image. This hopefully, would have made me my own texture stamp to use in blender. I went ahead and attempted to make a second one, for the edges where white fur meets black fur, to try and create a smoother randomised transition between the two.
​
As I mentioned briefly above, I knew these likely weren't going to be perfect, but hopefully better than the online texturing method as once I begin to add fur, if there are any harsh details, these will be covered by the distortion the fur causes.
This is my final textured model, the black and white patches look much more uniform in this final image, compared to the first test above with the online texturing.
​
I added some very light paws to the feet, I am hoping that adding fur will be able to hide the transition lines on the paws, and make it seem a bit more professional. If not I will take my UV mesh into photoshop after testing the fur, if I need to, to make it look more uniform.
​
I have plans to take this model back into photoshop/texturing, if I get time. I am happy with the texturing, and starting the fur I can see that it looks really good, but I want to add some additional details to exposed skin such as the nose and paws if I can.
This is currently the final texturing for Roxy (apart from the feet/pads, as for some reason they aren't showing up on the UV mesh). Looking closely, there is lots of lines and joining points in her fur, it doesn't look very good on the mesh, but luckily, the fur is covering a lot of this up. It creates a nice variation of black ion her fur, compared to me not texturing her at all.
​
In places you can also see where I have drawn on details, such as her legs where she has her "socks" and I did want solid white in these places. I also went back over some portions of her nose/ cheeks, where the texture was slightly fuzzy, just trying to tighten up the details slightly.
ADDING HAIR
Adding and grooming the hair consisted of a lot of back and forth. I was making lots of small changes and going back into 'object mode' to see how it had applied itself and how it looked. In this first screenshot, I hadn't yet set vertex groups, weight painted or changed the blender particle settings much yet. The hairs on the face were TOO random, and there was hair emitting from the nose. I had whiskers though!
Nothing major appears different in this screenshot. I had been playing around with the hair length on the face, following along with my reference pictures, but in turn this made the particles even more random. I had also corrected the shading on the whiskers, giving them a slightly translucent appearance.
Here, the fur starts to look much better, I had added my vertex groups for density, which made the particle system avoid putting hair on the nose, and around the very edges of the eyes, to stop hair spawning inside the eyeball.
​
I had increased the length of the hair around the eyes, to attempt to give some better shape to the eyes themselves, rather than them looking like they were popping out of the cats head.
This is the hair group I had for the face at this point. The general direction was almost all correct, I did go back and tweak bits slightly, after I realized my cat looked a little bit lumpy, where I had hairs on multiple lengths overlapping.
​
​
This was the state of my cat at this point. I didn't get any progress pictures after this one, as I was so focused on grooming the hair. I had multiple issues while doing my model, that I'm sure could have been solved with a quick google search. while grooming hair at the front of the body, if my camera was at a specific angle, I would (without realizing) also be manipulating the hair at the back of the body, this caused me to continue to go back and forth to try and amend hairs I had accidentally groomed into the wrong position, or into the body model, which meant there were bald/glitching patches on my model, that were quite hard to see, especially if I was working in textured mode.
FINAL SUBMISSION
This render isn't actually the final face of my cat. But it is the one I submitted in my project. I was having issues with my rendering (short for I left rendering too late, and didn't have time to render a close-up of the face before the deadline). The hair is still slightly untamed in this image, but it showed a very good representation of the facial hair and structure. I played around trying to create my own cat eyes, but I also didn't want to spend too long on this, as it wasn't a main aspect of my model, purely for aesthetic purposes. So they are slightly odd looking, as I didn't compete the shading process for them, as I still had more grooming to complete, but I wanted some sort of eyes on my model for rendering.
This image outlines the lack of grooming I did the best, around the front leg, back legs and the outline of the face, you'll notice there are odd clumps of hair, that don't look natural really. I had plans to tame this, but I ran out of time, and ultimately I'd rather have a cat with fur than a cat with bald patches, or fur that was merging into the skin, as I feel like that would look worse.
A good example of the variation, would be focusing on the quality of fur around the back, top of the head, chest and tail, as at the time of submission, these were fully groomed.
These two renders are by far my favorite out of all of them. I really like the angles they show, as they show nice details, as well as a very nice silhouette of the cat. I've also (purely by coincidence) managed to hide the imperfections in my work with the angles I've chosen.
​
One thing that is slightly irritating to me, Is the lack of volume in the inner legs, if I had known to bulk up the inner front legs more during modelling then I would have, but also I should have elongated the hair on the inside of the legs to create a denser hair, making them look thicker.
​
Apart from the front legs, I think my cat has very good form and texture in general, there are a few areas for improvement, but as a first attempt at modelling an animal, and a first attempt at using the particle hair system (properly) I am really happy with my outcome. I really like the texture as well that come through with the hair, although my texture painting was EXTREMELY rough, It looks really nice now that the hair has distributed it slightly.
​
I was nicely surprised by how nice the paws came out on my final model. As said multiple times in my blog, I struggled horrendously with the feet, and after adding fur, I feel like they look really good, a little bit simplistic yes, but overall look very nice.
IN CONCLUSION
Much like my reflection for the Production Arts pathway, I absolutely loved this project. I learnt a lot of new skills in Blender, and got to develop on skills that I had briefly touched on in previous units. I struggled with some aspects of the project, mostly retopology, which took a lot of time and effort, but I came out of it successfully with a lot of knowledge learnt from the mistakes I had made.
My final model has the successful likeliness of a cat, whether that be my cat or her hypothetical 'sibling' in this case. If I had been able to do this project again, I would have spent more time 'grooming' to make the hair look less patchy and clumped. I also would have gone back and added the fly away hairs over the main fur, as well as focusing further on the detail in the fur, specifically around the ears, where the hair is longer and a lot finer, as well as the joints and underbelly around the legs.
​
The two outcomes that I submitted for this unit, were both very successful, especially in the very limited amount of time we had. Of course now its submitted, I would go back and change almost everything given the opportunity, now I have the knowledge I do now.
There are definitely some elements within this project that I would research further outside of class, but I feel as though I have a good understanding of the basis of modelling animals now, as I mentioned within the evaluation for my concept art, if I had a little more time, I would have definitely managed to feel completely happy with my final submission. I would have completed the grooming on my cat, perfecting the base hair, as well as adding the fly-aways and longer hair groups where they were needed.
​
After the submission I am still continuing to work on my cat model, when I get the chance, as I am really determined to get this 3D model to a stage that I am happy with. Things that I am currently working on include, fixing the eyes to look more realistic, grooming the hair to completion so it doesn't look as rough, retexturing the nose so it isn't as blurry, adjusting my vertex groups to allow for less hair distribution in some areas, rather than a point blank NO hair or ALL hair, in places like the bridge of the nose, at the edges of the eyes, around the mouth, and by the feet.
MY RESEARCH
Weight Painting: The children hairs in my particle system were not following my original vertex groups, and also the weight painting, I did some research and figured out along side BOTH of these setting telling Blender where to place children hairs and where not to, there is also a third setting (probably more)… I found this thread that helped leave the nose hairless
https://blenderartists.org/t/weight-painting-and-particle-generation/429221
​
Whiskers: I wanted the whiskers to have their own texture, not follow the texturing of the cat, as most cats tend to have either solidly white or black whiskers, or somewhere in between, not half and half (at least Roxy just has white whiskers) and so I needed to find out how to separately colour her whiskers
POST SUBMISSION
So far post submission, I've managed to smooth out the majority of the main body hair. I still have some work to do on the feet however. I have also re-worked a lot of the hair on the face, it looked okay from the front but from the side in some of my close ups it was very clumpy. I took further detail in my weight painting, correcting areas around the nose and eyes that were emitting hair when they shouldn't have been. As well as changing the eye texture to something slightly more appropriate, and fixing the texture on the nose to not be as blurry.
​
I have also began to add the flyaway hairs in my model, which is helping to add variation in my hair particles that I was trying to achieve with one particle system.
TESTING RIGGING
