3D visualisations – Vilma Saarela

Vetehinen

The tale so far can be found here!

Semester two ended with me finally deciding on the character design of Vetehinen. She was to become half human, half fish and sit in a lake ready to drown her victims who dared to come too close. I was super happy with her design, considering how much I dislike character design in general, so I was more than happy to finally move on to modelling.

So ZBrush called, once again.

The modelling process

I started the modelling process as any other time, from sphere to face, from blocks to body and so on. However, for the first time ever I used my own drawings as guidance, which was quite exciting. So far I had only used other designs or actual photos as reference, but not drawings I had done myself.

This resulted in some problems in the sculpting process, as I noticed that as pretty as the drawing was, it wasn’t exactly the best possible reference to have. The side view and front view didn’t match 100%, as what often seems to happen when trying to convert the 2D drawing into 3D space. The issue wasn’t anything game-breaking of course, so I was able to just change the sculpt as I went to look plausible in 3D.

I had to reference the ears with another image, as my 2D drawing wasn’t working in 3D sculpt. I had planned on having the ears quite cartoony, but as the time went on I realised that I kept steering towards realism more and more. She was still going to be a little bit stylised, but retaining most of realism as possible.

The upper torso was quite easy to model, as by now I’ve already managed to gather mental reference of the anatomy in my head. It was actually the tail and spikes that ended up taking the most time. At this point of time I had planned to only use this sculpt as the 3D print and create other characters, so I didn’t pay too much attention on small details like the inside of her mouth.

This changed, however, after one Tuesday meeting and some concerns about the time and the limit of it. In the end, I ditched the other characters and decided to focus only on Vetehinen, which of course meant I’d need to somewhat go through the animation pipeline of making animated characters.

Preparing the model for visualisations

From the sculpt done in ZBrush I needed to bring it to Blender for retopo, unwrap and rigging. Since the model also had to work in digital form, as well as in print I had to go through the whole painful process. Before I could hop into that step however, I had to open up the sculpt’s mouth and at the same time I also addressed some of the sculpting issues the original sculpt had. I really didn’t like the face she had before and took a really long time trying to figure out what exactly was it that was so off-putting about it. I made some changes and in the end was way happier on the sculpt. The next step was retopo.

At this point I had completely smoked Maya out of my pipeline so I really wanted to do everything in Blender and learn to use these sides of the software as well. I had already bought the retopoflow for Blender way before, so I finally got to use it for character modelling too. This plugin has a very neat feature that lets you just click and drag rows of polygons so the process on the tail for example was super speedy. Of course the finer details are still fiddly and time-consuming, but overall the process is a lot enjoyable than in Maya that is completely manual.

The next question was how I would project the high detailed mesh to the low poly version. Normally I would do this in ZBrush and use a normal map, but since I’ve been having so poor results so far with the Going Live project I wanted to try something else. I found a video about the exact pipeline and was keen to try it.

The trick was to use two different modifiers; subdivision surface and multi resolution. With these I was able to use the low res version of the model in the viewport and save my laptop from burning, but have all the fine details of the high resolution sculpt in render. It wasn’t exactly the most ideal approach with rigging, but since I will only use that for posing it was okay.

I then unwrapped the model and started to rig using the notes I had from last semester, as well as watching through some more youtube videos about the process to remind myself what exactly I was supposed to do.

Blender rigging is pretty straightforward and easy, just adding one bone after another and naming them so that you can easily mirror the rig to the other side. I pretty much only needed the possibility to pose the character in simple way, move the eyes and add some emotion on the face. As I had decided not to point the camera straight on her face for a close-up I could just use shape keys (blend shapes) for the face and ditch any detailed facial rigs.

I was able to follow the videos and my notes pretty easily and in the end I had a proper working rig with somewhat wonky parts. I had a few meltdowns during weight painting, as I kept accidentally painting nearly invisible points to the other side of the model and then racking my brain trying to understand which bone exactly was the one that was affecting the wrong side. I did take the rig to Duncan to see if there was something to be done for the overall wonkiness, but since I decided to use the multires modifier to project the details I was actually weight painting on the lowest resolution, leading to some sharp corners on the model when I bent it. Instead of doing everything again, I just decided to leave it as it was, pose the model and then use Blender sculpting to polish it. Since the model was not going to move, that was perfectly fine.

After all of that, I posed the character first for the 3D print, sent it to ZBrush for further work (as seen here) and then saved the rig file several times for each of the 3D visualisations. This is due to the lack of referencing like Maya has.

Texturing

After I was done with the rig and uv’s, I was finally able to move on to texturing. Funnily enough, a step I used to dislike has actually turned to be one of my favourite things. These days I really enjoy being able to launch Substance Painter and start hand painting the textures on, instead just using nodes in Blender.

For Vetehinen’s textures, I really wanted to focus on the hand paintery effect so I made sure to reference a painting of the romantic era whilst I worked. Portraits of that time usually had quite strong shadows in the eye sockets and at the sides of the nose, creating almost sculpture like shape of the face. The skin was smooth with redness all around. I made sure to add these details to my version too.

Another reference I added was for the tail. It was quite funny drawing away with a fish right in front of my face for hours straight. But with this trick I was able to colour match near perfectly. It was especially satisfying adding the red fins contrasting all the other colours. The only pre-made material I used from SP was a fish scale material, so that I was able to add height to the model.

From there I exported the materials to Blender and kept going back and forward between the softwares as I made tweaks according how the textures looked with the render. As I had to use Eevee because of the paint filter, the priority was for the textures to work with it. At this point I also added particle hair to Vetehinen’s eyebrows and eyelashes and added a node curve hair. In the end, I only used the node hair for the ‘Drowning’, and for the remaining two I opted for Blender’s hair particles and dynamic simulation. The difference in the hairs can be seen in the Eevee screenshots and the final turnaround that was done with Cycles.

I'm drowning

First of the set was the “drowning”. As I was so pressed with time I started to plan about simply not showing the bottom of the rig so that I wouldn’t have to make her any legs. By changing the camera angle to a frog’s perspective I would be able to crop the legs entirely. This was of course a good quick fix, but it just didn’t work in terms of storytelling. After some brainstorming at the group tutorial we came up with the idea of keeping the original 2D layout, but giving the water more space.

I would need legs for this bit of course, but instead of going through all the effort, I’d just do a little surgery for ZBrush premade model, frankenstein it to my model and then simply show a silhouette of the model for extra drama. I could add some surface light reflections for some detail, but overall it would be an easy feat.

Kalevala paintings from the Akseli Kallen-Gallela exhibition in Helsinki, 2022.

It was indeed as easy as that, I took the legs from one of the premade models in ZBrush, cut them off and exported them to Blender. Over there it was easy to line them up with my rig after I had posed it the way I wanted. I then just cut off the tail and only left the torso in place. I didn’t even need to join the two as the model would be only seen as a silhouette anyway.

There was some talk about adding a dress to the character, and at first I even thought about adding one until decided otherwise. There was the question as to why she would be swimming naked in the first place, but the weird truth is that Finnish people, for some reason, tend to swim naked even in modern day and age. And especially in the past, as can be seen in the paintings above. Finnish women were almost exclusively portrayed in their natural state in these paintings, so with these facts in mind the choice ended up being the same for my work too.

I wanted to have some movement in her hair, so since I was using the node hair I was able to just animate the normal node to make the hair wavy. In addition of the hair, there would also be an ocean modifier on top of her, so that I could have movement in the water as well. The scene kept looking super dull, so I even tried adding some volumetric fog as well as a workaround for water acoustics using alpha texture on the spotlight.

In the end I added the paint filter and all hell broke loose. Since I didn’t have a proper background the filter had nothing to latch on to and the end result was a green mess. I had to add a plane and quickly add some colour variation in Substance Painter.

The plane did it’s magic and the filter worked again. And it was painfully slow. The filter was super heavy to use, and even with a scene with barely nothing in it, everything took a long time to do and the render itself was around 5 minutes per frame. I left my laptop to render overnight and in the morning imported the image sequence to After Effects.

At this point it’s safe to say I’m really not comfortable with this software at all. I’m skilled at video editing, but I’ve done my editing in Final Cut Pro, as the footage I usually edit is actual video footage from shoots, not animated stuff. So it took me a good while to stumble about finding the correct buttons and effects for the purposes I needed. I was able to turn the character only sequence into a silhouette and added some rim lights to be shone from above. The background water was a little too quick as it changed on every frame, so I slowed it down quite a lot and added some motion blur to try to make the effect less eye-hurting.

After adding a super simple png spotlight I exported the whole sequence from AE and brought it to Final Cut Pro for colour correction and other small details. And finally I was done.

The final result is quite good. I added some sound clips from the YleArkisto’s freesound library to further sell the scene. I think this piece is the weakest of the three, but still I do quite like it.

Foliage

Before I could move on to the next two pieces, I had to create a small asset library for foliage. Blender has a very neat tool called Sapling. With it it’s super easy creating trees and bushes, as there are quite many presets to choose from and an endless amount of customisation. I quickly created two different pine trees, a spruce, a birch and a small bush. These are all the types of trees you can find in Finland. I used NatureGate once more for reference.

After the trees, I went ahead and created a tall and peculiar-looking plant we have in Finland, usually close to areas where there’s water. For smaller foliage, I used the nature pack that I had created after CGGeek’s tutorial a few years ago. The process of making those assets was so time-consuming, that I’m happy to keep reusing them when ever I can.

After completing the set I just saved them into their own collections so that I was able to append the collections to my scenes later on.

Lonely

The second piece was the “loneliness”. The layout design on this took a step forward and then back again. After seeing the 3D layout we all agreed that the original plan of giving more space for the story to unfold around the character was more suitable so that’s what I ended up going with too.

I also loved the photo Sean sent me of the unknown mermaid by Donato Arts. I wanted to bring in the waterline for this piece too, so that all three had the element in them as a combining factor.

The set building was very straightforward, I just used the original thumbnail as a guide as I brought in the elements from my foliage asset collections. Thanks to this asset collection the two remaining scenes were very quick to set up. I added a spotlight to shine on the character, just like the original painting had, to frame the subject and lead the eye towards it. There’s also something peculiar of how the first thought of a spotlight leads to the feeling of loneliness.

I used the pose studies I had taken and drawn earlier this semester and sat the character down in her designated spot.

Since the rig wasn’t 100% functional in terms of the weight painting, I had to go in with the sculpt tools after I nailed down the pose. This was of course not a big deal, I really only needed to use smoothing and nudging to get rid of weird dents and wrinkles. At this point I also ditched the node hair and opted to use hair dynamics instead.

Once I was happy with how the hair was laying down, I turned the particles to mesh and added a hair texture. I took a moment to look at the model after this and was very happy with the pose and her facial expression. Especially her left hand and how she had it delicately placed on top of her tail was a nice detail I wouldn’t had thought of without the life shoot session with the model.

I once again added the paint filter on top, animated some movement to the water and foliage and rendered the sequence. I had one sequence for the background and one single render without the render for the character. I ended up using AKVIS oil painting filter for the character renders, as the oil paint filter in Blender kept producing too filtered end results. The filter had a 30 day trial so I was able to use it for both of my remaining works.

I then brought all the pieces to After Effects and in there did some masking and layering as I wanted to slow down the background foliage a little compared to the water. I again did the rest of the colour correction and effects in Final Cut Pro, being more comfortable with it.

The final render turned out very pretty in my opinion, with the character lit by the spotlight as the rest of the scene is being left in the soft shadow. The paint filter effect on the rocks is especially nice and contrasts well with the sharp detail of her face. I purposefully left her tail to be more painterly and face sharper, so that her facial features would be easier to read.

Anger

The last but not least piece was the “Anger”. This piece kept the same plan pretty much from the beginning from the moment I saw the oil painting about the night sky by the lake. I wanted to add some spook factor by replacing her eyes with simple white emissive material. This was something I had planned to use on my Ajattara piece, so I was happy to recycle the idea to here.

I wanted the scene to be creepy, but not too much. It still needed to be artistic and easy to read, so instead of going full on darkness I decided to let some light in.

Again the scene building was a piece of cake thanks to the foliage asset collection I made earlier. I was able to quickly plop down the ground and the lake. This time I didn’t want to use the ocean modifier as it creates too big waves to fit into a small lake like mine. Instead I opted to follow after a tutorial by Gokumon. Instead of simulating water all I needed to do was animate some texture settings to make it appear as if the surface of the water was moving. This was such a brilliant trick that saved so much time in rendering!

I added the moon and underneath it a sun lamp, so that I was able to frame the character with light. I then went ahead and turned the world light down to make the scene very dark and eerie, but not too dark so that it would still be easy to see the details. Same as in the piece before, I locked down the pose I wanted and sculped the details in.

I then used hair dynamics again to make her hair fall down and even in front of her face. We went full on Ring during our photoshoot so I wanted to emulate the same feeling here. With the glowing eyes behind the hair strands the fear factor was definitely there. It was fun zooming in to look at the model from the sides too, she really did appear if she would’ve been dragging herself out of the lake to sneak attack like a crocodile.

I once again animated the foliage, rendered the sequence separately with the character and added the AKVIS oil paint filter in Photoshop before bringing them all to After Effects. I quickly exported the scene to Final Cut Pro and played with overlays, lights and audio. I was actually not liking the result at all whilst in After Effects, so I’m glad I remembered my old friend. In there I was able to add so much more to the end result and in the end really liked how the piece turned out.

And there they were, the final three. Even though I lost so much time during this semester due to stressful news from home, I still managed to finish them all, which I’m very proud of. I do feel that if I were to be in this stage a few weeks back I would most likely find things to fix and better and the end results would’ve been even better, but as of now I’m very happy.

I’m hoping to get some valuable feedback during assessment and make changes if needed so that they can all shine individually and as a story at the Masters Show.