Experimenting with Houdini to Groom a Dog in a Ghost Costume

Introduction

Hi everyone, I'm Maxime Schyns, a 3D Artist from Belgium. For the past 10 years or so, I've worked on various projects as a Character/Creature Artist, mostly animated movies and TV shows, but also some VFX works here and there.

Characters and creatures have been a passion for as long as I can remember. When I was a kid, I spent a lot of time drawing them, mostly inspired by graphic novels. I also grew up binge-watching VHS/DVD of old Disney and stop motion movies, like many people growing up in the 90s/early 20s.

Although at that point, I had no idea this could become my job in the future. Many years later, I was pretty much set on going into something completely different after graduating high school, but some life events got in the way, and I decided last minute to go into Computer Graphic Design.

I struggled with that choice for some time, and my first years were, quite frankly, rather chaotic, and I failed classes multiple times. It took a visit to the FMX festival in Stuttgart at the time and a lot of side experimentations to realize that this could be my job.  

So I picked it up from there to finish my studies and finally jumped into the 3D industry. Fast forward 10 years later and here I am, making a spooky dog for Halloween.

Trick or Treat Workflow

The idea for this project came rather simply from pictures I saw on Instagram. At that point, I was experimenting in Houdini and looking for a small project to keep testing the grooming workflow.

And on top of that, I try each year to do a quick Halloween project. It's a fun theme, and it also gives me a deadline because if not, my personal projects tend to go on and on until I get bored. I might have 20 or more unfinished characters and creatures on my computer. Halloween is then quite perfect to avoid filling up that drive.

For this project, as I wanted my focus to be mostly on Houdini, I started the modeling with a basemesh inside ZBrush. I also knew that the body would not be seen in the final images, so I only did some sculpting tests on the face and quickly moved on to Marvelous Designer for the cloth.

Here, the process is incredibly simple. I started with just a circle, dropped it on the model, and then cut out holes for the ears, eyes, and nose. Just like you would do in real life.

Back to ZBrush, where I did the retopology and detailed the models a bit. Since this project was not going to be animated, I didn't bother doing a clean retopo and went for a controlled ZRemesh. It's way faster in this case, but definitely not something I would do in production.

Texturing and Grooming

Once everything is done inside ZBrush, I exported the low-poly to Maya, did some UVs, modeled the basket, and then sent everything to Substance 3D Painter for the texturing. As I said before, since my focus was not on these parts, I went quick and dirty.

Over the years, I've created a bunch of smart materials that work for me and that I can re-adjust depending on my needs. I used them mostly for the clothes and the basket in this case. The dog itself was a mix of hand-painting and masks to add some more details.

The entire thing was going to be covered in fur, so I mostly focused on putting the right colors and not detailing it too much, knowing that I would more than likely have to do some back and forth in lookdev.

While I was working on the texturing, I also started the process in Houdini. I am quite new to that software, so my process isn't exactly perfect at this point. But the way it works makes it so much fun to experiment, and the documentation on the SideFX website is super helpful!

The groom of the fur is quite simple. I split it into different parts to have more control for specific areas like the ears, the tail, the whiskers, and a few extras for the cloth.

Then I create guides for each part. There are different approaches to this process, and personally, I find it much more enjoyable to sculpt it directly. Since the groom itself isn't very detailed and difficult, I decided to restrict the shaping to the broader details of the fur only, keeping the flow as simple as possible. That way I have a bit more freedom in the hairgen process.

Once the guides are done, it's a combination of masks and a handful of nodes to get the desired results. One of the most enjoyable things about grooming in Houdini is the ability to paint masks directly in the software, which allows you to do pretty much anything you want, really.

Whenever I work on a groom, I tend to work the same way as sculpting or texturing. First, working on the larger effects and slowly building up the details. Overall, this groom didn't require anything difficult, so the graph ended up quite small.

Now that I am happy with the groom overall, time to move on to the last part, the lookdev process. Once again, since I aimed to experiment, I decided to go for another test inside Solaris. My process is still messy and unoptimized at this point, so I won't go too much into details on the technical part.

But from a more artistic point of view, whenever I render a personal project, I like to do a bunch of tests with different lighting. For that, I picked a few HDRI I liked from my library and had a few area lights here and there if I needed to accentuate something, soften some shadows, etc.

It is such a fun part because it allows you to see how your character reacts in different lighting scenarios. It guides you toward some fixes if needed, and more importantly, it finally shows the project as a whole.

I rarely do any post-production anymore, so usually when I am happy with the overall look of it, I render a few turns, a wide view, a close-up, and voila, the project is done!

Conclusion

As I mentioned before, this project was most and foremost about experimenting with Houdini. It was quite a jump at first because of the way everything is structured, but doing small projects and small steps really helped me understand more and more about it.

And in the end, the groom process itself keeps the same principles whether you use Houdini, Yeti, XGen, or something else. Once you know more or less what the tools do, you can pick things up quite quickly.

The biggest advice I would give is to keep it simple first and not be afraid to experiment. That's, at least to me, the best way to learn anything and avoid falling into the tutorial hell or being stuck with personal projects that are way too big to finish. I've done my fair share of that, so I can say I know from experience.

Maxime Schyns, 3D Artist

Interview conducted by Gloria Levine