Blender 2.8 Part 3: Self Learning though Personal Projects

I had just wrapped up some mini tutorials in Blender 2.8 to get a feel for the software. Modeling was done mostly with shortcuts, navigation relied on the Middle Mouse Button, and some aspects of the Beta were still a bit work in progress. Throughout all of these changes from Maya, I felt as if everything was coming together mentally and I was ready to start working on an honest to goodness game asset.

Stopping

The majority of my learning experience was taking place at my then employer. Due to a number of factors (business cycle, new corporate governance, new strategies, etc) the creative department had a slowdown in work that was coming our way. I had accomplished what I had wanted to (or was able to) as far as pipeline work was concerned and I needed to find a way to not waste away my time while I was in the office and doing my Blender work there was the logical choice.

The main driver behind learning at work was, of course, the commute I had. I have a few commute articles over at my personal blog, some of them a bit ranty, but the short version was that I had three to four hours of total commute time A DAY to and from work. Doing work on the train was difficult (I tried) and learning a new software would have been impossible without racking up an expensive cell phone bill. I chose to reserve my weekends for relaxation and recovery, so I rarely did much work at all in any capacity.

The downside of course is that when work comes in, work always takes precedence (and it should). It was about the time I was wrapping up my ship tutorial that we started getting work again and thus, I wouldn’t really touch Blender for about a month.

Starting: The Ember Mug

During the wave of work I got at this time, there was a day or two where I had an empty task queue. On one of those days, I opted to get the latest Beta and do a super small and super easy standalone asset.

This is the reference I used.

That’s right, I grabbed a render of the Ember Mug and decided that I was going to make that.

This project was a short, 2 or so day exercise to get me back into Blender. I had some muscle memory to re-expose and some new aspects of the software I wanted to explore as it related to Eevee, Blender’s real time renderer. I didn’t really have a grand plan or a goal to fully practice my usual workflow.

My final render from Eevee.

In the end, this project was instrumental in furthering my knowledge and desire to keep using Blender.

• Modeling went quick, partly due to the subject matter and partly due to the tool itself.
• Blender’s real time renderer outputs solid images
• Working with materials was easy and, while the interface is a touch odd, very similar to both Maya and Unreal in flexibility.
• I used Blender’s lighting to a limited degree and enjoyed it, although setting up IBL’s to work properly is really stupid and needs a better implementation, like though a Dome or Sphere Light.

I wrapped this up in two or so days and went back to work until the next slowdown occurred.

Game Dev 101: The Barrel

Technically, this is an Oil Drum.

My ref, by Uline.

When the next slow down at work occurred, it lasted on and off for a few weeks. I felt this was a good time to dive into using my traditional asset pipeline in Blender. I didn’t want any complex or difficult assets yet (and to a degree, still don’t) but I did want something that would serve as good practice. Additionally, I had been modeling a bit more in detail (compared to relying mostly on normal maps) at work and I wanted to carry that though to my personal work. When I factored all of these elements together, I decided that a simple oil drum would be my next project.

My personal workflow will vary depending on what I’m doing, but it generally looks like this today:

• I decide on a project and gather reference in PureRef
• I build out the basic shapes and will either take them to Zbrush or will continue to model into a ‘mid poly’. I think of ‘Mid Poly’ models as assets that are easy to tune and add details to, becoming your High Poly. Ideally, you can edit them down to being the final LOD0 Low Poly model. That does not always work out however.
• If I’m using Zbrush, I may start the project in Zbrush (organic assets) or use Zbrush as a finishing touch (hard surface or stylized assets). Zbrush is very fluid in the modeling stage of my pipeline and it can be used anywhere at any time before I’m baking mesh maps.
• I finish the High Poly model.
• I’ll create my Low Poly from the Mid Poly or make a brand new model (as needed) and UV it.
• I export both models to Marmoset Toolbag 3 for texture bakes. It is at this stage where I may go back to my modeling tool and add geometry if my bakes are a little off, or adjust my UV’s if I have overlapping bake information.
• Open Substance Painter and import my baked texture maps to use as Mesh Maps.
• Texture time!
• Finally I export everything and test in either my modeling tool or my game engine/renderer (Unreal/Marmoset).

This was more or less my process for the Oil Drum. The only major difference was that it was all done in Blender.

High Poly

Low Poly

UV’s, which could probably have been done a bit smarter.

One of the major struggles I had with this project was the UV process in Blender. This project was done with one of the early Release Candidates, meaning the UV tools were mostly in place. I found unwrapping pretty intuitive (mark a seam, tell Blender to unwrap) but outside of that, it was difficult to really work with the UV’s in a consistent, friendly way. It took a lot of time for me to work it out and I often sent an FBX over to Maya to confirm everything was working as intended.

My biggest take away from this project was discovering what I needed to add to Blender for it to truly replace Maya for me: UV tools. I knew that I’d need more than the included Magic UV Plugin and Texel Density Checker. The second takeaway was that Blender is absolutely capable of outputting Marmoset quality renders from Eevee.

Textures done in Substance Painter. This is rendered in Marmoset.

This is from Eevee, Blender’s Real Time Viewport.

Eevee looks better doesn’t it? I semi-lazily tried to match Marmoset to Eevee and wasn’t able to, although part of that relates to camera tone mapping. Now take a look at what Cycles can do.

Cycles is Blender’s path traced renderer.

With all of these learning projects completed, it was becoming time to take my work with Blender into Prime Time. I had not yet completed a real portfolio project with the tool. During my long commutes I occasionally would dream up small ‘mini projects’ that would be doable either on my commute or at work during slow periods. Some of them were complex and overlong while others didn’t really inspire passion in me by the time I was able to start them.

One project always held my attention. It’s based on my daily routine and it’s something I’ve done since I was a teen.

I pick things up, and I put them down. In the conclusion of this series, I detail my most recent portfolio project: The Platform.