As a CG artist, there are established ways of doing most things – ways which gives you reliable results every time and work great in a team. On the other hand, every CG aritst has a bag of hacks which if exposed, would make them go straight to CG prison. In this article, we’ll cover some of our favorite hacks which can save you a lot of time if used at the right time.
#1 – Render at Lower Resolution and Scale Up
Rendering can take an awfully long time and in some cases you can speed it up significantly by simply rendering at a lower resolution and scaling it up. The reason this can work well is due to how sharp CG renders are straight out of the engine. If your goal is photo-realism, you’ll have to add a subtle blur to your renders, otherwise it will seem far too artificial. In a lot of cases, particularly if there’s animation involved, there’s literally no way to tell that you rendered at a lower resolution and scaled it up – particularly if the end result is for video going on YouTube where it will get a lot of extra compression.
Left: Rendered at 1920×1080 – 8:40 min
Right: Rendered at 1400×788 – 4:47 min
While there are quality differences, (hence this being a hack) the difference in render time is significant.
Restrictions: There is a quality loss and for the highest quality you want to render out at the same resolution as what it will be viewed at.
#2 – Fix it in Post
Getting exactly what you want in 3D can be really tricky, and while I recommend that you should take your 3D render as far as possible, sometimes fixing things in post will make your life a lot easier. The advantage of doing post work is that you’re controlling the image directly. Instead of changing the light and hope you get the desired SSS, you’re straightup painting in the SSS where you want it. Can’t get the highlight in the eyes to be exactly where you want it to be? Painting time!
I had a really hard time in the render to get the eyes to look right in terms of lighting and reflections. Instead of spending days in 3D trying to get it to look right, I took the render into Photoshop and painted on top of it.
Restrictions: This is an incredibly dirty approach and really only works for still images. If you’re updating your render scene (new lighting, model updates, camera angle), you’ll have to redo all your overpainting.
#3 – Render Decimated Models Instead of Using Displacement Maps
The ‘correct’ way of rendering a model from ZBrush is to make sure the model has proper topology, UVs and multiple subdivision levels and then using displacement maps to bring out all the nice details. This takes a lot of time and doesn’t at all allow for quick changes. Introducing: Decimating your models! This hack involves simply reducing the polycount of your model while preserving the details using Decimation Master in ZBrush – which allows you to render the raw model instead of worrying about all the technical details.
The model after decimation.
This works really well for still images, background props (rocks, trees) and anything else where you need quick renders. Decimation Master in ZBrush also allows you to preserve your UVs, meaning you can still plug in texture maps like albedo should you have those.
Decimated rendered model on the left – Original ZBrush sculpt on the right. There would be no visual difference between rendering this model with a displacement map compared to decimating it.
Restrictions: Makes the scene a lot heavier than using a displacement map as you need to load in the actual geometry instead of a low res mesh being displaced by a texture map. It doesn’t work at all if you’re dealing with deformation.
#4 – Blend Different Models in ZBrush with Clay Brushes
If you have multiple models in ZBrush and you want to blend them together, here’s a really handy hack for you! If you use the clay brushes in ZBrush, you can simply sculpt over the different models (in the same subtool) and they will start to blend together. This doesn’t actually merge them, like dynamesh would, but rather visually blends them.
Below we’ve blended the beard with the head model using the clay buildup brush. The models have to be in the same subtool.
The topology is messy, but visually it works quite well. I do this a fair bit when it comes to concepting where the shape is what matters – not topology.
Restrictions: The end result is messy and not very flexible. It’s often better to merge everything, Zremesh it all, reproject and then sculpt over the intersections.
#5 – Use Scans as a Starting Point
There’s an ever-growing library of amazing scans online, such as ThreeDScans that offers high quality and free scans of models scanned in museums. Instead of building your scenes from scratch, you can use scans as a starting point to quickly block out your scenes and models. If you want more realistic scans, 3dScanStore offers a lot of scans of real humans. Once you start using scans as a base, you’ll never want to sculpt or model anything from scratch again, as it makes it significantly faster and your starting point is much stronger. I personally find that using scans of humans as a starting point for creatures is incredibly helpful.
Using a scan is different than using a base mesh in the sense that a scan is highly defined while a base mesh is generic in nature.
Full scene made using only scans from MyMiniFactory
Restrictions: You’re heavily restricted by what scans you can find. You want to be careful with using scans too early in your learning, as you can easily become too dependant on them. If you can only concept sculpt using scans, you’ll be severely limited the moment you don’t have a scan available for your subject.