Creating Film Resolution Assets for Unreal Engine
Taught by Liam Whitehouse
- Course Number:
- Software Version:
- Original Run Date:
- January 2022
- 3 hours 40 minutes
Liam is a New Zealand based Senior VFX Generalist, who has been in the 3D industry since 2003. He has worked on various projects over the past 18 years including most recently the photoreal canyon environments for The Falcon and the Winter Soldier on Disney Plus, and the hollow earth environments for Godzilla vs Kong, as well as VFX work on other upcoming high profile feature films.
Class 1: OverviewOverview of photogrammetry of complex real-world props and the workflow to get the assets into Unreal Engine 5 as a Nanite high resolution textured asset for Virtual Production or NextGen Games.
Class 2: PhotographyHow to photograph a complex prop asset to capture the entire surface in a very accurate 3d model and solve and also very accurate and sharp textures covering the whole model 360 degrees including occluded parts of the model like the inside of nylon loops and pockets blocked from view, images are provided and students can use the same techniques to photograph/capture any model using iPhone/android smartphones or any DSLR.
Class 3: Reality Capture, Part 1We will take our sharp photos from the capture session and load them into Reality Capture which is a smart 3D photogrammetry software that works by building models with textures from multiple photographs from different angles. In Reality Capture, we will Align the photos into a 3D solve model, then mesh and uv the model before using the smart re-mesh downscaling system to make a 5 million polygon “Nanite” ready with 8k textures. We will also export the model for optional re-topology in Zbrush and Maya for upgraded sub-d quad geo, UV’s and for creating animated character models.
Class 4: Reality Capture, Part 2A look at how to modify and clean a Reality Capture Scan mesh using the sculpting, z-remesher, polygroups, smoothing brush and Dynamesh tools. We will remove part of the scanned geometry so that a 3D character can put their arms inside the vest and take a look at re-topology to make a quad lower res mesh automatically in Zbrush without having to hand-build re-topo geo onto the model as well as methods for fixing holes and repairing geo.
Class 5: RetopologyA look at hand retopology in Maya to create a standard pipeline friendly 3D quad subD Asset that would be used by texture artists, riggers, animators and other DCC software for hero player/foreground/closeup characters. This is useful for hero models which need to be animated or deformed or have cloth or soft body simulations run on them as the Zbrush high-resolution static mesh geometry is too heavy to use for animation and fx. This model can have new UV’s and then be subdivided in Maya and re-imported into Reality Capture to re-project the textures from the master model onto the re-topo model.
Class 6: Unreal EngineSetting up the Unreal Engine 5 Project to look at the finished Nanite hero assets we have created. We will take a look at the difference between a 21 Million polygon model, a 5 Million Polygon Model and a 150k Polygon model to see the differences in visual detail using UE5 Nanite and also the disk size requirements for each resolution of asset. We will also connect the texture maps and import a test character from the Epic Games Marketplace to test our prop vest onto a real character and test the stunning real-time performance and photoreal quality of lighting and geometry handling of UE5 and Nanite geometry streaming engine.