Hello everyone!
I’m a beginner with Unreal, and have general knowledge with 3D modeling (have done it 5-10 years ago on a very simple level) and some better knowledge with the process of photogrammetry. I want to create and sell props in Unreal Marketplace. I would like to create models both for games and for movies/tv (meaning that models can be used in both use cases), and I hope the process is fairly similar, apart from maybe the final poly count. I mainly want to scan household items, appliances, tools, backyard items etc. In the future I might want to start scanning bigger things, but for now I’ll keep it small and relatively simple.
I have questions about the workflow from start to finish. I mostly have the reconstruction process down, but I lack knowledge about the process after exporting the model from RealityCapture. I know that’s where the domain of this sub ends, but again, I hope to have some luck by posting here.
Here’s my understanding of the process, with questions:
- Creating a model in Reality Capture
After meshing in High detail, I usually get models ranging from 50 do 100 million triangles. These are wood logs, axes, shovels etc. I would assume this is way too many polygons for both film and games. I couldn’t find what the rule of thumb for polygon counts is. What should I simplify my model to (reduce triangle count) in order for the model to still be considered „high poly“ and will be on par with other photoscanned assets from the Marketplace? From what I can tell, similar props which are considered “high detail” in the Marketplace have around 50K triangles.
- Textures
I have to export diffuse and normal textures from reality capture, right? If I want to have roughness or metallic maps, I should paint them seperately either in Photoshop or some 3D painting program?
Also, what textures are considered necessary for the model to have? Only the diffuse and normal or something else too?
- Editing the model before importing to Unreal engine
After exporting .fbx from Reality Capture, do I absolutely NEED to edit the model in some other software like zBrush or is the model good to go as-is? Let’s say I don’t want to have roughness or metallic maps for this.
If I need to edit the model in software like zBrush, what excatly do I need to do here? This part of the process is the least im familiar with. I’ve found tutorials talking about texture baking, simplifying and then increasing the poly count back again, etc., what exactly do I need to do here? What is my goal to do with this step, and what is the process to achieve that goal? Give me at least some keywords to search, or some tutorials, if not the step-by-step process.
- Unreal Engine
I import the model with textures to Unreal, create a Master material with Material instances for each model, make sure my folder hierarchy is correct and then I’m ready to submit to Marketplace?
As for LODs, there is an option to export them straight from RealityCapture, but should this ideally be done in Unreal after importing the high res model? Do I even need to create LODs (Megascans don’t have them from what I can tell)? Can I provide the buyer with a full-quality model and let them handle and edit the model how they want to, and I just provide the materials and 16K textures? Can I provide them with a full-quality model and then also include a separate model that is simplified with a few aditional layers of LODs? Don’t know what is the norm here. I will enable Nanite, but from what I can tell, LODs should still be created.
If I’m forgetting anything, please tell me any additional information you might think of. I know it’s a lot of questions, but I really can’t find any beginner friendly workflow tutorial for selling props from photoscans, from RealityCapture to Marketplace.