I’ve decided to do a wee post mortem on the PET scene, as there were a lot of issues in it and I’d like to learn from them! This was essentially an exercise in “doing things I ask people not to do”.
I am not good at planning. In uni all my projects were totally overscoped and at work I tend to jump head first into solving problems instead of taking a step back. Its something I’m working on.
To help projects get finished, I need:
- Have a clear idea of what the project actually is and what the finished product will look like
- Have a list of assets or tasks
- Stick to one part at a time
A major blocker to finishing this was bad organisation. Files were not in one location, were not named well and were not easy to find. Due to this, I’ve rearranged my project folders into a much nicer format, with separated work and final folders.
Next project I need to:
- Use the new folder structure on my pc
- Use a sensible naming convention that ensures uniqueness across all resources
- Use a sensible folder structure in unreal
The models in the scene were definetley worse than I can normally make. I didn’t spend enough time on them as I was anxious to finish. I skipped some steps which I shouldn’t have!
- UV AS YOU GO (been telling myself this for years, yet I always forget!)
- Bake AO as you go
- Create full maps for everything
- Don’t guess PBR values (c’mon…)
- Use an export tool (I have a maya one, but haven’t made/downloaded a max one yet)
Shader/Texture Authoring and Organisation
The amount of materials in the scene was disgusting…instancing exists for a reason!!!
- Use generic “ubershaders” as much as possible
- Anticipate needs beforehand then adapt to the project, don’t make them up as you go
- Have a limit on number of shaders (“I do optimization at work, so fuck it” is not a good attitude…)
- Be smart with channels and compression – lots of things could have been packed into channels of one file or used certain compression types to ditch/favor certain channels