Hello everyone!
I’m thinking about starting a new game project heavily inspired by The Witcher 3. I absolutely love how that game looks, feels, and handles its massive, seamless world. I know a massive team of over 1,500 people worked on it using their own custom engine (REDengine), but instead of being discouraged by that, I really want to discover the techniques and workflows within Godot to create a similar seamless experience for the player.
That’s why I’m coming to the forums today. I want to make sure I lay down a proper architectural foundation before diving in head-first, especially since open-world games require serious optimization.
Right now, I am just planning to use placeholder assets (like simple colored blocks) to see if the core technical concepts can even be made by a solo developer in Godot. So far, I’ve found a few amazing plugins that seem like they will heavily lift the load:
Terrain3D (for high-performance landscape rendering)
ProtonScatter (for procedural asset placement)
Open World Database (OWDB) (for data management and chunk loading).
I would love to hear from anyone who has used this specific stack of plugins (Terrain3D + OWDB + ProtonScatter) or has tackled large-scale 3D environments in Godot. What are the biggest “gotchas” or architectural pitfalls I should expect when trying to replicate a AAA open-world feel on an indie scale?
The Witcher is based on a richly developed world inspired by literary works.
Are you planning to develop this game on your own? What is your experience with game development and programming?
The biggest challenge is filling the world. If you automate it, the world will become repetitive (this is actually noticeable in the “Out of the Ashes” demo), and doing it manually requires a lot of people (The Witcher 3).
How many millions of dollars budget, or the equivalent in man hours from a proven team do you have access to ?
Sure you can make a tech demo or prototype of a CRPG with all sorts of game systems, sure. As others have already said, filling it will be most of the effort, as even a semi open world that’s mostly procedural will need a lot of handmade content to be even a half dozen hours long.
I fear you are seriously underestimating the amount of work and effort making a game of that breadth and scope.
Are you experienced ? If not, start with small projects you can finish.
Starting a game is easy. Finishing and shipping one is very hard. Having it a commercial success is the rarest of things.
That said, game development, the art of making a game is in and of itself something incredibly fun and rewarding, if you can be satisfied with that, but you need to learn as you go. Setting yourself up for little successes by starting with small projects you can actually finish all in a year or two.
Granted, from what you’ve wrote, I assume you have little to no experience with game development, programming and design in particular.
Thanks for the feedback and the link to the documentation!
To clarify where I’m coming from: I’m not jumping into this blindly hoping to release an MMO next week. I’ve spent time with Godot, GDScript, and the editor via Zenva, YouTube, and various GameDev courses. While I haven’t published a commercial title yet, I have a solid grasp of the basics, and my “budget” for this project is simply my own free time.
Seeing the release of the REDkit for The Witcher 3 is what really inspired me. I’m treating this purely as a technical hobbyist exploration. My goal right now isn’t to build a finished game, but to experiment with placeholder blocks to see how far a solo developer can push Godot’s open-world architecture.
Instead of the absolute basics, I’m specifically looking to discuss the high-level engineering challenges of a large-scale map.
What inspired you in Witcher is probably not data streaming, but all the artistic aspects of the world. And that you won’t be able to replicate without already mentioned millions of resources.
I downloaded the plugins to start benchmarking, and it turns out Terrain3D and ProtonScatter do not want to play nice together in the editor. Separately, they work flawlessly. But the moment I try to generate or update terrain while ProtonScatter is active, the entire Godot editor artifacts completely and becomes unusable.
It looks like a severe rendering/Vulkan conflict, likely caused by ProtonScatter trying to dynamically sample heights while Terrain3D is updating its storage textures on the GPU.
Has anyone else encountered this specific editor instability when pairing these two plugins?
Have you unpacked The Witcher 3 yet? Did you have enough disk space? These are just the finished resources. The source files will take up an order of magnitude more space.
From personal experience — there will be more problems down the road.
Thankfully, disk space isn’t an issue! I have plenty of room on an external SSD, which is definitely a prerequisite for dealing with large-scale source files and unpacks.
You need to work with these files, not just keep them in an archive. There was some solid advice here: start with a small game in a small location and expand gradually.
Once you’ve filled the world, you’ll need to bring it to life — the second insurmountable task
What are you basing your game on? What will be the foundation — what’s the idea?
my goal right now isn’t to build a finished, shippable game with a complex story. I’m focusing strictly on the technical foundation.
Think of it like building a car engine: I want to see if the engine can handle high speeds (seamless data streaming, chunk loading, and rendering) before I even think about what color to paint the car or where the passengers are going. The “idea” right now is purely a mechanical prototype—a player character moving through a massive, seamlessly streamed environment filled with placeholder blocks.
If the technical foundation can’t handle streaming data smoothly, then the best story and art in the world won’t save it.
If anyone has insights on stabilizing Godot’s editor when sampling heights for massive scatter networks, I’d love to focus on that!
That is a fair analogy, but even if you’re building a bicycle, you still have to figure out how the chain and gears fit together! For an open-world project, data streaming and chunk management are those basic gears.
But out of curiosity, were you an advisor for The Witcher 4? Because CD Projekt Red actually did exactly that—they ditched REDengine to use Unreal Engine 5 for the next game!
I haven’t even played any Witcher. I think I tried Witcher 2 at some point but got bored after 5 minutes and never started it again, although I particularly remember not liking how the game looked.
This happened because of internal studio scandals and conflicts, not because Unreal is better. They fired the old team, and the new one wasn’t able to master the more advanced REDengine.
You can check out Unigine. It seems to support large open spaces.
Filling the world and making it interactive will be a challenge but the biggest challenge will be yourself dont start creating systems that look cool but don’t really help the game that causes burnout fast but it’s achievable as a solo dev good luck
