Need pro landscape advice
Hello there,
My friend is awaiting solutions from me and I'm so desperately lost after doing my researches for the subject. I have a very difficult time with sorting out a landscape for a walking simulator game such as dear Esther. To be honest, It's something I want to do but with all my 4 years experience with terrain stuff I'm not able to find a solution on this one. I mean I know Battlefield 3's Alborz Mountains map is 8 kilometers square, I know the level of detail there, I know how they populated the distance but this is a tier above that. Please share with me any chances I can achieve exactly what he wants.
Restrictions:
1. My task is to create the entire Omaha beach meaning 9x9 kilometers for UE4 with it being based on the real beach so no randomness anywhere.
2. It has to be made out of landscape so no use of geometry or distant assets.
3. Can't use programs such as World Machine 2 since we can't deliver details at a man scale like small narrow paths etc. at a 9k map in WM2 using the node based graph so everything has to be hand sculpted.
4. It has to be fully detailed all across the 9km square to a point you can do comparisons between UE4 shots and real life pictures and say Damn It's Close!
Where did the allies first make landfall, where they crawled up the hills etc. It has to look identical due to the historical importance of this..
Now my concerns:
1. The player would barely explore 1.5 km of the 9km. The whole point of having 9k landscape is so they look at any direction and they get details as if it was real life and say oh look I see the details that are 5-6 km away from me.
3. There are at least 4 landscapes to be made with similar situation. To be fully designed until they are ready to publish with the game. What kind of time frame would you give for such a thing if it was asked from you? (This is one important question)
4. Such a big landscape would drain a lot of FPS due to number of components even though it's still a bare landscape. And target audience aren't people with high end machines.
5. Since it's going to be made out of landscape and no distant assets put together, we need to not use aggressive LoDs for the landscape or else it'd look so bad when you are miles away from other areas, it flattens the details. That again drains more FPS on top of what mentioned above.
6. Making a tiled terrain and loading/unloading would be an option and still the view distance and the amount of detail is so high so we can't unload that much of the work that's in front of the player view. Yet the performance is supposed to be in lines of medium to high end machines. I'm running my prototype landscape on an i7 4790k with 8Gigs and I have no smooth results. It'd probably crash if scene was designed with assets.
I'm keep refreshing this page for replies.
Thanks.
My friend is awaiting solutions from me and I'm so desperately lost after doing my researches for the subject. I have a very difficult time with sorting out a landscape for a walking simulator game such as dear Esther. To be honest, It's something I want to do but with all my 4 years experience with terrain stuff I'm not able to find a solution on this one. I mean I know Battlefield 3's Alborz Mountains map is 8 kilometers square, I know the level of detail there, I know how they populated the distance but this is a tier above that. Please share with me any chances I can achieve exactly what he wants.
Restrictions:
1. My task is to create the entire Omaha beach meaning 9x9 kilometers for UE4 with it being based on the real beach so no randomness anywhere.
2. It has to be made out of landscape so no use of geometry or distant assets.
3. Can't use programs such as World Machine 2 since we can't deliver details at a man scale like small narrow paths etc. at a 9k map in WM2 using the node based graph so everything has to be hand sculpted.
4. It has to be fully detailed all across the 9km square to a point you can do comparisons between UE4 shots and real life pictures and say Damn It's Close!
Where did the allies first make landfall, where they crawled up the hills etc. It has to look identical due to the historical importance of this..
Now my concerns:
1. The player would barely explore 1.5 km of the 9km. The whole point of having 9k landscape is so they look at any direction and they get details as if it was real life and say oh look I see the details that are 5-6 km away from me.
3. There are at least 4 landscapes to be made with similar situation. To be fully designed until they are ready to publish with the game. What kind of time frame would you give for such a thing if it was asked from you? (This is one important question)
4. Such a big landscape would drain a lot of FPS due to number of components even though it's still a bare landscape. And target audience aren't people with high end machines.
5. Since it's going to be made out of landscape and no distant assets put together, we need to not use aggressive LoDs for the landscape or else it'd look so bad when you are miles away from other areas, it flattens the details. That again drains more FPS on top of what mentioned above.
6. Making a tiled terrain and loading/unloading would be an option and still the view distance and the amount of detail is so high so we can't unload that much of the work that's in front of the player view. Yet the performance is supposed to be in lines of medium to high end machines. I'm running my prototype landscape on an i7 4790k with 8Gigs and I have no smooth results. It'd probably crash if scene was designed with assets.
I'm keep refreshing this page for replies.
Thanks.