I hear you about decimals.
Thatās great news.
The only issue Iām having with multiplayer is world origin rebasing and custom gravity. Iām not sure how double precision would be an issue with multiplayer either.
Yeah I was waiting for this tech, and glad its finally arriving, hopefully soon.
Finally some talk about this!
I was waiting for this tech for a few years now so finally seeing this brings some hope.
What worries me is the fact that they removed any mentions of this - the youtube video is gone/private and the Large World Coordinates mention on the site is also gone ?
From what i see on GitHub - there are LWC (Large World Coordinates) related commits every couple of days so my hopes are up.
Mentions about Large World Coordinates were just removedā¦
Thatās ā¦ very wrong.
First, read this: What Every Computer Scientist Should Know About Floating-Point Arithmetic
Second, realize that any integer can be a fixed-precision decimal number. You can say āthe lowest 10 bits are fractional bitsā and now your integer has a resolution slightly better than 1/1000th of a whole unit. You will have to do some adjustment for magnitude when doing multiply/divide, but itās pretty straightforward.
int64 for the āhigher levelā coordinate would make sense for a few reasons, including supporting equal precision no matter how far from the āoriginā you are. For some kinds of games/simulations, this is highly useful!
The main drawback with fixed precision values is that you need to put away a fair number of bits for the decimal part to support the smallest possible velocity at the highest possible framerate without quantization artifacts. float64 gives you that āfor freeā as long as you stay āclose enoughā to the origin, but, really, you have the same precision problem, just with 54 bits of mantissa (including the implicit leading 1) instead of 64 bits, so it actually hits at smaller radii.
Thanks for the update. Iāll keep an eye on GitHub commit messages.
Keeping track of this thread for any news regarding Large World Coordinates within Unrealā¦ canāt wait for that to be a reality!
Interesting
Update time:
Zak Middleton doing some heavy lifting:
https://github.com/EpicGames/UnrealEngine/commit/75107c146f316d15bc910cb6b7a27502e0d5a2db
If you ever see this - thank you for your work and everyone helping.
Just wanted to throw my hat in here and say I too have been lurking and waiting for something like this. Want to see this become reality for UE5.
page not found, what was it?
Itās a private repository on GitHub, which means only certain people can access it. The content of the page was showing various source code changes adding support for 64-bit floats.
To gain access to the repository (youāll need a GitHub account), follow this:
Thanks for the callout
Itās been fun to work on, and I hope it enables even more awesome creations in UE5
-Zak
I canāt really say as to when it will land, but weāre actively working on it and have made a lot of progress already in UE5. Many teams are collaborating (Physics, Rendering, Gameplay, Networking, etc) to get it to a good state.
Good to see it being worked on from all directions, hope the work goes well and fast because when this hits in combination with Nanite - there wonāt we any other engine that can do that amount of detail on such scale (built-in, not custom made)
Just when you think UE couldnāt possibly get any more awesome!
Its not just changing float to double. physics and rendering is a big challenge.
I wait for Dounreal 5
I canāt thank you enough and your team for adding this, Iāve been waiting for this feature in a quality engine such as yours to add this for over 2 years now. Keep up the amazing work