Fantastic. A glimpse into the future.
A nice waterfall:
Haha! Just watching the videos is fun, though for practical purposes are there any plans for multi-grid/importance sampled sims? (Multiscale simulation detail where needed, I imagine just querying generating and querying SDFs would give you about a perfect approximation of where itās needed). And also any plans for caching sim results? Non interactive areas would benefit greatly from caching, EG a big unreachable river down below a bridge or even just a distant one.
Or really I guess there are a thousand ways to reduce this to utterly practical use, like just using it to spawn water particles from splashes or using it for particle based cloth sim. Actually that last one would be the most interesting, as youād vastly reduce the number of simulated particles needed and could support pretty much ubiquitous cloth sim. Anyway looking forward to it being available, good work!
Perhaps a naive question but do you intend to release this as a general plugin and if so, do you have any sort of estimate when it might be available?
Excellent work.
I have had an interest in realistic water in Unreal Engine for a long time.
My target to build a realistic forest river (stream) that interact with rocks, have foam, etc, like in this video:
It is seems that this is achievable with your plugin.
Hope that you will release it on Unreal Engine Marketplace soon.
Awesome! will you release on marketplace as blueprint ?
Yes, Iāve already mentioned it in my previous posts. At the moment we are still focusing on quality. Weāve improved results in the last month quite a bit, but it is not really visible in this waterfall video:).
I didnāt really think about that option. I guess we would quickly run out of memory. If we find any practical applications, we might do it.
If you mean general = available for anyone to buy / use, then yes. If you mean general = including all other physical phenomena, then the answer is that we will try to incorporate as much as we can. If it turns out to be beyond our reach with limited (human) resources, we will try to incorporate it in existing physics engines.
We will not release it before it is practically applicable, that means before we implement all approximations planned and include additional features, at least constrained rigid bodies (ragdoll etc) . Itās hard to estimate (maybe 6-12 months), our further steps might also depend on potential customers.
We havenāt decided yet, how is it going to be released. Possibly on the marketplace. I will have to read the terms and conditions.
Hello,
Looking awesome. I am considering using this plug-in for two scenarios in my project Crusoe (link below):
- A piping system for agriculture: Connect several poles with each other and let the water flow through them; the poles are placed horiozontally and will be open on the top so player will see the water flow.
- Rather unlikely but Iād like to explore the idea: Water on the beach. I am considering adding water physics to the beaches and near the rocks such that foam will spawn when in contact with them. Somes ocean waves that will propagate are also in planning. Can you system handle waves (well)?
A note to drop: If I do use your plug-in it would be integrated in a few months time; it is not a core system of the game but would be a nice add-on for the visual quality.
Thanks for the plug-in and your time. I wish you much success with your work because itās great.
Cheers,
Univise
That would probably already work well.
We plan to try an simulation on the top of Gerstner waves - that means near objects in water (ships) or new boundaries (beach), we would simulate using our method and the result would be continuously blender to waves.
our video was accepted to SIGGRAPHā17 Real-time live session
you can find our submission video (2min limit) here:
it contains a compilation of our old scenes, but as you can see, the quality improved quite a bit (for example swirls are formed nicely)
Saw your demo at Real Time Live last night. Very interesting. Nice work, good demo.
> if there is any interest in the game development community
> I would gladly sacrifice the chamfers on my stone wall assets or the density of my grass to implement this.
Yes and Yes 
you can find my siggraph real-time live demo here
the same scenes + one additional color-mixing scene + a few effects (tornado on the sand castle) + some description
@lubor_ Congrats on the Real Time Live presentation! I couldnāt attend SIGGRAPH this year, but we found your work very exciting 
I didnāt realize you were on our forums actually. Are you looking at a UE4 integration? Iād be very happy to discuss that more with you if you were interested. Iām curious about any problems you have run into, or architectural ideas you have for the engine. We would like to eventually run more physics on the GPU (as an option, anyway), though we need to make sure things are implemented in a cross-platform way. Good luck with everything!
@ any ETA for an Unreal Engine plugin for this at the Marketplace? Thanks
The state is as follows. With our CUDA implementation, we successfully managed to simulate fluids and visualize it in Unreal. However, we run into this problem: https://answers.unrealengine.com/queā¦-from-gpu.html . It did not seem to be possible to add a fluid into the rendering pipeline without transferring CUDA data from GPU to CPU and then back to GPU. This is a big problem for fluids with foam, where we simply can not transfer 1M foam positions every frame. If we directly edit the buffer from the main thread, it flickers every now and then, probably because we write to the buffer exactly, when the rendering thread renders it. We converted our code into openCL, which runs a bit slower on NVidia cards, but works cross-platform, but we didnāt manage to resolve the data transfer issue there either. We concluded (based on various topics and articles on internet), that this is possible only if we implement our method in Compute Shaders and running it in rendering thread, which we did not do yet.
To be honest, game engines were not of the highest priority recently. We needed to focus on stuff that generates (or will soon generate) revenue, and it is not from the gaming community. This is not of a great surprise, because as it is, it requires too much computing power for simulations of reasonable scale. This might change in the near future (some stuff under progress). We are quite a small team and rewriting to compute shaders was not of sufficiently high priority. The progress might get faster if we manage to get an investment.
We got it to work with VR. You can find the video here:
Itās really lots of fun to play with. Unfortunately, the video canāt really capture the VR experience, but it is really awesome.
That looks fantastic. What kind of rig are you running to get 60-90fps from that scene in VR? Iām really looking forward to more tactile physics sims being VR viable - really opens up some amazing new interaction possibilities!
the same as previous simulations - NVidia Pascal Titan X (the older version)
runs the same on GTX1080
anyway, we are still trying to improve the quality of our simulations
it starts to look comparable to offline simulations
This looks amazing! Please release this soon.
Iād be curious as to if you could simulate a person walking on a thin layer of sand?