getting better and better
2 Likes
- New functionality of volume upscaling has been added and fixed few times
- Projection kernels has been rewriten for better stability with adaptive domain and to achieve more similar results
- A lot of small fixes, features and improvements
41 Likes
Added possibility of smoke sourcing from Obj sequence.
So one step closer to blender addon
37 Likes
moderation notice, frustrated ranting isnt gonna be as productive as one may think,
i saw ton mentioning some of the videos in this thread recently so i may not be the worst idea to see if the foundations feeling have changed on this
7 Likes
It is worth clarifying the difference between:
- Poorly made engine.
- An old engine that no one has done for a long time.
Please take into account that there are really not enough people. There are places in the code that have not been changed by anyone since the time when my computer was still considered a novelty. So yeah, they might not be as good now.
And everyone is glad that someone is busy in this area now.
1 Like
Where did Ton mention some of the videos here? Even though I agree that frustrated ranting isn’t productive, it is also frustrating to see this person who’s put so much effort into creating a new GPU based smoke sim, a real-time one at that, get rejected, at the same time as I keep seeing Pablo mention during nearly every Blender Today that they’re lacking in people and want us to apply 
8 Likes
@ bnzs We have a clear code of conduct and I ask you to respect that. You can share your oppinion and say that you don’t understand the decision taken but ranting and being impolite is not okay and won’t be tolerated on here.
5 Likes
The aspect of this that I find very confusing, is that the Blender Institute is usually known for taking in talented, but inexperienced developers under their wing and nurturing their talent.
@michal1000w has clearly done an incredible job on his own already in isolation. Imagine with some guidance from Brecht and the other senior devs, along with the Blender Studio crew and it could turn out to be give Embergen a run for its money!
3 Likes
Let’s still leave the way his skills were assessed to those who participated in this.
What you guys seem to miss is that Blender is not looking for “a physics” developer but a “senior physics” developer with certain experience and knowledge.
4 Likes
Small question,
Since a Senior physics developer is really hard to come by, would it be useful to have a Junior physics developer until a senior one becomes available ?
7 Likes
Michael’s system is even now very powerful and extremely fast compared to the existing one. If he would further develop and integrate this system as a Blender developer, many people in the community could already benefit from it.
If a senior developer is found in the future, the simulation team could grow.
The demand in new simulation features is high and maybe isnt fulfilled by just one senior physics dev
5 Likes
Sure, I understand that, but he could be hired for this one specific feature. The Blender Institute has given short term grants to developers in the past for specific features such as lineart, XR, retopo, etc.
Even when you hire a senior dev, it would a large task to create a realtime solver such as this from scratch. Most likely this senior dev will be very busy with the general physics node integration and architecture, and not have time to write individual solvers. Btw, I thought I had read that Lukas Tonne had been hired for this role anyway, or maybe that is for general geonodes?
Likewise, I think the Institute should be trying to hire other well known physics devs in the community to write other solvers, such the team that made the Fracture branch and the Flip Fluids addon dev. There is no way one senior dev is going to be able to do this all themselves! There needs to a team. Just look at the sheer number of developers that work on Houdini.
4 Likes
I think the problem and the key different with Embergen is the fact that blender is intended to have other simulation solver. all of these solvers should share code and communicate with each other very fast, e.g a cloth sim reacting to smoke. this is why a senior engineer is needed.
Because a senior engineer will have prior knowledge in how such interconnecting system should work. what data structures to use, so they could communicate across multiple solvers, and so on.
If the BF hire him and accept his implementation, this will come at the risk of having to change things a lot down the line. and the longer they work with only him, the bigger the risk of having to replace or change a larger portion of it.
Specially since I think the BF don’t have anyone with the experience to review and approve the implementation as a future-proof architecture.
7 Likes
You describe a unified solver, which would be fantastic to have in Blender, but this is a much bigger task than improving existing features and bring them up to date.
I guess a unified solver would require multiple very experienced devs for design and coding.
Sure, I agree with what you say and there needs to be future proofing to get an architecture where the solvers can all “talk” to each other. The main point I was trying to make is that it will take a team of specialists to get Blender to a professional level of modern physics capability.
Who knows how much longer it will take to find the right senior dev and if they can really even afford to pay for such a specialist role. In the meantime, Michal could at least be on some sort of provisional grant under the supervision of the geonodes team and do some preliminary work. By the time a senior dev is found, the opportunity could be gone and he has been snapped up by Adobedesk or lost interest.
Anyway, I will not be commenting further on this issue, as this may be making the OP uncomfortable. Apologies to Michal for derailing the thread…kudos on your great work so far!
3 Likes
Note that a senior physics developer has been hired already.
I believe Lukas will specifically be working on physics solvers (for geometry nodes).
17 Likes
Another test after little tweaks of the high-res mode
34 Likes
I do have to say that with the addition of a test model in the BG, it just gives a whole new perspective on the simulation
5 Likes
Sorry for the unrelated comment…
I don’t know if you plan to sell your addon, but I think that could be a great way for you to make money and be able to develop it further.
Check Flip Fluid, which is a fantastic fluid sim addon for Blender.
They have 10900+ sales at 79$ !
Mantaflow in Blender is great ! But if someone need a bit more fonctionnality, then they would have more options 
3 Likes