Видео

As explained in description of the video, languages Python + Taichi (+ Numba) from scratch.

I explain this in the description : Gauges : From up left to the right : * Level of filling of reaction chamber * Temperature in chamber * Start/stop light Below the reactor left to right : * Flow speed at exit of chamber * Flow speed at exit of nozzle * Thrust at exit of nozzle

@ ok thanks

yes the cohesion is a bit too strong. I think I should look at this. But Mars is destroyed completly, and do not forget that the mass of earth is ten (10) times the one of Mars, so it is not completly abnormal to see Earth not exploding like Mars. However Earth is not far from the explosion because his diameter increases a lot after the collision.

@ yeah?

this is still VERY fast for an SPH simulation (I've seen your other video, 1 hour for 80k particles is pretty decent imo)

In fact the SPH calculations are done by Taichi running on the GPU CUDA (Nvidia RTX 4060). Python just encapsulates Taichi. For the launching of the program and mouse detection I use Numba (a "Just in Time compiled" & paralell on CPU sub-program of Python). For the display again Taichi, for the recording of the images and then the generation of the video I use Moviepy when the simulation is finished. Taichi is a "Just in Time compiled" langages and is massively parallel on the GPU . So the simulation IS compiled ! In fact for this one it takes 0.7 second par loop (or step as you want). Here it has 10000 steps so it took 116 minutes (during the night...) but the result is a long video of 9 minutes.

@ ooh, I have a 4060 too

nvm I just dumb, I looked at the dec

Thanks !

Or many galaxies... I do not know !

I am not sure to understand what do you meand by "textures", I do nothing special for that in fact... if you mean drawing seas and continents I do not know how to do that

@ no I meant where are the textures for the planets

There is the gravity first. I added the new feature I talk about in the description but only if two particles are very close and with a low cinetique energy.

3200 steps (loops of the program) * 1.45 seconds per step / 60 = 77 minutes... thanks to Taichi !

I checked what you said and you are right in term of physic. I will try this ! Thanks. uk.pinterest.com/pin/191895634108480447/ .....I tried "blue azur", "blue electric" and "blue" and it is not OK, I have the feeling to see ... water. I think this is coming from two facts : 1) A star emits the blue color at a temperature above 6000 °K. The star Rigel is blue beacause it is at 11000 °K but this temperature does not existe on earth (except in some research laboratories). 2) for a human being in his day to day life, blue means mainly water and water in the nature is naturally cold. This why painters call "blue" the "cold color". I think I will keep white for the "maximum" for instance still for a human being the sun is white because his surface is at 5770°K. But I will try to change the intermediates colors. commercialledlights.com/blog/lighting-articles/color-temperature/ Feel free to tell me your opinion. Thanks again

Thanks !

I tried to explain this in my description, but it is a little bit too long : Two gauges are on the left : above the chamber it shows the filling level of the reaction chamber. The red part indicate if it is below 50 %. The one under the chamber indicate the ejection's speed of the burning hot flow of gaz measured after the exit of the nozzle. So... I added a description panel directly in the top left of the screen for my next post (today), I hope this will clarify these gauges. Thanks for giving me this idea !

Yes me too. But I must find how a particle bounce on such a deflector, for instance a cone to have a better flow behind the sphere...

A little bit more than 100 K but if I collide two stars I can have 200 K... and wait 1 hour to have a video of 3 minutes length. Taichi on Nvidia GeForce RTX 4060 Laptop GPU.

These colors show the temperature, grey the lowest, then red, orange, yellow, peachpuff, purple, lemonchiffon for the hottest. I tried to be realistic....

Yes unfortunatly I know, but I am at 1 second per iterartion/ step/loop of the program on my graphic card (I have a Nvidia RTX laptop 4060) for calculus and display with Taichi... and too impatient to wait hours before being able to check the result....

No, it is written from nothing, I realized this with Python, plus Numpy and Numba to initialise planets, stars, or galaxies, and the simulation is done with the Taichi language (Taichi works as subprogram of Python) for the simulation itself and graphical display. Nothing esle... Taichi is able to use directly the graphic card (I have a Nvidia RTX laptop 4060) for calculus and display.

use opensph or spacesim instead you can make much better planet collisions with opensph or spacesim then this@@tristandeme

Sorry... but no. It is much more interesting for me to do it from scratch. Spacesim or Opensph are boring, you use something made by someone else, as a basic user, and then what ??? You learn nothing !

ok@@tristandeme

@@tristandeme Um, but I don’t understand about the second reply?

Dont you mean space sim?

Real

Thanks. do you mean add more explanations on the algorithms I use ? It is SPH comparing pair of paticles (so N**2 vulume of calculus it become quickly very long after 100 K particles) and I try to apply Navier-Stokes SPH theory.

@@tristandeme Yeah basically I was. But also you can kind of commercialize this if there’s enough demand for it or work with other people that are building SPH simulation software like SpaceSim

@@tristandeme like making it public use and possible raymarched

i mean, there wouldnt be much of a point to making knockoff spacesim

Yes

Yes you are the first !