Anastomosis won't export

[mr_fish]

I had a previous attempt hit 10 minutes and still be at 0%

I have a 9950x3d and 128gb ram

[TyXanders]

Personnally i can’t repro base on this information.

[MarcSaunier]

I just exported a 8K 3 nodes test terrain : mountain range / Anastomosis / Sandstone

Although it’s not listed as an heavy node, anastomosis is the one, from far, who took the most time to build using 100% of cpu (i9 13900 K). but it built and exported.
Maybe you didn’t wait long enough.

[mr_fish]

Oh wow, more than 10 minutes for 1%. Think I’ll just leave it out. Do you know if there’s anyway to get a similar effect without using that node?

[MarcSaunier]

Honestly I don’t but maybe some gurus here do !
However baking nodes, the build/export took 1 sec with the test terrain.

[TyXanders]

@MarcSaunier

Although it’s not listed as an heavy node

Just an FYI it should be listed as a heavy ( this is relative to the graph), by default the quality is as low as it can get, but the modes and values can change accordingly

Here is the difference between quality modes min and max and downcutting on the same graph

[TyXanders]

If you’re using it early for shaping, bake it, use the statistics tab to help gauge your nodes as well as graph tooltips.

not all nodes modes have the same performance, so think carefuly about their values, modes and impact

[MarcSaunier]

Erf… Look at this:
Anastomosis full quality, listed as heavy node on your screenshot and not on mine

[TyXanders]

Try again in a 4k viewport

[MarcSaunier]

• Now it’s listed as heavy node but… mountain range not anymore ?!?
• So, you have to be in a 4K viewport to have the “nodes informations” ( heavy node etc..) working correctly ?
  
  

[TyXanders]

This would be because of input complexity too, the simulation itself is heavy, but also depending on the complexity and samples of the input/to process, it too could increase or decrease the heaviness.

If you blur an incoming terrain keeping most of its base shape and add an anastomosis node with the same settings on each.

You may be able get an idea of the separate cost from the input Vs node.

You should also see this difference if you compare a baked and non baked input too

[MarcSaunier]

This is way less simple than what I thought. I should have studied more younger :)

[TyXanders]

Bless you, You’ve got this and you’ll get there.

The main thing here is where and how to optimise where it is heavy.
The tools you have available are:

• Baking stable upstreams
• Masking smaller sections for a heavy process to process
• using less detailed inputs
• using it earlier in the graph

You may even be able to use results from low level pass of information and then transpose the flat results into a higher fidelity pass in a section.

Not something I’ve tried but it sounds plausible along side a graphics EQ node.

[MarcSaunier]

1st two (baking, smaller sections) already indeed change a lot the game.