Are there any dForce Simulation Benchmarks?

Sevrin

Like many, I use dForce clothing and transmapped dForce hair quite a bit.  I was wondering if anyone had compared GPU generations or models for the impact they have on simulation speed.

oddbob

I don't think there are any suitable assets provided with DS or the essentials packages to build a test scene. Outside of that there's a lot of variables that could make an impact.

Pendraia

I've been comparing various settings in the surface tab but haven't looked at the speed.

Expozures

oddbob said:

I don't think there are any suitable assets provided with DS or the essentials packages to build a test scene. Outside of that there's a lot of variables that could make an impact.

 Agreed.  There are some hair assets, for example that I have that take only a minute or two to complete, and some that take several minutes.  I think it all depends on the vendor and how they made the asset.

marble

I upgraded from a GTX 1070 to a RTX 3090 and it made a huge difference to my render times but I hardly noticed a difference in dForce simulation times (one of my regular complaints here, in the forlorn hope that the message will somehow get back to the DAZ development team). I have Marvelous Designer 8 (which I really ought to use more) and version 8 was before GPU assisted simulation and it is STILL orders of magnitude faster than dForce. AND the cloth can be tugged and pulled during the simulation. If DAZ were to employ people to develop a simulation plugin that would come close to the MD product (just the sim, not the clothing creation) I would pay good money for it. There was VWD which seemed pretty good but it constantly crashed my DAZ Studio.

kyoto kid

..all I know  with a Mwxwell Titan-X  I'm deaing with 15 - 20 min sims on average. A real workflow breaker.

Sevrin

oddbob said:

I don't think there are any suitable assets provided with DS or the essentials packages to build a test scene. Outside of that there's a lot of variables that could make an impact.

There are no official assets for rendering benchmarks, either, and yet, it's being done.  Eliminating variables is part of the process of setting up a benchmark

It could done as simply as rendering something with a typically long simulation time , say, Linday's Classic Long Curly Hair with dForce for Genesis 8 Female(s) on Basic G8F in the A-pose from start to finish with a defined set of simulation settings.  There's no need to overthink it.

memcneil70

If you look into the contents of dForce Starter Essentials, there is of course the dress, shoes, but also a hair (even if it is a mohawk), a primitive flag, sheet drop, blocking fan scene. Could someone make something out of that for testing?

Maybe with G9 a dForce hair suitable for a woman might be added to G9 Starter Essentials? Or more dForce clothing?

 

Pendraia

Sevrin said:

oddbob said:

I don't think there are any suitable assets provided with DS or the essentials packages to build a test scene. Outside of that there's a lot of variables that could make an impact.

There are no official assets for rendering benchmarks, either, and yet, it's being done.  Eliminating variables is part of the process of setting up a benchmark

It could done as simply as rendering something with a typically long simulation time , say, Linday's Classic Long Curly Hair with dForce for Genesis 8 Female(s) on Basic G8F in the A-pose from start to finish with a defined set of simulation settings.  There's no need to overthink it.

This is pretty much what I've been doing while testing the different settings in the surface tab. I'm using a base mesh, very simple pair of pants I made in Zbrush. Same pose consistently and just trying the different settings.

One thing I discovered though reading rbtwhiz's documentation is the collision layer setting requires more than one item though. So I'm currently reading his info and trying to work out how the different settings work.