Water material

96 views
Skip to first unread message

Zohar Levi

unread,
Apr 29, 2021, 8:58:42 AM4/29/21
to pbrt
I have a water simulation that produces triangle meshes, and I'd like to render them.

I tried dambreak0.pbrt from the demo scenes, and it took 6hr:

Can you recommend something specific that would take a couple of minutes?
For example, Maya Arnold result, less than a minute:

---

By the way, can you recommend video lectures of a Physically Based Rendering course?

Matt Pharr

unread,
May 6, 2021, 10:09:59 AM5/6/21
to pbrt
Try the "volpath" integrator and fewer pixel samples. dambreak0 uses the bidirectional path tracer with a very high maximum depth (which is quite slow) and also 8192spp. Here is dambreak0 rendered with "volpath" and 16spp, which renders in 15s on my system here.

Matt

dambreak0.png

--
You received this message because you are subscribed to the Google Groups "pbrt" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pbrt+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pbrt/88037ad0-ca03-4802-bb5c-5450f349e1b8n%40googlegroups.com.

Zohar Levi

unread,
May 10, 2021, 9:56:36 AM5/10/21
to pb...@googlegroups.com

It's grainy. I played with the spp, and 256spp (12min) is reasonable (in terms of quality, but still slow).

Isn't there an (image processing) algorithm to smooth such images so, e.g., the 16spp wouldn't look grainy?
How does Maya/Arnold do it?
You received this message because you are subscribed to a topic in the Google Groups "pbrt" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/pbrt/Z5JtxpObEgE/unsubscribe.
To unsubscribe from this group and all its topics, send an email to pbrt+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pbrt/CAO4dBWTNUhHO8%3DHcMQ9pXwYAUCmGqDZUwz9A3BvZ-35Sv77guA%40mail.gmail.com.

Matt Pharr

unread,
May 10, 2021, 8:27:45 PM5/10/21
to pbrt
On Mon, May 10, 2021 at 6:56 AM Zohar Levi <zoh...@gmail.com> wrote:
It's grainy. I played with the spp, and 256spp (12min) is reasonable (in terms of quality, but still slow).

FWIW pbrt isn't designed to compete with commercial rendrers, which have hundreds of years of engineering behind them and are designed for performance over understandability.

Isn't there an (image processing) algorithm to smooth such images so, e.g., the 16spp wouldn't look grainy?

The "a posteri" approaches in section 2 of this may be useful: https://cseweb.ucsd.edu/~ravir/STAR.pdf

How does Maya/Arnold do it?

Extensive attention to optimization, including hand-coded intrinsics implementations of critical paths (and presumably different design trade-offs in terms of abstraction versus efficiency.) 
 

Zohar Levi

unread,
May 11, 2021, 5:38:52 PM5/11/21
to pb...@googlegroups.com

Okay, thanks.
Reply all
Reply to author
Forward
0 new messages