I’ve played with the solution of harveymoon, but at 163 840 pixels it’s very laggy and undoable… My laptop has a 4080. GPU and CPU runs under 12%. I don’t know where is my bottleneck but it’s caused by the group SOP…
For my future project I would need wayyy more than 163 840 pixels…
It’s definitely a CPU bottleneck. You wouldn’t happen to have an 8 core CPU would you? TouchDesigner is single threaded so you can only max out one core of the 8 - hence the 12%
I would recommend investigating RayTK. I just whipped up an example file that can render almost 200,000 points and 1200 universes of DMX:
It runs fine at 60FPS on my 3070.
Here’s the file with RayTK 0.36 already embedded in it.
In this case the banana point clouds would be the XYZ locations of the LEDs. and then there’s some RayTK operators moving through them and colored by the point material operator. It all gets downloaded into CHOPs and split into the ~1200 universes, just not hooked up to a DMXout CHOP. The load on the DMXout CHOP may be a little too much for the Windows network stack, but you could split this to two files and keep them in sync easily.
Make sure to run it in TD 2022.35320 or earlier as it’s still a little buggy in the 2023 versions yet.
Wow thanks, i’ll try it out ! I’ll let you know!
Yes my CPU is a 16 cores and 24 logic, I didn’t know TD only use one.
I’ve run dmx output test to see where I’m capped, and I’ve reached about 4200 universe before getting delay and noticable skips. But, for some reason, I needed an external 1gbs network card as windows capped my internal network card for some reasons…
(btw, I think we’ve met at the Montreal TD meetup 2 years ago. I was the guy with the orange glasses.)
