Hello !

So just for fun I wanted to build a graphical EQ. And it works just fine and it's lovely to watch

But maybe im going about this all wrong. Im creating 1 new channel for each “range” of the music input with a “Band EQ” in chops.
But boy does these eat memory, when I have 12 of these Houdini goes 10GB ram used.
Is there any smarter way to achive this without it eating so much memory?

I've attached a picture also to show what I mean

/M

Try turning on “Unload” for all the CHOPs except the last one (ie. math1). You might need to restart Houdini.

10GB???? oh my God!

it will take some time for i experience the power of CHOPs :cry:

additionaly to unloading
if you have stereo audio try deleting one channel right after file chop unless you need both channels
you also may want to trim your audio just to scene length before processing

using some nodes in Time Slice mode may also reduce memory usage

and have a look at spectrum CHOP it may be what you need for such effect

Sty
10GB???? oh my God!

it will take some time for i experience the power of CHOPs :cry:

do not be afraid of CHOPS
in SOP you can easily make houdini want to use 100000000…..GB of memory and I bet you are using them all the time

in CHOPS it is the same, you just need to be aware of your steps so you avoid wasting memory
it especially applies to working with audio where the sample rates are very high

So much replies, thanks all !

I tested “unload” but that didnt do much for my case a slight lowering in the memory, not enough.

“time slice” took it from 10GB all the way down to 200MB in memory but then no realtime feedback when pressing play.

and the Spectrum CHOP sofar im either not understand or it doesnt do it for me. But I will investigate further after a trip to the food place

/M

After som testing I put a “Resample” chop just after the File node, and this dramaticaly decreased the memory while maintining realtime preview.

And actually decreasing the sample rate with the “Resample” CHOP made the graphics smoother, though still responsive enough for it to look good

Thanks for all the answers

/M

yes resample will decrease memory, but you lose high frequencies which can be very useful but off course all depends on what you are doing
however i agree that default audio frequency may be vay higher than needed

also take look at pitch CHOP, it can create channels with different frequencies for you to avoid duplicating equalisers

then after getting the frequencies (either way) you may want to filter them to get smoother results, or resample more or add Envelope chop or Lag chop to tweak the channels a bit
do not forget to resample it to your FPS at the end to get as memory effective channels as possible, you can also lock the last node if you are satisfied or save it to disk and load with another file node to avoid recalculating and memory usage in the future

EDIT: i forgot to mention that if you are processing audio this way, with intention to save it to disk and use later, you may want to turn Time Slice off, i have not much experience with it for such case. It is ment to be used for realtime processing i think