Im especially interrested in low latency settings that are good since i dont really know what the go to set ups are for you hardcore users when you want to do "realtime" work.
The highest quality settings are also of interrest.
Im asking about this since im working on something that i think will be of interrest for many of you guys and gals using nebula. Maybe ill post some pics later.
i do not work with low latencys (real time usage of nebula), to say that first. but when you want to lower the cpu hit and you do not need the distortion kernels calculated, then go to the kernelpage and set the number of kernels to 1. if the first kernel is set to timed operation or splithybrid (timed and freqd), set it to freqd only. this will give you a very low cpu hit, when the kernellength is not too high.
for highest quality, there are some threads you can search for. basically it is about the timed operation. you can set the timed kernel to 30 - 50 ms length, if you want for all kernels, when your cpu is capable to do that. some programs react unpredictable, so you have to test yourself, what programs will work and with which kernels. when i remember correctly, nebula is not meant to do all kernels in timed mode, so there are limitations. maybe someone can chime in who knows the correct information about that.
system 1: windows 8 32 bit - samplitude prox, tracktion6, reaper system 2: mac osx yosemite - reaper(32+64bit), tracktion6(32+64bit)
both systems on: macbook pro (late 2009), core 2 duo 3,06 ghz, 4 gb ram, graphic: nvidia geforce 9600M GT 512 MB
I think timed is limited to 30ms because of cpu ,you can change it in the xml if you want it to go higher.
I've found apart from timed mode mod . Getting nebula to run at 0.5 ms program rate should cover audible transients .
so you could mod that too. You should hear better depth and room due to transients being captured. Like slow preamps don't pickup the room sounds so quick because of the round off, thats a little like nebula running slower.
Last edited by Tim Petherick on Tue Aug 07, 2012 2:28 am, edited 1 time in total.
This is exactly what im looking for. I dont know if this img link will work but if it does it illustrates what im currently up to.
On the pic im running 7 instances of the CLC eq impulses and i have implemented automatic switching between rendering modes and realtime modes and some stuff that makes life a bit easier. Im not really sure as if this is something that you guys would fint interesting but it might. Another point with the plug is to provide one interface to a whole strip. All impulses will get bypass buttons to make it possible to A/B setting without PDC change. The big silver knob drives all instances.
I dont feel quite confident with the plugin just yet but in a couple of weeks i might let you try out a beta or something.
It will only support 64bit coreII nebula unfortunately (due to lazyness i guess).
since im wrapping everything in one single plug i dont want the users to have to mod xml files. Id rather find general solutions available to all users to maximize audio quality and usability. This is something i would like to find somekind of consensus around but since applying FX or doing all kinds of offline rendering automaticly switches to different settings i think that cpu power shouldnt be a problem.
The plug is currently working quite nice but the switches between quality modes is a bit sluggish since conversion between kernels takes some time.
One more goal with this is also be to be able to find settings that makes it possible to lower latency to 128 for realtime and still be able to EQ and set compressors decently.
I must also add that at the current state all settings and impulse control has to be hardcoded so this pluin will probably never be flexible enough to give users own control over all settings in different modes.
GUI currently looks like crap - i know.. Ill try and remedy that.
I think ill get this thing working a bit faster than i expected. I am a bit disappointed by cpu load of over 40% 96kHz 24bit on the lowest quality settings. I tried to do only clean with 10ms of FREQD engine. Is there other tricks that could lower the load further?
Fixed some stuff regarding automatic samplingrate switching.
Does anyone use a frequency spectrum when EQ:in? I think ill leave it anyway, cant do any harm i guess.