Convolution technology has provided new levels of realism for plug-in reverb, but has been unable to capture non-linear or time-variant effects such as compression, tape saturation, distortion and flanging. Until now, that is...
You might not expect a single multi–effect plug-in to warrant an in–depth SOS review. But when that plug-in arrives with a 6GB convolution library emulating several hundred items of vintage gear such as preamps, tape machines, compressors and chorus/flanging effects along with the more usual equalisers, filters and reverbs, you sit up and take notice.
The free taster version of this plug-in, which I discussed in May 2007’s PC Notes, has created a lot of interest, since it seems to do what we were previously told was impossible. Not only can it capture the dynamics of non-linear devices such as compressors, but also ‘time-variant’ processors such as chorus, flangers and phasers. Until now, non-linear convolution was only available on hardware products such as Focusrite’s Liquid Mix, while, to my knowledge, time-variant devices have never before been captured using this method.
Nebula’s convolution technique is based on ‘Diagonal Volterra Kernels’; in effect, tiny instantaneous snapshots of real-world effects. It processes lots of these kernels at once, and then, rather like a wavetable synth, can scroll through an array of slightly different waveforms to morph them, using a ‘vector engine’ controlled by LFOs, envelopes and so on. For instance, by mapping the input level of the signal to the position in the kernel array, you can reproduce level-dependent processes such as compression or distortion, while scrolling backwards and forwards through a kernel array using an LFO reproduces cyclic sounds such as chorusing and flanging.
Nebula 3 runs on Windows XP, 2000 and Vista, but not Windows 98. It was delivered to me on two DVDs, although you can download the whole thing if you prefer. The first DVD contains the plug-in and a 1GB core of Programs and Vectors (Programs contain the engine instructions, while Vectors contain the kernels themselves), while the second DVD contains an additional 4GB of larger Programs and Vectors encompassing the longer reverbs. The entire library expands to almost 6GB and doesn’t have to be installed in your Windows partition if you’d rather keep it elsewhere. On installation, only a small number of presets are usable, but when you email your serial number to Acustica, they will return a unique Nebula3License.exe file that you run to unlock the rest.
The Nebula 3 graphic interface is a vast improvement on that of the Nebula 2 Free version, with a more sophisticated look, and larger, clearer level meters with dB calibration and bigger fader caps. It’s obvious from its ‘hardware rack sampler/synth with LCD menu window’ design what sort of gear inspires the team who created it. From left to right, you get input and output faders with peak-reading meters alongside, then the main LCD window with a status bar across the top and a menu bar across the bottom, plus context-sensitive info in between, and overflow and bypass LEDs beneath it, while the right-hand side contains the eight preset sliders, with a Mini LCD above them displaying their current function and value.
The quickest way to get started is to click on the Fast menu option and choose one of the presets that then fill the central area of the main LCD. You can view all of them in one vast, alphabetically sorted list across switched pages, but mercifully they can also be organised into nine categories, comprising Preamps, Compressors, Equalisers, Filters, Tape, Mics, Amplifiers, TMV (TiMe Variant) and Reverbs.
As soon as the preset finishes loading, any sliders that it uses will have parameter names above them, with any that remain unnamed doing nothing at all. Some effect parameters, such as Rate, Wet, Dry, and compressor time constants are self–explanatory, and most have obvious calibrations such as dB, Hz, ms or percent, but others require a little more explanation.
Slider 8 is always Liqdt, for ‘Liquidity’, which changes the way the individual chunks of sound are mixed together to make the final audio stream. A low value makes the preset sound closer to the original gear, but at the risk of occasional clicks or continuous low-level grating noises, while a higher value lengthens the crossfades between the chunks and reduces the likelihood of audio drop-outs when automating the controls, but may produce phasing artifacts. All of the supplied presets have their Liquidity value selected by ear for the best sound, but a good rule of thumb is to keep Liquidity as low as possible, unless you want it as a special effect or hear unwanted artifacts in the background.
For more detailed information about the currently loaded preset you click on on ‘Prog’. I suspect that few users will ever need to adjust the various kernel engine parameters in the Kern page, or most global settings in the Mast page, many of which are for development troubleshooting only. However, I did find it useful to change the default meter ballistics from RMS to peak. The MIDI page lets you map MIDI controllers to the eight main sliders for automation purposes.
A potential source of confusion for newcomers is that two versions of the plug-in are supplied. One is designed for use with the shorter effects, while the other is ‘optimised’ for longer reverbs, but both look identical, and in fact are identical, except that each declares a different latency to the host (194 and 16,450 samples respectively). The reason for this is that the VST2 standard doesn’t cater for a plug–in whose latency changes from preset to preset. In practice, you can still load long reverb presets into the version of Nebula 3 intended for shorter effects, but their tails will be truncated (which might be useful if you just want early reflections). If you do find yourself mixing up the two versions, it’s possible to load in new ‘skins’ in order to change the appearance of one or both.
I spent a long time both auditioning and examining the audio quality of the Nebula 3 plug-in, and with a few reservations I was very impressed. Compared with Nebula 2 Free, the maximum lengths of the reverbs are no longer truncated to a few seconds, and many presets have been resampled with greater resolution.
Preset names have been changed to protect the innocent, but not enough to prevent ‘gearheads’ from being able to guess at the original gear that’s been sampled; examples include the Boeing 747 compressor, StateOfLogic EQ, and the Strudel A81 tape machine.
There’s a colossal 76 preamp presets on offer, ranging from cheap DI boxes through to expensive analogue console channels, sometimes driven hard to provide analogue distortion effects. Some have almost ruler–straight frequency responses, while others offer characterful bumps and ripples, more extreme low-frequency roll–off, or obvious presence peaks. Some of the analogue and tube models have been captured with varying Drive levels so you can adjust this to taste with a slider. Some even provide separate Original, Odd and Even sliders so you can balance up these relative harmonic contributions.
The effects are often subtle, but I particularly liked the punch of the M5 Roadster Class A preamp on drums, the bite of the 110 Solaire with about 20dB Drive on almost any material, and the tightening effect that the Los Angeles 2A had on drums and guitar. For more obvious tailoring, the range of BEH 12AX7 Tube preamps was very effective, offering variable drive with a variety of instrument EQ and limit settings. I was particularly intrigued by the sound of the NIV Summing Amp, which added a slight intimacy to percussion and guitars, bringing them forward in the mix. This ties in with other people’s opinions of summing amps, although I suspect it may have more to do with the tiny amount of HF boost I measured from about 3kHz rising to a peak of 1.5dB at 18kHz!
All nine of the Amp presets are guitar models with radical EQ tailoring to suit these instruments, ranging from FND Bassman (no prizes for guessing that one!), through models by Peavey and Zoom and more obscure ones such as Kong Kong and Mars Attack. They are all recorded ‘dry’ with no room sound, but offer plenty of pleasing harmonic distortion (mostly third and fifth, but occasionally second as well), adjustable using the Drive slider. I found them all useful for adding radically different tone colours to synths, guitars and snare drums (they don’t suit kick drums because they roll off so much LF), but you can’t use the Drive control much above the halfway mark without encountering nasty distortion break-up.
Like the amps, the small selection of nine mics offers a range of very different tone colours for adding character to existing sounds, with models from ElectronicVox, Goldmann, KappaG, New man and Sure (did you recognise them all?). None will make your recordings really sound as if they used those mics, but I found them great for treating guitars and synths, and, for what it’s worth, I preferred the Goldmann TLM II mic for vocals!
The 13 Tape presets are more of a mixed bag. The Strudel A81 presets offer useful but subtle frequency-response ripples and saturation, but very few ‘Teka 3340-S’ tape decks would have been sold if they really sounded like the two presets on offer here. These nevertheless provide very useful lo-fi thickening effects, and the Sonic Tape selection takes this lo-fi theme even further with some cassette emulations, while the Tape & Bass preset is great for extreme kick–drum sounds. A couple even seem to have some of the test sweep signal embedded in them, giving a strange yet very pleasing level-related ‘thwip’ on each drumbeat. Overall, this section contains plenty of creative and unusual effects, but would benefit from some more straightforward tape-saturation presets.
Currently there are only 12 compressors in the library, but more will apparently be added later on. They cover some tasty DIGI Tube compressors, two ‘enhanced’ Nightmare compressors and prosumer hardware from ‘Yakuzi’ that’s capable of those pumping effects so popular in dance music. Like the tape selection, there are more extreme offerings for special effects: an unusual French compressor, a couple of really weird Fruitjoy models with a mid-range that changes radically with different Ratio settings, plus ‘Do U wanna comp?’ offering Numerical punch compression. I suspect that for most people the highlight will be the ‘Boeing 747’, based on a popular ‘high-end dual-mono Class A compressor’.
Attack and release controls are present on the compressors, as you’d expect, but they are also featured on all of the preamp, amp, tape, and mic presets, where their functions relate to the audio engine. Like the Liquidity control, the lower their value the closer you get to original gear’s sound, but sometimes you need to raise them to avoid unwanted low-level artifacts. For instance, my frequency response and distortion measurements showed that many of the presets in these four categories benefit from a Release setting of 2ms or above, which also allowed me to reduce the Liquidity setting to minimum for more realism. Reducing the Attack time to the minimum 1ms setting also cleaned up the sound of some presets.
In this release I received 39 filters and 29 EQs. There are low-pass, high-pass and band-pass filters, all with a cutoff frequency slider and many with additional resonance control. They are captured from Emu, Moog and SSL gear, among other brands, and there are lots of different flavours on offer, from surgical roll-off through to more characterful four-pole synth filters. Of the latter, I enjoyed the ANT and VIR digital selections, and they seemed reasonably smooth when swept through the range.
I suspect that the EQs will be of more interest to most purchasers, since they encompass quite a few desirable hardware items including StateOfLogic and some really old German and Danish consoles. A few (like the four Fruitjoy offerings) offer fixed-frequency bass, mid, and treble bands in the same preset, but the majority are parametric EQs, so you only get one band per preset. This approach makes sense, since a single EQ band can sometimes take 8 percent of an E6600 2.4GHz dual-core processor, and if you do need several bands on one sound you can cascade multiple Nebula plug-ins, CPU permitting. Personal favourites included ‘2055 Angels EQ’ and ‘4KG to Heaven’.
The version of the library sent to me contained a massive 127 reverb presets, covering hardware such as an EMT 140 plate, Lexicon PCM70, Digitech units, Alesis Quadraverb and Kurzweil 2500, plus various others, addressing a wide range of sonic territory, from short ambiences to rooms, chambers, plates, larger halls and churches.
Despite a little truncation in a few presets, the audio quality proved to be consistently high: this is an excellent reverb collection, particularly for the price, and will win many admirers. However, this is one area where Nebula faces stiff competition, since there are already many convolution–based reverb plug-ins available that provide very realistic static snapshots of acoustic spaces, and Nebula’s CPU overhead was up to three times that of my other convolution reverbs.
This final category is the one that should really make Nebula special, since I know of no other plug-in that can capture hardware effects such as phasing, chorus, flanging, phasing, tremolo and wah-wah. There were 109 presets of this type in my version of the library, covering a huge variety of hardware units, plus specially created numerical effects such as the Ultra WahWah and PhazerWorks collections. Some, like the ‘PongChampion’ panner, are even rate–synchronised to your song.
I was impressed by the PhazerWorks and Ultra WahWah collections, plus the ASR, CSL, JAH and Micro phasers, which all provided lots of character and smooth sweeps, with rates adjustable over a reasonable range. However, others, including many of the DIG, KRZ, QUA, ‘70’ presets and the WWW Numerical flangers, had obvious ‘staircase stepping’ responses that occasionally sounded like chugging steam trains. This has nothing to do with the Nebula engine — I suspect these were early captures with fewer steps that can now be improved on. Personally I’d prefer to lose these and stick with a smaller range of consistently higher quality sounds.
Nebula provides a huge collection of effects, mostly good, some great, some downright weird yet useful, along with a fair few duff ones that ought to be culled to raise the overall standard. Personally, I think the preamp collection alone is worth the asking price, followed by the EQs and the character compressors. Others will particularly relish the filters and flangers, or the reverbs, but there’s plenty for everyone to enjoy.
Nebula 3 is an ambitious product with a few rough edges, but it’s still being updated on a regular basis. Just before I finished this review a new version appeared that lets you explore the inner workings of the kernel engine, such as its LFOs, envelopes, and many other functions, so you can fundamentally modify presets or create new ones. Those who master these controls should be able to create radical new effects such as level-sensitive chorus or cyclic distortion — the tempo-sync’ed ring modulation and flanging of the existing ‘Come To Hell’ preset hints at the possibilities — but it’s a daunting task.
CPU overheads are significantly higher than with most other plug-ins; on my dual–core PC the hungriest ones tend to be linear phase EQs and convolution reverbs, which on average consume around three percent of my total CPU, but a typical Nebula 3 EQ or chorus preset took some five percent, its reverbs took between five and 10 percent, and some of the more detailed amp and preamp models containing higher numbers of kernels could consume up to 20 percent. It’s hardly surprising that Acustica Audio recommend bouncing tracks down once Nebula has added its magic!
Overall, I’m really impressed by the scope and audio quality of Nebula 3. It covers a huge amount of sonic ground, and is incredible value for money. I’m confident that it will sell in large numbers to musicians who have never owned any of the gear being emulated, as well as to many that have.
I’m quite sure the vast majority of Nebula users will stick to the presets, but if you want to capture responses from your own hardware, you can do so using the stand-alone NebulaWdeconvolver utility. This generates a test–tone signal that you pass through the hardware in question via your audio interface outputs and inputs, after first adjusting the interface’s input gain to optimise the returning signal.
Dynamic captures require you to decide on the number of levels and the dB step between them, but fortunately the more in-depth choices, such as kernel lengths, are mostly decided when you select the most appropriate template (dynamic, reverb, time-variant, and so on). Finally, you enter a name and description for your preset and choose a file name for it, click the Run button to generate the test signal and capture the audio result, and then copy the program and vector files into the appropriate Nebula folders. You may need to tweak the Nebula engine parameters to get best results, but the Acustica Audio forums offer lots of help if you get stuck.