ValhallaPlate: On Diffusion

One of the goals for ValhallaPlate was to have a minimal user interface. The goal was to have as many controls as necessary to sculpt beautiful plate reverberation, but no more. Any parameters that were deemed superfluous, or were even on the edge of being so, were removed from the interface.

The Diffusion control has become a standard feature of most reverb plugins.  But ValhallaPlate is the first Valhalla reverb to not have a diffusion control.  Why?  Read on!

Why is Diffusion important in music?

Diffusion is used in reverbs to adjust the initial echo density of the reverb.  The Diffusion control is present in most algorithmic reverbs as a way of dealing with artifacts. A diffusion value that sounds good on vocals may not sound good on drums, and vice versa. By dialing in a given Diffusion value, the user can figure out the right compromise for the input signal. The Size control is often used in conjunction with the Diffusion control, to balance out grain vs. metallic coloration.

How do you make a Diffusion plugin?

In an earlier blog post, I discuss how the Diffusion control commonly works in algorithmic reverbs. The most common technique is to use several short cascaded allpass delays to build up the initial echo density of the reverb, where the Diffusion control adjusts the allpass coefficients of these delays. These cascaded delays give rise to various sonic artifacts, depending on the number of delays, their lengths, and the allpass coefficient settings:

  1. If there are only a few allpass delays, the initial onset of the reverberation will be sparse, but with a natural attack/decay.
  2. Using a larger number of allpass delays will result in a slower attack time for the reverb, that can sound unnatural.
  3. A higher valued allpass coefficient (i.e. somewhere between 0.7 and 0.99) will result in a faster attack and a higher echo density, but will also create a metallic sound.

Why did you leave Diffusion out of Valhalla Plate?

ValhallaPlate doesn’t have a diffusion control, because it doesn’t need one. During the development process, I figured out a new diffusion technique, that resulted in a sharp attack with no grain, while avoiding the metallic sound commonly found with cascaded allpass diffusors. The result is a reverb that works on pretty much any input signal, without any adjustments necessary. The simplicity of the GUI stems from solving issues at the algorithmic level.

The improved diffusion technique used in ValhallaPlate is reflected in the presets available for the plugin. The presets are arranged by decay times: Huge, Large, Medium and Small. There are no specific preset categories for drums, or vocals, because the same ValhallaPlate settings work equally well on both. The presets names in many reverbs are a reflection of “this sounds good on this source,” but this is also a way of saying “yeah, you probably want to use it on this source, as opposed to something else.” ValhallaPlate doesn’t need to be handled that gingerly. Put ValhallaPlate on whatever source you want, and adjust the SIZE and EQ controls to get the sound you want.

I view ValhallaPlate as being closely related to an SM57. Or a hammer. No need to be delicate with the tool. No need to think about things too much. It just works.

Comments (12)

  • nicholas.murphy

    Thanks for the explanation Sean! I was wondering about that. It really makes sense and mimics how a real plate works. Not a whole lot of options on those!

    Are the new algorithms in any way related to ValhallaRoom, which I believe you said had no allpass delays in the early section? You still had a diffusion parameter but it seems to work quite differently from most classic designs (such as used in VVV). Just wondering…I’m sure you don’t want to say too much about your design.

    Thanks again for this and your other recent articles; I’ve learned a lot of what I know about reverb from your blog.

    Nick

    Reply
    • Sean Costello

      The ValhallaPlate algorithms aren’t that closely related to the ValhallaRoom algorithms. I’m using a different technique in ValhallaPlate to generate echo density. I wish I could talk about the techniques used, as I like talking about this stuff, but it wouldn’t make business sense to discuss this stuff in much detail.

      The ValhallaRoom Diffusion parameter would allow the initial echo density to be adjusted, but there are ways of doing this besides allpass delays.

  • You know, I’ve noticed that every new reverb from Valhalla has fewer and fewer parameters. Valhalla Room from 2011 has 19 parameters; Valhalla Vintage Verb from 2012 (late 2014 for the finalized six additional algorithms) has 15 parameters. And Valhalla Plate has 11 parameters. At this rate, Valhalla 2016 will have six parameters (someone will have to come out with new a Eventide 2016 clone in 2016), and Valhalla 2017-2018 will have all of one parameter (a combined decay/time depth, and yes Waves does indeed have a one-knob “Wetter” reverb plugin).

    Personally, I like having a lot of parameters to adjust; one of the great things about an algorithmic reverb is that it can be really adjustable (and considering that the Lexicon 224 did a lot better than the less customizable QRS and Eventide 2016 means the market agreed with me, at least in the 1980s). I love how the Yamaha “Echo Room” algorithm (from the SPX900 and SPX990) does not have just a single “size” parameters, but three adjustable size parameters: Width, Height, Depth (as well as wall variance), and even allows different reverb decay amounts for the width, height, and depth; alas, Yamaha no longer sells a new reverb with that “Echo Room” (the MOXF6 and other keyboards have a simplified form of it called “Space Simulator” which is not quite as flexible).

    Reply
    • Sean Costello

      If I can ever come up with a great reverb algorithm that only requires six parameters, I will totally release it! It might take a while, though. It took me three years to figure out how to create a reverb algorithm that works with only 12 parameters.

      “Je n’ai fait celle-ci plus longue que parce que je n’ai pas eu le loisir de la faire plus courte.”

      It is worth noting that the original Lexicon 224 only had 6 parameters. The Diffusion, Mode Optimization and Decay Optimization parameters were only added at a later date. I would argue that having an “ideal” solution for diffusion (i.e. fast build of echo density, without adding metallic coloration or affecting the attack time) would render the Diffusion parameter unnecessary.

    • Thanks for the correction on the 224. The 2016 did it with nine parameters; the original QRS did it with eight parameters. I’m sure you can do it with seven parameters — just figure out a way to remove standing waves without potential pitch-warping modulation (the devil is in the details). Then we can have mix, predelay, decay, size, brightness (low-cut to flat to high-cut, affects decay), position, and color (continuous: low bit, low bandwidth to high-bit, high-bandwidth).

      If I ever were to design a reverb (like when my child is in my college), I would have predelay (either milliseconds, BPM, and option to be MIDI synced), low frequency decay, mid-frequency decay, high-frequency decay, low frequency crossover, high frequency crossover, width, height, depth, wall angles, diffusion, density, er1 diffusion, er1 level, er1 delay, er1 feedback, er1 pan/position, er1 input pan (left, center, right), er2 diffusion/level/delay/feedback/pan/input pan, tail modulation depth, tail modulation rate, tail modulation shape, stereo blend amount (dual mono to stereo reverb knob), output width (mono to stereo to out-of-phase put in the other channel), color, position, etc. etc. etc. But, then again, I would have to document all of those parameters.

    • nicholas.murphy

      As far as I’m concerned, reverbs don’t even need the MIX parameter. I think a lot of bad-sounding reverbs I’ve heard in (mostly amateur) productions was due to not filtering the frequency extremes out of the reverb send/return. It’s pretty obvious when someone just slapped a reverb on a track as an insert…IME there are frequencies in just about every input signal that you don’t want to send into a reverb. I usually put a HPF at anywhere from 200-600 Hz and a LPF at anywhere from 4-10 kHz before the reverb, which helps immensely. Filtering is one of many things that can be done before/after a reverb when it is used on a send.

  • I’ve been looking for a good mastering reverb for live choral music. Thanks for releasing ValhallaPlate–it’s *perfect*. I can get a really neutral sounding ‘verb that adds just that little depth without conflicting with the natural room sound from the recording. (This use is a great example of where an insert ‘verb makes sense, so thanks for the MIX and H/LPFs.) I find it fascinating how different, yet usable, the different metals (“mode”) are–Brass is my favorite, Unobtainium is amazingly colorless. I haven’t even started exploring new 1.5 modes. 🙂

    Reply
  • Ken Karsh

    I bought UberMod and Plate. Sean, I’m a fan! These are great!!! Thanks!!!!!

    Reply
  • robert.logan

    Thank you for sharing that, Sean. Your plugins sound beautiful and I understand the elegance in reducing parameter options as you have done here. This is an absolutely fantastic plugin for so many scenarios.

    However, I really do think there is a place for detailed control and lots of options – and not necessarily to get conventionally ‘pleasing’ results, but for exploration and sound design purposes. For a simple example relevant here, setting diffusion and density to ‘0’ on certain older Alesis reverb boxes (with a huge decay) – specifically to get that long weird grainy web of delays (the kind of sound we are trying to avoid! ;-)…) Of course there is room for both worlds, but it just seems the move towards simplicity and the reduction of control has become somewhat universal now in plugins, with some developers (not you!) and users exhibiting a real lack of understanding towards those who love to tweak (the reverb you’ve carved being as much the music as anything else.) Reaktor and MaxMSP are great for creating your own reverbs, but it’d be a shame if most effects from most companies continued in that trajectory of being so ‘tailored’ and reduced (but great sounding for certain purposes) so as to render the only options for real tweakability those more deep-end programs.

    Sam Trenholme, I would buy your plugin in a heartbeat! But Sean, I have all of yours, and like I said, they have been such a benefit to music creation. What I wrote above is more in response to a trend I am noticing than to your excellent plugs, which do what they are designed to do exceptionally well.

    Reply
  • George

    I agree with Robert’s take.

    With Plate in particular, I wish there was more control over the stereo/mono aspects of the different modes, independent from whether the mode, and also ability to put the output of all of the plates into the reverb chamber – the way that one of the modes does.

    Having specific control over density and other traditional params, across all plugins, instead of them seeming to be a bit lumped together with other params in some cases, would be good.

    My biggest takeaways after living with Plate and Room for awhile, is that I wish they had the tweakability of the Ubermod demo I discovered later on. Or maybe I’m just the type of user who would have enjoyed crafting reverbs with Ubermod more to begin with.

    The sound engines themselves are very nice.

    Reply
  • RAFAEL

    I’ve only ever used stock plug ins when recording with a daw in the past. I usually play live electronic music with hardware until recently. I’ve been struggling to find a reverb plugin which matches the smooth and lush aesthetic I hear in my head. Valhalla Plate is the first plugin I’ve ever purchased! Smooth and lush for days…

    Reply
  • Craig Freeman

    Yeah, I gotta say, after 35 of doing this, Valhalla Plate is the best ‘verb I’ve ever heard. Yes I heard Lexicons in the ’80s.

    Reply

Leave a Reply

Your email address will not be published. Required fields are marked *