epicVerb 1.5 released

[susiwong]

Will there be an epicverb version with more parameters? As far as i can hear, it sound like hall algorithm with a medium to large size room. Does the room size shrink when minimizing the reverb time?

Perceived room size is influenced to a large degree by the ER controls, this thing is a lot deeper than it seems at first. The presets are a good starting point for your own experiments.
What I'd recommend is looping a single percussive sound like a sidestick with 2 or 3 sec silence inbetween, that way you can hear changes quite well. That's also handy if you want to approximate other reverbs, I've been able to get really close to the UA Plate 140 for example, widely seen as one of the best reverbs of its kind anywhere.
Ymmv,
susiwong

[MaliceX]

Besides, this is a really dope reverb!!!

To quote Wikipedia: A dope, a person who is stupid.

Now I thought EpicVerb was surprisingly good.

Dope in the context of "addiction". For good reason too!

[bootsie]

Will there be an epicverb version with more parameters? As far as i can hear, it sound like hall algorithm with a medium to large size room. Does the room size shrink when minimizing the reverb time?

Besides, this is a really dope reverb!!!

The reverb time parameter affects technically speaking the delay line filtering in the feedback loop, whereas the "size" is actually affected by the the different ER models and ER shaping.

On my own I'm not that much into the more parameter thing but more interested in "no-compromise" algorithms when thinking about potential future developments of epicVerb. All native reverbs today are still containing significant compromises to keep the CPU low, whereas HW or DSP based reverbs can take advantage of brute force methods such as 1000+ delay line computations and the like.

Which leads to the million dollar question to you: Would such a native "no-compromise" reverb be appreciated even if it consumes let's say a whole single core of a modern cpu or maybe 50% of the whole system or so? Where is the acceptance threshold level (cpu consumption wise) for such a device? Or is low cpu still the most important requirement?

I'm really interested in your opinion on that ...

nice weekend,
Bootsy

[Lotuzia]

Will there be an epicverb version with more parameters? As far as i can hear, it sound like hall algorithm with a medium to large size room. Does the room size shrink when minimizing the reverb time?

Besides, this is a really dope reverb!!!

The reverb time parameter affects technically speaking the delay line filtering in the feedback loop, whereas the "size" is actually affected by the the different ER models and ER shaping.

On my own I'm not that much into the more parameter thing but more interested in "no-compromise" algorithms when thinking about potential future developments of epicVerb. All native reverbs today are still containing significant compromises to keep the CPU low, whereas HW or DSP based reverbs can take advantage of brute force methods such as 1000+ delay line computations and the like.

Which leads to the million dollar question to you: Would such a native "no-compromise" reverb be appreciated even if it consumes let's say a whole single core of a modern cpu or maybe 50% of the whole system or so? Where is the acceptance threshold level (cpu consumption wise) for such a device? Or is low cpu still the most important requirement?

I'm really interested in your opinion on that ...

nice weekend,
Bootsy

Very very basic answer I'm afraid :

Implementing several "quality modes" with according Cpu gluttony

So that people can choose best quality if their comp allow it, or for rendering/freezing, then avarage or low quality to insert in a project and real time tweaks/automation.

Well, just the first idea wich crossed my head

LtZ

[susiwong]

Would such a native "no-compromise" reverb be appreciated even if it consumes let's say a whole single core of a modern cpu or maybe 50% of the whole system or so? Where is the acceptance threshold level (cpu consumption wise) for such a device?

YES !
though this will vary extremely from user to user.
Lotuzia gave one sensible answer, if that's an option technically.
Personally the sweet spot would be around half of a current 3 GHz core so I could run two main reverbs without rendering. Seeing that epicVerb currently hardly registers on the ASIO meter that should be a considerable increase in power, about 5x to 10x.
I still find it hard to believe a 480L from the '80s for example has that kind of computing power, even taking into account a specialized Lexichip might be more efficient at these tasks. But what do I know ?
Another option could possibly be Cuda.
And even if the best quality could only be achieved with offline rendering I'd gladly do that, provided there's a realtime preview mode.

Ymmv,
susiwong

[Christian Schüler]

Did you actually listen to some 480L impulses?
They're quite grainy and the echo and mode density is sub-par by today's standards.

The computing power of that device should be below 1% of a current processor core, no more than 30 MFlop/s (that's million floating point operations per second). By comparison, a single core of a current processor can do more than 4 GFlop/s!

[susiwong]

Did you actually listen to some 480L impulses?
They're quite grainy and the echo and mode density is sub-par by today's standards.

The computing power of that device should be below 1% of a current processor core, no more than 30 MFlop/s (that's million floating point operations per second). By comparison, a single core of a current processor can do more than 4 GFlop/s!

Impulses and quite some experience with real 480s, R-880 and TC 6000, yes.
Impulses, even great ones, come close but can't duplicate the effortless way the real things sit in a mix, especially the 480 still would be my first choice for vocals, sax, guitar, snare etc, the graininess (which is even more obvious in the 224XL, another great verb apart from the mediocre noise floor) is just part of the character imho, ymmv.
Unfortunately there's the price for these units, still hope to find a used R-880 at a decent price one day ...
Cheers,
susiwong

[Christian Schüler]

There you go.
Careful engineering goes a long way ahead of brute processing power. Think of that for hardware devices, every hour spent on engineering and optimizing the DSP code means the product can run on cheaper hardware, amortizing cost in sales. The hardware devices are the least likely to contain a brute force approach, because they cannot afford it. Its the a similar situation with PC games vs console games.

[susiwong]

There you go.
Careful engineering goes a long way ahead of brute processing power. Think of that for hardware devices, every hour spent on engineering and optimizing the DSP code means the product can run on cheaper hardware, amortizing cost in sales. The hardware devices are the least likely to contain a brute force approach, because they cannot afford it. Its the a similar situation with PC games vs console games.

Very true imho, but that's only talking from common sense, I'm no programmer.
What I'd hope for from Bootsie if he would consider such a big task is mainly raising the quality bar even further, CPU efficiency is far less important in this case for me. And he is a detail freak with very good ears, so ...
And I'm not talking about 1:1 emulation, epicVerb's flavour already is so much up my alley it's become my main verb even though I own Plate 140, CSR and a few other good commercial and free ones, if it can be improved even further at the cost of CPU I'm all for it.
I'm a fool for reverbs, if I hadn't been burned by TC's pathetic driver support before (and if they had x64 drivers at all ) I'd probably have added the VSS3 and DVR to my collection long ago.
Ymmv,
susiwong

[mabian]

So, if I understand correctly, it seems we already have in EpicVerb something that does far more calculations that the legendary 480L and company did?

Does Epicverb really need improvements?

- Mario

[Christian Schüler]

And I'm not talking about 1:1 emulation, epicVerb's flavour already is so much up my alley it's become my main verb even though I own Plate 140, ...

If I may also say so, you also just gave proof in how subjective these reverb issues are. The Lexicon sound is in no way a special flower, it's just been around for long enough to establish itself. But could be replaced with anything equivalent for the same amount of satisfaction.

[susiwong]

Does Epicverb really need improvements?

wrong question
epicverb is among the best ITB verbs atm imho.
The abovementioned hardware verbs still are considerably better than anything ITB, again imho.
So improvements are certainly possible.
The real question is, would Bootsie take on such a monstrous task (we're talking hundreds of hours probably), and what would be the CPU load then ?
The plug should run decently on a current Intel C2D with 2.5 to 3 GHz which probably is the mainstream atm imho.
Cheers,
susiwong

[filterfilter]

Will there be an epicverb version with more parameters? As far as i can hear, it sound like hall algorithm with a medium to large size room. Does the room size shrink when minimizing the reverb time?

Besides, this is a really dope reverb!!!

The reverb time parameter affects technically speaking the delay line filtering in the feedback loop, whereas the "size" is actually affected by the the different ER models and ER shaping.

On my own I'm not that much into the more parameter thing but more interested in "no-compromise" algorithms when thinking about potential future developments of epicVerb. All native reverbs today are still containing significant compromises to keep the CPU low, whereas HW or DSP based reverbs can take advantage of brute force methods such as 1000+ delay line computations and the like.

Which leads to the million dollar question to you: Would such a native "no-compromise" reverb be appreciated even if it consumes let's say a whole single core of a modern cpu or maybe 50% of the whole system or so? Where is the acceptance threshold level (cpu consumption wise) for such a device? Or is low cpu still the most important requirement?

I'm really interested in your opinion on that ...

nice weekend,
Bootsy

Definitely!! There are so many average plugins! What we really need is sound quality! I think native plugins are becoming very good these days...look at for example Waldorf Largo, Softube FET-Compressor or the SPL EQ Rangers. A high-quality-reverb would definitely be appreciated!!

[susiwong]

And I'm not talking about 1:1 emulation, epicVerb's flavour already is so much up my alley it's become my main verb even though I own Plate 140, ...

If I may also say so, you also just gave proof in how subjective these reverb issues are. The Lexicon sound is in no way a special flower, it's just been around for long enough to establish itself. But could be replaced with anything equivalent for the same amount of satisfaction.

We certainly agree here.
It's about acquired taste and also musical genre more than anything.
Personally the 480 and R-880 are my favourites, I know lots of folks who would prefer a TC 6000 or maybe Bricasti, fine !
You can do a fine record with a Strat or a Gibson, still most guitar players prefer one or the other.
Cheers,
susiwong

[bootsie]

Did you actually listen to some 480L impulses?
They're quite grainy and the echo and mode density is sub-par by today's standards.

The computing power of that device should be below 1% of a current processor core, no more than 30 MFlop/s (that's million floating point operations per second). By comparison, a single core of a current processor can do more than 4 GFlop/s!

I actually wasn't refering to one of those dinosaurs but to the todays generation of quality highend outboard reverbs. And this generation is a whole different story concerning dsp and memory architecture and the demand for it. E.g. just have a look in this thread over at gearslutz where the developer of the Bricasti unit talks a little about it (other talks elsewehere in that forum or on the net) and you get the picture what I'm talking about: http://www.gearslutz.com/board/3647089-post17.html

Don't get me wrong: I'm a strong believer in clever invention and to do that first but some stuff can't be "optimized away" and needs serious computational ressources.

To all contributors in this thread here: please do me a favour and don't repeat those silly discussions about specific ancient devices and taste and which is better or not and such a like. It was already here. No need to do it again. Not in this thread. Ok? Do it in another thread if you are in the need

peace,
Bootsy