MFreeformEqualizer Morphing Shapes?

[vectorwarrior]

So, i was hoping to morph the EQ shape in MFreeformEQ between two or more shapes. Then modulating that. Unfortunately I can't seem to find a way to do it. Firstly the standalone plug doesn't have any MPs or modulators. So I tried doing it in MXXX, switching FreeformEQ out of 'drawing' mode, selecting all the points, then MP learning and trying to slightly move all the points so they are added to the MP. Unfortunately this only adds the first 32 points to the MP... and there are a LOT of points.

Can anyone think of a way of doing this? I could just set up two (or more) different Freeforms in parallel and crossfade between them, but obviously crossfading volume isn't the same as truly morphing the EQ shape. It's the morphing itself that I find interesting.

Ultimately, my number one dream would be to have the ability to constantly be auto-EQ'ing a signal, so it's constantly updating the EQ curve to match the input against the side-chain. I've asked Vojtech for that, but he's obviously pretty busy. That would be a sweeeeet alternative to something like MMorph. But maybe that's a different thread

[sirmonkey]

My first impression is that your idea to use cross-fading would work fine. You do have me thinking, though.
First, with a freeform graph, you can't automate a Q setting, band shape/width, gain/attenuation... etc.
Since you can draw any arbitrary shape, the EQ curve isn't based on discreet bands that are specified in the usual way. So, exactly how the morphing would occur between 2 different shapes would be complicated, and also very subjective as to how that would be implemented.
However, cross-fading seems as perfect as one can get. At least that's what I think so far.
I figure that if you have 2 EQ settings gain matched, and choose a transfer curve in MRatio that you like, you would effectively have the closest thing to true morphing, if there is any 100% objective definition of that.

[jmg8]

Try FreeformAnalogEQ. with the maximum Q setting at 20 and bands at 32, it can be very detailed. you can learn the 32 bands with a MP, then morph between banks.

[jmg8]

And, if you need the auto features, you can create shapes in FreeformEQ and copy/paste into FreeformAnalogEQ.

[vectorwarrior]

Yeah, I tried using the analog eqs but the problem with the analog eqs is that they have different types, so when morphing from one shape to another they don't smoothly glide, they pop when the eq shape changes e.g. when a peak eq changes to shelving eq

[DarkStar]

Will this do what you want? It may depend on how many nodes / points you need.

In FreeEq and draw your first curve
Switch to slot B
Draw your second curve
Then open the Morph button and drag accordingly

As in:

Morph button lets you morph between the A, B, C and D settings. Morphing only affects those parameters that can be automated or modulated; that does include most of the parameters however. When you click this button, an X/Y graph is shown allowing you to drag the position indicator to any position between the letters A, B, C and D. The closer you drag the indicator to one of the letters, the closer the actual settings are to that preset.

Please note that this will overwrite and change the preset that is currently selected, so it is best to select a new preset e.g. 'E', then use the morphing method. This way you will define the settings for A, B,C and D, morph between them, and store the result in 'E' without any modification of the original A, B, C and D presets.

Please note that the ABCD morphing itself cannot be automated and that, while morphing, the changes to the underlying parameters are not notified to the host (there may be hundreds of change events).

[DarkStar]

Yep, forget my idea. This is the way to go:

Try FreeformAnalogEQ. with the maximum Q setting at 20 and bands at 32, it can be very detailed. you can learn the 32 bands with a MP, then morph between banks.

And if you need more, how about MXXX, with a FreeformEq on each band, with an MP targetting the appropriate points? I do not know how many target parameters an MP can have, but you could have 1 MP for each band and a "Super" MP controlling those MPs.

Concept, in MXXX:

Code: Select all

$eNrtWFtz2jgUfs+v0Gj3LQUkWb7N2O4AgTZpaLwhm9C8KViAEyM5sgiXX79jcwlQw3Q73czsNMOLOdI5+s7t08X7OBsn4IWrLJbCh7iKIOCiL6NYDH040YOKAz8GJ16n1+tlXOtYDDPwhw8xBA0moqacCO1DA4KmFFrJJOEqawn2kPDIhwiCppJZJl+46siI+5DQqFGTfb01cDNPuQ-rgiVyCMGZmt9xXR8MeF9n5yKd6E8sFsV6rSjWUnXjBfchNl30wbKMbWmLZXMf2qb7wTYxBJfxONbIh6RqIIyI67qmZa3E2IdGlbi2bWID2ZjmYvFUAL687qZJrDUvbGIfYmS4+2LiQ5JLO2x22b5qJLL-tMRlEAg6jR0nriZ624sO11xlTZkkLM141NVccbkcmSQ6Dpli490pBazbdYYck7gugYG3CWAeuvmCKwK6Yyn1SPAs8yGqIhOCu1hEcroM8WcmBAy8UPFBnGiuYjEMvFDGQoN2woZZw4eYYghmeRlsx6x2aBrdn1bbNV4PEdgUA4Yg5OwpK766IzltykSqW5ZMeCHL1b-3KTjxHpiI0KrmSgPe5Qnvax6FyWSYRxl9yOOz-JetP0B7kiRf2Zj7sK04H0g1bj1PWBIvuAIYgv0hCJaJOxcRn2EfVjAEvSIV33z4ar+72xNl+AKvrjjL8rF4HC-4tZyuHR4qlo4ORXc3CWDuwwqqIoKI45SlBK2ULJMatoMxcjdKxCCYHFVysWMQ6rim466VDNuxTXJYiVSxadnEodQwX1dyHYyPKhkIU9ugJrXoxiejTOFQia1itq4QbPz-y-7Vl8CrT7R8LcyunKg+L8azODvjg1isCOGGqSHXJSOBV49YqmMp2oo-T7jozzuxqJWJ2awWeLtLtMQLT2TKD-leUpQElYVqle-DQz+c9GIRWixSOwS2thO2wLuXcpw3XN6xVYsuk7TTsBvB5iMXFVyz4hx8pKd-D84pSdC6jYlrYRO71HKcIkGoipDhGkd1qIUxdizHphu+wK5tHOSLZV-3cOFlYcFB2HJN7FBsbCw4ru0espCXWRVZhFIXWXRDOMQkln1Uh7iW6ZoUmU6pR2Vl+s5N79xUCvYoNzn-iprwiprIkdb-fY5DqIpwWaJ+sInX0yhyLdum1LTtzdkHU4KOHJiMquWaNnKoY2-OS9gktnVUxyWOSSg1CX1lL2ITo7QQy6rtnWLeKaYU7D7F-DCh5PervXtnsCdA+UX-aXXRJxB8+vt8eede3ugb8bCQddk4b+UiOasRJp6yNZN0+JB1pEpHEHRTvkpidxrr-ugmzifkd9y8KooaNqzVv6WlkKs+FzrL63gNC7Rm+npZQwWXXTMx5D6kEGymLFfOnykABhVQwnOgArojlnJQActUEPDtFQcM-rytU52lxgteVK4WneL3FE5936ttVvnPMRl7mEiYnDI1Q2jhnKvGjN2fvQzeGhPdw6TD0MWfH1P86NyTSo5v0fpVmMhPxklcnYrZU2aQm9Mvf92dJah5+fjWmPbjFNZVTxp1lzxWstS5v-xKxfVbYzL3MPUu04uz+rnCi0pz-txu65cw+VWYjJ-M3UPjy9m08zzCN6dOdNH51mg-8LfGtJ+7x4to0pJhnTyedu3etPL11o7fGtN+7uJWe2hG9ZgsKvZg-OxEF6H8DlNtj9C-k2TBiVfv6-iFh4pnXBePqjnywGsyzYdSzc81H5+LgawFXnOiFBd6Rez5vlQPl3rFZr5lJzerUnDNdP70C2v5ZlOsRwLvLo70qLPef8H2ESJ3LTe1nlJbK514ta136Cw4+QfaJPvW

[DarkStar]

Any good?

[vectorwarrior]

Any good? Not really I'm afraid, I already discussed the issues with both options above, namely that analog freeform won't morph well because the eq types will switch during the morph, and that there aren't enough MPs to control all the points when using the real freeform. The best option is actually the ABCD morph, but as you know, that can't be automated or modulated.

Oh well, thanks anyway!

[jmg8]

You only need 1 MP in banks mode to control all the points though. Or maybe I am missing something?

[jmg8]

Also, the analog freeform issue might have a solution. Could you add the shape parameter for all the bands to the "Lock" set them all to peak. This way they will not change type when morphing.

[vectorwarrior]

You only need 1 MP in banks mode to control all the points though. Or maybe I am missing something?

As I explained in the original post, 1 MP isn't anywhere near enough to cover all the points. It can only store 32 points and it looks to be in the hundreds for a full spectrum graph.

As for using lock, that would work, BUT the original purpose is to use the auto-EQ to match one spectrum to another. There is no way to force the EQ matcher to only use peak filters, so it's going to attempt to use shelving filters in the match. If i lock those shelving filters to peak, it'll sound totally different. So yeah, I can lock it, but it then it wont be the actual EQ curve that i want.

Ha ha, it sounds like I'm being super picky here, but the gist is, I can't morph between two different EQ-matched curves. Maybe I should have been clearer about that from the beginning

[jmg8]

As I explained in the original post, 1 MP isn't anywhere near enough to cover all the points. It can only store 32 points and it looks to be in the hundreds for a full spectrum graph.


As for using lock, that would work, BUT the original purpose is to use the auto-EQ to match one spectrum to another. There is no way to force the EQ matcher to only use peak filters, so it's going to attempt to use shelving filters in the match. If i lock those shelving filters to peak, it'll sound totally different. So yeah, I can lock it, but it then it wont be the actual EQ curve that i want.

Ha ha, it sounds like I'm being super picky here, but the gist is, I can't morph between two different EQ-matched curves. Maybe I should have been clearer about that from the beginning

The reason I wrote about MP, is that you said 1MP isn't enough. But you are wrong here, as far as I know, the MP can have an unlimited number of target parameters. The limitation is not with the MPs but rather the EQ itself only has the first 32 points available as targets. So that's why the crossed wires happened here.

As for the lock trick, I just tried it and it is really quite accurate, even with only peak filters. I managed to get the shapes very close using 32 bands, Q=20, Mode=Default.
But how accurate do you need it to be? I analysed some audio using free from EQ, with 5% smoothing, then copied it over to free form analog EQ and it almost exactly copies. However if you need the smoothness lower than 5%, I can see how it might get less accurate.
If you haven't tried it, I think you should, as it might just do the job.

But you have me thinking now of other ways to achieve this.
Does the ratio in convolution interpolate between two IRs, or simply crossfade?
If the former, you could use freeform to create 2 IRs and then use convolution to morph between the two. Just an idea?......

[vectorwarrior]

Ah, I see your point about the point limit in the EQ being the issue with freeform EQ.

When you say 'the lock trick' can you clarify please? In freeform analog I am not seeing any settings to lock anything in either standalone or MXXX. The only thing I'm aware of is the parameter lock, but there is no way to force it to peak. In fact, in freeform analog, you can't really 'see' what EQ's are being used, it's all hidden behind the scenes and you only get to control the input curve.

What am I missing here? So, I'd use Freeform EQ to match two signals, and then copy/paste the curve into FreeformAnalogEQ. Do that multiple times and 'save' the curves into banks of a MP... but it's this last section that I'm drawing a blank on. Sorry for the questions, it sounds like you may be on to something here.

[vectorwarrior]

Oh, one minor bug this has highlighted. The ABCD morphing doesn't work correctly with MFreeformAnalogEQ. It changes the shape of the destination curve (the green one) but doesn't actually create a final output curve.