Ooooh... Charts and Graphs Redux

[laxu]

Test of one of the new products on the market. Green trace is the Axe-Fx III. Blue trace is new product. 5153 amp model. Gain at 6. BMT at noon.

View attachment 129200

This also shows the improvement in aliasing due to the new triode algorithm in the Axe-Fx III. The aliasing is so low now that it's almost in the noise floor. The hum from the virtual power supply is higher than the aliasing components.

What do the FM3 and FM9 look like in relation to this, since they are more comparable products? I know they don't have the new triode algo yet but mainly wondering if they perform the same as the Axe-Fx 3 or not.

 

[JJunkie]

This is an impressive tone buzzword to post ratio

i originally had all 7 of the bad system dwarves in there, but had to edit them out or else it would be the forum equivalent of my marshall goes to 18

 

[IronSean]

What do the FM3 and FM9 look like in relation to this, since they are more comparable products? I know they don't have the new triode algo yet but mainly wondering if they perform the same as the Axe-Fx 3 or not.

If you go back a few pages there was a graph of the Axe FX against other competitors. When that amp block got ported to the FM series they would have matched it. Currently they're likely somewhere between there and here until the triode algorithm is ported.

The amp block modelling is the same to my knowledge, so the aliasing should also be the same.

 

[unix-guy]

If you go back a few pages there was a graph of the Axe FX against other competitors. When that amp block got ported to the FM series they would have matched it. Currently they're likely somewhere between there and here until the triode algorithm is ported.

The amp block modelling is the same to my knowledge, so the aliasing should also be the same.

I'm not sure that's true. The aliasing level is related to oversampling rate, from what I recall in the past and that can be affected by the hardware power.

If I'm wrong, I'm sure Cliff will Batman slap me!

 

[James Freeman]

If you have enough power you can also apply a brickwall filter after every stage that generates harmonics, ie. every triode.

 

[Greg Ferguson]

I'm not sure that's true. The aliasing level is related to oversampling rate, from what I recall in the past and that can be affected by the hardware power.

If I'm wrong, I'm sure Cliff will Batman slap me!

The FX3 has oversampling super-powers because it has the big south-forty to play in…



The Wiki tells us…

(Axe-Fx III) "The reason I haven't added a third amp block is that I would have to reduce the oversample rate for all the amp blocks when three are in use and this would be detrimental to sound quality." [8]

(FM3) "We could potentially do two amp blocks but at reduced quality and I don't want to do that. Part of the problem with other modelers is that they don't oversample enough (and use single-precision in places where you need double-precision). Then you get complaints of artifacts and ear fatigue and all the other things associated with inadequate sample rate and word length. The vast majority of users only use one amp block so we wanted to make something with one very high quality "Ares" amp block." [9]

That leads me to believe that if the Oversampling Mode is not "Best Quality" but instead is "Min. Latency", it's equivalent to what the FM units have. And, at that point, they're generating the same output.

---

PS -
Oversampling Mode was added in Cygnus 19.02…

• Added “Oversampling Mode” to Global menu. This parameter allows selecting between BEST PERFORMANCE and MIN. LATENCY.

but then further into the thread Cliff said…

There's no difference in CPU usage. Oversample Mode selects between different interpolation/decimation filter kernels.

So maybe he'll chime in. Or not. He has to maintain some mystery (AKA proprietary information).

 

[GlennO]

I'm not sure that's true. The aliasing level is related to oversampling rate, from what I recall in the past and that can be affected by the hardware power.

If I'm wrong, I'm sure Cliff will Batman slap me!

I don't think he has ever said if the modeling is the exactly the same on the different models, but yes, aliasing is primarily determined by the sample rate (48KHz for Fractal) and oversampling, and better specs for both of those requires more cpu power.

I'd hazard a guess the oversampling is the same on the different units, but anyone who has both an FM3/FM9 and an Axe-FX III could do an aliasing comparison and see if there's a difference (using an appropriate firmware version to exclude the effect of the recent triode change).

 

[DLC86]

Already done:
https://thegearforum.com/threads/modelers-and-aliasing.1815/post-59756

FM9 and FM3 are slightly worse in aliasing performance than the axe fx III, but still a lot better than that new product for sure.

And no, setting oversampling mode to "min. latency" doesn't put the aliasing of the axe fx at the same level as the FM3/9, it's just 2-3 dB higher than "best quality"

 

[FractalAudio]

If you have enough power you can also apply a brickwall filter after every stage that generates harmonics, ie. every triode.

Two reasons you don't want to do this:
1. It adds latency (or phase distortion if you use a minimum-phase filter).
2. Those high frequencies actually interact if they are distorted in subsequent stages. Filtering them out removes that interaction.

 

[randocaster]

Two reasons you don't want to do this:
1. It adds latency (or phase distortion if you use a minimum-phase filter).
2. Those high frequencies actually interact if they are distorted in subsequent stages. Filtering them out removes that interaction.

Just use your sledge. Look at the horror on the squirrels faces.

 

[rr32btg]

Test of one of the new products on the market. Green trace is the Axe-Fx III. Blue trace is new product. 5153 amp model. Gain at 6. BMT at noon.

View attachment 129200

This also shows the improvement in aliasing due to the new triode algorithm in the Axe-Fx III. The aliasing is so low now that it's almost in the noise floor. The hum from the virtual power supply is higher than the aliasing components.

1. Will the aliasing level be different when using two Amp blocks? There was rumors that with two amp blocks there are some trade offs
2. Will the aliasing level be different with different values of Global Settings -> Oversampling Mode parameter?