Jason Sadites' video claiming (proving) that an IR is nothing but just a static eq curve

Status
Not open for further replies.

luckymethod

Member
Messages
500

What do you think?
He is wrong. EQs are a special case of an IR, but IRs capture the evolution of that "filter" over time while EQ has just "one point" and that's it. Can you make an EQ sound similar to an IR? Yes, but it's not gonna be exactly the same. EQs are cheaper to process, so if the loss isn't noticeable, then you can spare some processing by using an EQ instead of an IR.

Point is, THEY ARE NOT THE SAME THING, but they can be swapped for each other if need be.

edit: I actually didn't watch the video, so I was replying to the message in the thread. It's pretty obvious Jason knows the difference so usual caveats apply.
 
Last edited:

mehegama

Member
Messages
246
The question is if one takes this eq curve and use another amp with its own characteristics, will the sound be the same as using the IR there? I.e. is it a matter of an overfitting model with bad predictive power?
Maybe someone more experienced might be able to help.
 

Jim Roseberry

Member
Messages
1,527
He is wrong. EQs are a special case of an IR, but IRs capture the evolution of that "filter" over time while EQ has just "one point" and that's it.
A 1024-sample Cab IR (48k) is equal to 21.33ms
A 2048-sample Cab IR is equal to 42.67ms
Double that for 4096-sample Cab IRs (not supported by many hardware modelers)... and you're still well under 100ms.
Not 100% static... but nowhere close to being dynamic over the duration of a typical guitar note/phrase.
 
Last edited:

DieSchmalle

Member
Messages
148
A 1024-sample Cab IR is equal to 21.33ms
A 2048-sample Cab IR is equal to 42.67ms
Double that for 4096-sample Cab IRs (not supported by many hardware modelers)... and you're still well under 100ms.
Not 100% static... but nowhere close to being dynamic over the duration of a typical guitar note/phrase.
IR's are linear time invariant functions. A dynamic effect like a compressor is timevariant - it's current behavior depends on it's previous behavior. That is not the case with IR's - Jason calls that static.
 

Gearzilla

Member
Messages
4,609
“The difference between an impulse response and a set of simple electrical (or virtual) filters is largely abstract. Any collection of filters will have its own impulse response, and one way of implementing that set of filters is via the same process that the Axe-Fx uses to implement cab sims, which is called convolution. In fact, all digital filtering processes involve some form of convolution.

With simple EQ-type filters, the total number of taps (points) required, and the associated processing burden, can be relatively small. This makes this type of filter easy to implement, but the price you pay is that the level of detail that can be introduced in the resulting frequency response is small. This makes it essentially impossible to successfully emulate the behavior of a guitar cab with simple EQ.

Using real-time convolution with an impulse response ("IR"), OTOH, has the possibility of duplicating all the response detail that your hearing is capable of recognizing. The price you pay for this is that, to achieve satisfactory results, the length of the IR must be sufficient. Adding more points creates a higher burden on the cpu. The 1024-point length of the Axe-Fx IRs is sufficient to capture the audible detail in the response of a guitar cab.

When you compare an IR-based cab sim to an EQ-based attempt to emulate the same cab, the IR - assuming it is long enough - will win every time. This is because much of the response of a guitar cab is due to multipath interference effects, which real-time convolution can duplicate and simple EQ cannot. Neither approach is "dynamic," however. Both processes are linear.”

—Jay Mitchell
 

BigBadOrange

Silver Supporting Member
Messages
1,044
Hmmm, as a hard rock up to high gain player, I though modeling was absolute garbage until IRs came out. Then it was like "oh, ok this might really work". Then it seemed every manufacturer sounded better and better. But turn off the IRs and even modern modelers sound like garbage, I admit I have never tried to use an eq, but the sound is so awful I can't imagine the effort being worth it. Granted, as a non clean player, cleans always sounded fine to me, even in the early days of modeling. Heck, Im ok with pluggin in direct for cleans. :p
 
Status
Not open for further replies.




Trending Topics

Top