fremen_mu a écrit :
totoleheron a écrit :
C'est cela oui.
Ben oui c'est cela, c'est pas parce qu'on n'en a pas parlé ici que certains ne le savent pas depuis l'an dernier. Je ne suis pas du tout électronicien, donc désolé pour les termes approximatifs, mais voilà, de mémoire, ce que j'avais compris à l'époque (facilement 2 ou 3 mois de cela) : le Kemper dispose d'un "modèle" d'ampli à lampes de base "idéal", et quand il profile un ampli existant, il mesure les différences d'avec cet ampli "idéal" là, et les applique. Et maintenant retourne souder
Concernant la vidéo d'Ola : je préfère celle où il compare l'Axe II et le POD HD - et je dis pas ça pour descendre le Kemper puisque je l'aime bien le Kemper moi
C'est cela oui. Heureusement que tu suis sinon on serait perdu surtout pour dire des âneries.
J'ai retrouvé le passage en question où l'on a appris avec certitude comment fonctionnait le kemper. Il y a bien eu un article un mois plus tôt mais très ambiguë. J'ai eu du mal à retrouver le passage car ca n'est pas hoss mais berlinchris qui a laché le morceau après avoir rencontré CK au guitar center de hambourg(ou au musikmesse?). C'était début septembre 2011, il y a 5 mois :
http://www.thegearpage.net/boa(...)02552
berlinchris
Citation:
I had a brief chat with Mr. Kemper yesterday and he told me that that they did check out some digital modellers. With some they were able to create a profile, with others they were not. Now why is that?
First, let´s take a look at how this thing works in general. The KPA has a built-in tube-amp. Or better said, some kind of digital simulation of a tube-amp with very, very good audio-data. This is what I was told in person. And this I guess is their secret recipe. They are using very good algorithms and from what I understand this was by far the hardest part to come up with.
Now, when profiling via those weird test signals, the KPA compares the "ideal" data, parameters, curves, values etc. inside with the returning data from the actual amp/cab/mic/console-chain. There you also have the reason why a profile is only 10kB small and why the unit can store thousands of Profiles. In the end it´s just a few values that are changed to create the sound of the actual amp. Sound incredibly simple, doesn´t it?
Small edit: profiling is happening in mono only (!) and without efx. Amps with external distortion pedals are ok, but the rest won´t work. You can not profile your full sound with stereo-spread, harmonizer, time-based-exf etc...
moi
Citation:
So why they don't do the same for all amps if they know how to collect accurate datas?
It's probably just reference IRs coming from another amp, this would explain the clean and crunch modes for the capture. Then they deconvolve both signals to get the difference. It makes sense even if i don't know what they do after.
Puis quelques jours plus tard sur le forum fractal:
Citation:
Quote Originally Posted by FractalAudio View Post
This isn't a prelude to profiling. As I've stated multiple times before, profiling has a plethora of shortcomings. You can't treat an amp as a black box. There are fundamental mathematical reasons why the model created when doing this is inferior to a physical model.
You are right even if the kemper claims to work in a different way.
Christophe kemper said at the musikmesse that in fact each tone is based on clean and crunch amp data carrefuly collected. (the question is "how it is collected?", IMO it's always IRs)
So the profiling process just gives the difference for each real amp.
Why not using the same concept but with your components modelling technology?
Let's take for example the FAS clean for all the clean amp models to profile then send a bunch of sine sweeps to your real amp, get the result and cook it with the FAS clean to get an approximativ diff.(deconvolution)
In this case the block which will render the result will use the FAS clean model with component modelling and will be convolued by your bunch of IRs.
Same for the crunch amp with a generic crunch model etc...
So it's the same concept than the kemper but using your component modelling technology acting as the tone core and which would take care of non linearities this time.
A new snapshot block working in this way would be...perhaps(probably)...interesting.
