Overloud announce upcoming new IR technology for their TH-U hoping to take things to the next level.

Details;


Fluid Convolution is a new generation DSP Technology for IR processing. It will be the technology behind the upcoming Overloud cabinets and IR processing products.


What is Fluid Convolution?
Convolution is the mathematical process used to apply an IR to an audio stream.
Fluid Convolution allows us to remove the digital artifacts of conventional IR processing, seamlessly switch between different IRs, and also do this with a lower CPU impact.


How does it work?
When the recorded audio sample rate differs from the IR sample rate, conventional convolution requires resampling of the audio in order to match the two sample rates. This is particularly true when oversampling is used, like in many guitar processors. This leads to an increased CPU load and some audio degradation due to the resampling process.
With the Fluid Convolution we store the IRs in the complex frequency domain, so no resample is needed anymore. The results are a more natural feel and a lower CPU impact.

The transition between IRs also happens seamlessly: when changing IRs or tweaking cabinet parameters, traditional audio engines typically jump between different impulses according to the new settings. With Fluid Convolution, everything will happen smoothly, with no gaps or artifacts, and with a new level of realism.


Where do I find Fluid Convolution?
The first Overloud product that will benefit from this new DSP Technology will be TH-U 1.3. It will be released on February 8th: the update will be free for all registered users of any TH-U version.

• thanks to the reduced CPU requirement we have been able to double the length of the IRs: this will produce a more accurate simulation and a more solid tone.
• in case of a different sample rate between the audio and the IR, the resulting tone will be more natural thanks to the absence of the resampling stage.
• changing the parameters, like moving the microphones or switching IRs, will happen much more smoothly: you’ll perceive a more natural feel when tweaking the models

What comes next?
During the Spring of 2021, Overloud will release a new product which will redefine the IR processing. Stay tuned!

 

Coming back here after it's been fully implemented in v1.4.0.
Sounds nice and I think I get the fluid thing. However I don't get the advantages of storing directly the frequency response.
No matter how you store an IR is still a finite digital signal.

 

Isn't this similar to Melda's MCabinet?

 

I'm still on Win 8. There is no TH-U joy in Mudville tonight!

 

In some features, yes. What a great plugin MCabinet is BTW.
However this technology is supposed to improve the sound and CPU usage of existing ones.

Here's what I get:
"The transition between IRs also happens seamlessly: when changing IRs or tweaking cabinet parameters, traditional audio engines typically jump between different impulses according to the new settings. With Fluid Convolution, everything will happen smoothly"
This can be done as some sort of advanced interpolation. I don't know how they do it but makes sense.

Here's what I don't get:
"When the recorded audio sample rate differs from the IR sample rate, conventional convolution requires resampling of the audio in order to match the two sample rates... with Fluid Convolution we store the IRs in the complex frequency domain, so no resample is needed anymore"

Don't get it

"thanks to the reduced CPU requirement we have been able to double the length of the IRs"
Even less

 

I don't use it either. As much of a metalhead ampsim lover as I am, TH-U has too much stuff both in disk size and number of amps and effects.

But I really admire Overloud plugins low CPU usage. Don't like the high RAM usage, even when it's not a problem for me anymore with my modest but well served in RAM new rig.