The Axe-FX Ultra-Res 2.0 impulses operate at 170 ms with some data compression so they are less taxing on the processor. They sound pretty convincing in a mix IMHO.
I shoot my cab pack impulses that you can find at
https://www.drbonkerssoundlab.com/ at 170 ms (exclusively for the Fractal packs) as well as 200 ms, and 500 ms for the Wave packs at 96 Khz and 32 bit. I convert down from those high resolution formats so that others do not have to for the various hardware or for software sessions that are recorded at the other dominant sample/bit depth specs. Some credit must go to
@ownhammer really pioneered the longer IR lengths as well as higher sample/bit depth and illustrated its benefits, so I must give a shout out and tip of the hat to Kevin for doing this work and making me a philosophical convert,
Since most hardware amp modelers' conversion tools can truncate longer impulse responses if they can't use them and the deconvoluted impulse responses don't seem to lose anything sonically in truncation too much.. There is no hardware modeler on the market yet that can make captures at 96 KHz 32 bit either at such long sweep lengths.
What is the difference of longer capture lengths versus shorter lengths? Visually, if you look at the convoluted waveforms for the various lengths at the same sample rate, when the sweep is longer in time duration, there is more information captured for both the low end and high end. Remember, it's not so much that you sweeping more frequencies beyond 20 KHz over that time duration, it's that you are taking a longer time to get there and as such the capturing apparatus is capturing more info in that time.
But can you hear a difference? In a double blind test with Mrs Bonkers selecting the same type of mic/cab combination file at 96 Khz/32bit, but just varying at different file lengths, it's not so much that you hear a bass boost or treble boost or the sudden appearance of possible room tone, but it's more about a smoother transition playing notes up and down the instrument neck, a more 3-D or realistic sound, and the low end feels more "solid" for lack of a better descriptor. Unique timbre is more identifiable when selecting different cabs too at the longer lengths with higher sample rates.
Is it a dramatic difference sonically or aurally? In my opinion it's maybe a 5% improvement when hearing it in context with other instruments.
Why do I personally chase this 5% when creating my cab packs if only DAW/emulation software standalone software jockeys can even utilize it? My feeling is that I am future-proofing my impulse response captures for the foreseeable future. At some point, someone is going to have the hardware processing power to handle longer file lengths or higher sample/bit depth rates. I do not want to have to track down and physically transport 250 lbs of SVT in a flight case to re-shoot it along with many other cabinets and spend yet another 100 to 200 hours in the shooting and cab re-mixing process on cab, if I already did work to bring to the musician public.
I wish some future version of Cab Lab would support longer lengths, higher sample rate/bit depths, and perhaps even wave file archives of the shot files, but that is not the reality at the present time.