Aren’t the dynacabs captured through a neve, believe fractal used a different board for the stock IRS.
The "stock IRs" or factory IRs are a large mixture of captures form various sources, Fractal and third party. You can see a detailed list here:
https://wiki.fractalaudio.com/wiki/index.php?title=Cabinet_models_list#Axe-Fx_III_and_FM3_and_FM9
There are also lots of third party IR makers doing lots of captures at very high quality as well. Both from an evolution of capture process and from it being good marketting, most new IRs from any maker are "the best they've ever captured with the latest and greatest capture process available to them".
Fractal's new process is likely improved from they're previous process in their eyes, because there is no reason for them to use anything but the best process they know to caputre something like DynaCab.
But they are still IRs. There's nothing that inherently prevents another IR from sounding better.
The workflow is very flexible, as it gives you better control over the exact mic positions to craft the sound you want. Definitely more control compared to plenty of third party libraries which only inlcude a handful of locations and mixes. And vastlly more flexible than the factory IRs which only include a few mics and positions for each speaker. So you might find it easier to find the exact balance you want in Dynacab and find it tends to sound better for that reason. But there's nothing technologically that prevents another IR (from factory or third party) from happening to hit the perfect balance and sound better. "Better" is subjective anyway.
Pretty sure the curve goes back to where it was. I saw it happen, but that could be because I didn’t save the preset in Dynacab mode.
Any change you make will revert if you don't save the preset then reload it. But if someone did test dynacab then go back to legacy and save the changed impedance curve that could explain a difference in sound. Or maybe the speaker modelling added to a recent FXIII firmware pointed out by someone else in this thread could be what people are hearing, if it was included in this firmware and not earlier ones.
The thing IMHO that makes the Dyna Cabs useful is using them to get a great sound within a mix. I’m no expert micing engineer, but I have always used two mics on a cab (SM57 & Condenser) and am somewhat able to dial the dyna cabs in easily. And I’m convinced the overall quality is better. One caveat…I own no 3rd party IR’s. I do have the Leon Todd and the few that came with Gift of Tone which are great. There may be some tweaking that needs done by the 3rd party IR vendors to bring them in line with 7.0.
Those going from 6.0 to 7.0 b3. I thought I remember reading updates being done by some IR vendors in between 6.0 and 6.02. Correct me if I’m wrong.
An IR captures a mic/preamp/speaker/room setup, no changes to underlying models would require any changes to an IR to bring it in line with 7.0. Most of these IRs are not captured with fractal in mind, but with all guitar recording use cases in mind from standalone modellers (all platforms) to profiles/captures to VSTs to a real amp plugged into a load box recording silently. If anything, they would be designed to work best with the real amp as that is the most accurate (caveat about load box impedance curves), and Fractal is designing to get their models to behave as closely as possible to those real amps (plus greater control over load box impedance, speaker compression, and other things a load box and IR don't always capture).
IR vendors typically don't update their IR pacts after capturing, unless they go back and re-do a pack with a new process. And the factory IRs don't update as that would change people's existing patches using that factory IR. Nothing in 7.0 should need any IR vendors to do anything different.