bmanic wrote: ↑
Sat May 05, 2018 1:39 pm
giancarlo wrote: ↑
Fri May 04, 2018 6:15 pm
yes they are unique. We didn't advertise it very much (similar to what an other developer did), but we have unique channels. The point, our ones are sampled and not based on a random variation on components. We are modeling a real object, channel after channel.
Indeed! However, where this "other developer" used these different channel tolerances to their advantage was with allowing true stereo processing.. so that you can get a strange or nice stereo width effect due to various tolerances of every single component in the model (this includes EQ and compressor timins too). This is NOT at all the case with Cream, which in my opinion is a bit of a lost opportunity. The user should be able to choose a dual-mono (aka true stereo, eh?) version of Cream where the user can then choose freely the pair of Line/mic channels separately for left and right.
This is easy setup in a host like Reaper and I've done it already just to check how it sounds. Unsurprisingly it sounds amazing.. so yeah, consider this a feature request for the future.
In short: Separately sampling tons of channels is only interesting if you can run them in dual-mono mode. Otherwise it's a bit of a wasted of time.
there is a logical difference, but quite important. In one case you have a system composed by subsystems, and you see that elements are "not perfect". Following the imperfection used for being selected you have let's say N different channels with small differences. In our case we sampled "exactly" a console, channel after channel. We are not modeling a "random" console, but exactly a real one.
The difference between components is based on the age of components, on a large sets of imperfections, and even by partial recapping. Normally these consoles were really high-end, and at the very beginning the difference caused by this sort of tolerance was almost absent. When we model a difference it is because that difference is in that specific real item, it is not a guess.
If you take in consideration the whole cake, cream represents the first time this kind of emulation is applied to a whole console (not just few channels, not just random imperfections, and if it is not based on random choices, in our case we can guarantee each channel is exactly the same as the original one, channel after channel, accurately, harmonic response included). If someone already did it and I'm not aware, sorry for the confusion (but I don't see how it could be possible on a real console, considering the crazy number of components... you should make an accurate assessment of ALL OF THEM, or try to converge the imperfections as a result guessing original values for ALL OF THEM). Even in that case it is "acustica" approach, we are famous for the accuracy, and we are speading this accuracy to a very high number of channels (line AND mics!) (and it is a magic thing which is extremely easy to perform using our tools, and verified as pretty accurate).
About true stereo: if you ask to me the effect is completely absent on this kind of console. Crosstalk was measured and neglegible. Crosstalk is the only element which could bring something of channel L to channel R, for instance. It was something in the -100dB ballpark.
So in this case true stereo is overkill, and it would double the resource usage using our technique. We want something to apply to all channels. Now: cpu load is very very low today, and I can confirm it will be LOWER at least on windows side. We'll tell you more about a possible timeframe for this kind of update.