I think there are (at least) a couple of things to be aware of here (which you quite possibly are already be aware of). First is that full sequence length on Pyramid is the length at which all the patterns in the sequence return back to their start points. So with the above example the full sequence length (for both S1 and S2) is 2 bars, if the 2 tracks/patterns in the example were instead 3 bars and 4 bars long the full length of the sequences would be 12 bars, and if they were 4 and 5 bars long instead, the full length of the sequences would be 20 bars. If you wanted S1’s full length to be 1 bar you would need to create a different pattern that was 1 bar in length for track 2 (which, of course, would not be the same pattern anymore, and would also require more of Pyramids project RAM to store). I guess this isn’t the sort of solution you would want (it would likely lead to needing to create more patterns and sequences).
The other thing to be aware of is what Pyramid might ‘actually’ doing in the background (I would say these are the things that enable it to do stuff like polymeters and polyrythmns). Of course, I can’t be sure what Pyramid is ‘actually’ doing in the background but I tend to think that once you press play its like Pyramid is constantly running the tracks/patterns in free run mode and with the sequence, when in perform mode, with ‘Instant’ (track mute delay and/or pattern delay) as its ‘basic’ setting(s). I would say ‘default’ settings but I don’t think ‘Instant’ is probably the default on a straight out of the box Pyramid. But I do see it as the ‘unaffected’, non delayed, and therefore 'normal/basic/‘default’ setting in that sense.
Delaying the mute/unmute and/or pattern change is a means to help you change states and still keep them in sync, or, if you like, to allow you to be a bit less precise than you’d have to be if there was only the ‘Instant’ setting available. With delay settings that are the same or less than the shortest pattern’s bar length (in that sequence) and when all patterns are in the relatch run mode I think you should get the behaviour that I think you are probably expecting. Once the delay settings are bigger than the shortest lengthed pattern you will get this apparent ‘problem’ occuring.
Anyway, the thing I want to try to get across is that when you press play on the Pyramid (at least the way I see it) this sets off all the tracks/patterns playing from their start points and they continue (in the background) independently looping according to their own length and time signature irrespective of what other settings (including the mute state, or run mode) that you have put in place and this, I think, is the basic behaviour of Pyramid. I see those other settings (like run mode or pattern/trackmute delays) as non-destructive ways to manipulate the basic behaviour (in a similar way to how Pyramid’s FX are non destructive). Personally I find this a helpful way to look at it in terms of understanding what Pyramid is doing. You may disagree on both its helpfulness and whether this is indeed ‘actually’ what is going on in the background. It sort of works for me, but that might just be me.
I suspect there isn’t a workround that you would be totally happy with. I think there are really only two ways that I can think of to get the desired behaviour that I think Alexb wanted (and I guess what you also would want). And I think Alexb had pretty much found them out. Those being, to use track mute (and pattern) delay settings that are as short as necessary (or as long as you can get away with) for your performance, or to use one of the song like Seq modes. But I guess that was not at all suitable for Alexb’s needs (and probably not yours either). Unfortunately, if you want to perform the sequence changes (live) I think you would probably have to be able to actually get to Pyramid’s controls to change the sequence somewhere near the time when you wanted it change if you were using tracks of different length within those sequences. There may be other (probably overly complex) ways to program the sorts of changes you might require in a given project by using an assortment of run modes, track lengths, pattern variations, a really good memory, some amazing foresight and planning skills, and probably more sequences than Pyramid allows for.
However, despite this, I think if you increase the length of your shorter patterns (beyond the length of mute/pattern delay(s) that you might need for your performance), you should be able to use longer delay settings and get closer to what I think you would want. It may seem rather inefficient to have an 2 bar pattern that is actually the same as a 1 bar pattern repeated (especially given the low amount of MIDI event Pyramid can store in its RAM). But if you really need to be able to cue up sequence changes in perform mode further in advance this may be what you’d need to do. I don’t think it would work in every situation, like if you needed to cue up sequences well in advance (mute/pattern delays longer than, say, 2 or 4 bars) and where you were already close to the limit of MIDI events that Pyramid could store in its RAM. But how often would you need (or want) to have more than a 1 or 2 bar delay to the sequence change? Obviously, abandoning the full sequence perform delay setting will mean you couldn’t just cue up the next sequence at anytime after the last sequence change (and then relax) but as you have found out that is not really possible the way you want with ‘full’ selected anyway, at least not unless you have all the patterns set to the same length and that if that is the length you want the sequence to play on until.
I would suggest not using the full sequence delay setting. Pick the shortest delay you feel you can manage. Ensure your shortest patterns (tracks, if your not using patterns) are longer than that delay. Alternatively, (and I wouldn’t advise this) you could consider using more sequences that are more or less duplicates of others but that contain different patterns on certain tracks that are of the requisite length so that the full sequence length is the length that you may require for a particular section of your ‘song’. This may require remembering (or predicting?) when you need to select that sequence rather than the original one and may not really be very practical. In all honesty, I think you should ditch the full setting when in perform mode.
I hope that this will be of some help. I’m no expert and It’s quite likely i have missed the point altogether. Apologies if thats the case and I hope someone on here can offer some better advice.
