Architect beta for macOS, Windows, and Linux. 0.10.5 now available

[colin@loomer]

Can we expand the number of lanes in the grid sequencer , 8 is a bit meager
Also , where are the in/outputs of the macro located ?

Yes, can do, although I'll probably keep the original as a Grid8 sequencer and add say a Grid16 brother module. 16 seems enough, right?

Do you mean the pins for the macro modules ? If so, they are under Built-in > Macro, and are the ones named * inlet and * outlet. They can only be created inside a macro.

[afta8]

Just playing with this, its great fun but it's crashed a couple of times when loading a VST before I could save, could I suggest you have a crash recovery or autosave options.

Also are there log files generated for when it does crash?

[Hez]

Quick couple of questions regarding MIDI sync to a host clock:

1. I've recorded Loomer's MIDI output into a new channel in Bitwig and found that all notes appear to be predelayed by a small amount, meaning that the first note of a sequence is dropped. The patch is very basic, just a grid step sequencer sent straight to MIDI output -> Host (ignore the aux sequencers, they come into play later). Is this expected behaviour or perhaps a peculiarity related to Architect's interaction with Bitwig?





2. I expanded the patch by hooking up the aux sequencers to modulate Divide and Count on the main grid step sequencer to get some IDM vibes going. The aux sequencer is synced to host (as is the main step seq), but the timing quickly drifts further after Divide has been increased above 1 a few times. Should I expect this kind of setup to be able to remain in sync naturally or do I need to add additional processing after the grid step sequencer to 're-quantise' things to the desired clock division?

Cheers

[colin@loomer]

Also are there log files generated for when it does crash?

Only the standard OS logs, but I'm adding additional logging in to help disagnose plug-in crashes.

[colin@loomer]

. Is this expected behaviour or perhaps a peculiarity related to Architect's interaction with Bitwig?

Not expected behaviour. It should be perfectly in time. MIDI messages to host should be the most reliable of all as they are timestamped alongside the audio stream and accurate to the sample level. Have you another host you can test this preset in to see if Bitwig are delaying it, for some reason?

Oh, sudden thought: have you got any plug-ins either in Architect or Bitwig that are adding plug-in delay? Maybe that could be pushing things off?

[colin@loomer]

I expanded the patch by hooking up the aux sequencers to modulate Divide and Count on the main grid step sequencer to get some IDM vibes going. The aux sequencer is synced to host (as is the main step seq), but the timing quickly drifts further after Divide has been increased above 1 a few times. Should I expect this kind of setup to be able to remain in sync naturally or do I need to add additional processing after the grid step sequencer to 're-quantise' things to the desired clock division?

Cheers

Yes, if you play with divide, count, speed, duration, etc, you will knock things off the metre. They will keep time with themselves, but not with a "global" clock. You can periodically resync with an event to a the modulation > pattern launch control inlet when the control action is set to restart. (Maybe it'd be handy to have something like this built-in?)

[ThomasHelzle]

OK, sounds like it's heading in the right direction. I'll have a look at the Bitwig scanner than and see what I can ape for the next scanning iteration.

Not so happy about the crash though: does this happen every time you run the app?

The FastSpring guy said that apparently the VATID dialog appears towards the end of a checkout, so it should already be there.

Yes, it crashes immediately when I click the icon every time. Architect shows up in Task Manager but only under Background processes, not in the Apps list on top (sorted by Name, Windows 8.1 x64).
My system has tons of stuff on it and two graphics cards. If I can run a debug build let me know.

I'll look at the order process again then.

Thanks and cheers,

Tom

[colin@loomer]

Thanks for your patience Tom. What I'll do is, as very sensibly suggested above, add concise logging to the next beta. That way, if things crash on other people's system, I can hopefully point myself in the right direction from the logs.

Should have really done this for the first release, but I guess we live and learn, right.

[ThomasHelzle]

No problem at all, the plugin works and that is what I will be using most - together with Bitwig Architect is dream team.

Cheers,

Tom

[Hez]

I expanded the patch by hooking up the aux sequencers to modulate Divide and Count on the main grid step sequencer to get some IDM vibes going. The aux sequencer is synced to host (as is the main step seq), but the timing quickly drifts further after Divide has been increased above 1 a few times. Should I expect this kind of setup to be able to remain in sync naturally or do I need to add additional processing after the grid step sequencer to 're-quantise' things to the desired clock division?

Cheers

Yes, if you play with divide, count, speed, duration, etc, you will knock things off the metre. They will keep time with themselves, but not with a "global" clock. You can periodically resync with an event to a the modulation > pattern launch control inlet when the control action is set to restart. (Maybe it'd be handy to have something like this built-in?)

Thanks for the swift responses Colin!

So I started a fresh project and added a fresh grid seq and everything works as expected, including modulating divide, as long as it doesn't go off into triplets or something weird.

As soon as you move divide by hand while the sequence is playing back, the sequence will be offset from the host global clock, which I guess is expected behaviour. What's slightly less expected is that if you set everything back to normal (divide 1, no modulation etc.), the sequencer will not resync nor start from step 1. It's pretty easy to check this by adding a grid seq, setting Duration to 1/4, starting host transport, messing with the divide a bit, stopping host transport, setting divide back to 1, then restarting host transport. When you start transport again the sequencer will start play back from some random position, which changes each time you repeat the above steps.

I guess the sequencer is retaining the last position it ended up in, even Resume is toggled off (I assume this is what this control is supposed to do)?

(I also just noticed that toggling transport on and off will reset the sequencers position to step 1)

Like you said I could work around this by making sure all sequencers have an event restarting them which triggers at least when playback is initiated, and more frequently than that if doing really crazy stuff with modulation of divide/count etc., but it does seem like the kind of functionality that might be worth building in as I imagine this is quite a standard expected behaviour.

EDIT: Apologies, I'm re-reading the quick start guide now and it's quite explicit about needing to reset events via the restore toggle - I was a bit presumptive in assuming that sequencers etc. would reset to step 1 automatically after screwing with their parameters! I will experiment with using events to reset things explicitly when doing crazy divides down into triplets etc. >_<

[colin@loomer]

Resume is for Gate and Toggle modes.

You are absolutely correct in your summary of the situation, though. I guess the easiest solution would be some kind of "sync with global clock" setting, per sequencer. So then you can set it to, say, 4 bars, and no matter how crazy your inner modulations are, every 4 bars the sequence will start restart from dot. Does that sound good?

[colin@loomer]

As an aside, tomorrow's beta will be later in arriving than the previous few have been.

I want to really ensure that the logging is concise enough that I can diagnose issues as swiftly as possible, which means lots of things that need to be logged, which means lots of work.

Thank you all.

[Hez]

Resume is for Gate and Toggle modes.

You are absolutely correct in your summary of the situation, though. I guess the easiest solution would be some kind of "sync with global clock" setting, per sequencer. So then you can set it to, say, 4 bars, and no matter how crazy your inner modulations are, every 4 bars the sequence will start restart from dot. Does that sound good?

Aha, I guess as the documentation gets fleshed out over the next few months some of these things will probably become a lot more obvious.

Sync with global clock sounds like a good idea; though in some ways I do like the idea of forcing the user to explicitly construct the required resets etc. as it teaches you how to do things properly

The main thing I think I'm trying to achieve with this specific patch is forcing the next sequencer note to fall on a given time division, synced with the global clock. E.g. every bar (host transport), no matter what the internal time position of the sequencer, it immediately jumps to the position of the next step, then resumes whatever crazy stuff is going on with dividers etc. putting stuff off grid, until the next bar at which point it resyncs itself etc.

Is that achievable currently? If not, maybe a Control Action like 'Jump to next step' would suffice? Then I could just hook up a metronome sending a signal every bar and resync myself every bar no matter how silly I get with my modulations

EDIT: Argh although having suggested that, that would mean I lose the Control Action to restart the sequencer when transport is started.

[ThomasHelzle]

Indeed the VAT-ID is in a later step and well hidden instead of where I would expect it (belongs to the other customer data IMO and not up there with the price and vouchers) but it worked and I'm now a proud owner of Architect. Thanks Colin!


I'm currently checking the new VST sorting options.
"By Location" is what I will use (until a search can be implemented) and that seems to work well mostly, although it seems your system does some "unfolding" where empty subfolders which contain another folder with the actual plugin are removed. Needs a bit of getting used to but is okay by me.

"By Format" I do not understand at all. I didn't scan my VST3 folder so it's almost only VST2s. I get hundreds of folders in the list which are almost all empty. Like 8 "Applied Acoustic Systems" folders where only one of them contains "Ultra Analog", all others are empty. And 55 "Airwindows" folders where only two of them contain a plugin. Not sure what this is supposed to do but doesn't look useful ATM?

"By Manufacturer" seems to work, but the sorting of the main list is strange: It goes from A-Z but then starts again with less Manufacturers going again from A-Z? I'd expect all manufacturers to be sorted in one go from A-Z?

"By Category" I'll probably never use. The list is less crowded now though, which is good, but still the sorting looks arbitrary to me, with for instance tkDelay and Sandman in "Reverb". Do the developers define those categories or how does this work?

"By Name" has the same sorting issue as "By Manufacturer", the list goes from A-Z, the last here being Zebrify, then it starts again with A, here that is audioM8 (part of Jamstix) and goes in a shorter list to Z again. From the plugins I can't make sense of what the difference is between the first and the second list part.

Again, all in the plugin using Bitwig as host and Windows 8.1 x64.

Cheers,

Tom

[colin@loomer]

Hmm, give me a while, let me think how you could achieve this... Worse case scenario, you get busy with some numeric and boolean tables and build your own sequencer!