setronic
Synthedit regular
Posts: 293
Posts: 293
|
Post by setronic on Feb 27, 2014 6:42:11 GMT 1
all the examples show a 16 step sequencer with 16 stepcounter outputs going to 16 junctions(AND/LevAdj/Multiply/etc.) with ie: gate/event inputs sequentially/ in order 1 to 16.
because of a question about a small but noticeable latency/delay, against ie: a kick drum/tick generated by the VST host (Orion), it was suggested that the order of the step outputs should be shifted back -1, to give a 'roll-in' action to the sequencer start, and let the first step crunch in, bang on, on sequencer 'run' - having been reset on 'stop', to its reset position, '0':
> on 'run', counter moves from '0'(which doesn't exist) to step 1, where await gate and pitch cv info, ready to be sent on through the junction with the step (nb: ? and clock? i'm using it, but not sure if it is needed, 3 way AND)
by shifting -1, step 16 becomes step 0 at reset. it is a nice roll-in action. but i can't tell yet whether it gives perfect synchro, though i am assured this is the way.
the trouble with my build of this is that the LED activity shows the counter coming to rest at step 16 - i could AND it somewhere -with the run switch? i'm trying to get a snappy, fast start/stop action. it needs to only come on at the first count.
(would i have to do this for BPMClock3?) ------------------------------------------------------------------------ the other method i've tried recently for the stepcounter is RH stepcounterB, and this seemed to work pretty well with straight pin order 1-16(as opposed to 16-15?)- is this because GUIBools is always running, needs no time to wake etc? or a shorter instruction? i've tried driving this with bpmclock2, and tbh, i think there's still a small delay(i need to check again).
|
|
|
Post by KlangManipulation on Mar 3, 2014 13:09:21 GMT 1
all the bpm-clocks i have tried are not precise enough or have different problems. the bpm-clock 2 has also a "transport-run" problem. oh and you should know: with different DAW you will get different synchronisation problems. for example cubase has a "preroll" count in time. FL-Studio starts directly if you hit play. so if you test your plugin in FL the same chematic wont work for cubase! btw. the steps are 0-15 and not 1-16! so you have to set a -1 if you want to start more early. if you work with SE1.1 you have to use bpm-tempo module for a precise "transport run" start
use a counter like "sequenze-control" ( like the prefab i gave you" and a many to one to build a stepsequenzer. thats a much better way than using stepcounter.
|
|
setronic
Synthedit regular
Posts: 293
Posts: 293
|
Post by setronic on Mar 3, 2014 23:23:17 GMT 1
do you mean the bpmclock(3?) in SE 1.1 syncs accurately? or not? CK's one is supposed to be correct, i don't have those. what transport run problem do you mean? not using it, it doesn't do anything until SE transport runs. what do you use it for?
tempo out controls an oscillator i use as clock, dividing it by 60000(m/sec)
not using bar start. seems it could be useful > i've been wondering how to ensure patterns play to end before change >?how would this work with different lengths? 'bar' remains the same.. -'pulse' doesn't get used either, but i've put it in as an option. i can have clock vary off host tempo-/+ 50%(or whatever) or run locked to the bpm in tempo divisions, could be a combination of both ie: 'slower'/faster @ 1/2 or 1/8 divisons. but still has that slight delay when it should be at bpm
so i tried to shorten everything as much as possible, and removed everything i could, and now it's tidier but the same. the way bools runs seems simpler, text too, so i had a little exploratory messaround with that, quite good. still not sure it synced any better though, and it still depends on the same clock.
so far, i've been running this without the offset step- i ilke the way that works but it gives me a new problem with the LED display to fix: last step lights at rest! (it still needs and AND with 'Run' in there, to each step's LED (Run isn't necessarily transport run, any 'On' button...etc) the result is a noticeable latency next to eg: host instrument kick drum. if you run 2 instruments using the same sequencer, they run in sync (when i first tested this, i noticed i could control the tempo of antto's vb-303 beta2.. very odd, must be something to do with SE)
i haven't tried the -1 thing in a plugin test yet. there are several sequencer prefabs here! - it is getting confusing. one is driven by DD bool tick with RHstepcounterB. have to try 'roll-in' with that. it is still the clock that determines the end result.
antto said write a sample accurate(?) clock sem like he did - since bpmclock2 is at 'block' rate(is this right?)
hey man, how about you make a sample accurate bpmclock?! hehehe
|
|
|
Post by KlangManipulation on Mar 4, 2014 16:11:26 GMT 1
no clock i had tested was working 100% fine (even ck-clocks have problems with FL-Studio. maybe they work with cubase but i dont know). i do it the same way like you with an BPM-Synced oscillator the big problen was the first pulse of the bpm-clock. it was never in time! from the second pulse on it was working fine. the transport run is needed for your vst-plugin to know when you press start and stop in your daw. so its very important if you work with sequencing stuff but the transport run of the bpm-clock2 was also not always in time. you can resync oscillators with barstart connected to sync. in some cases its necessary to do that. the bpm-tempo module in SE1.1 has a 100% working transport-run now. there is a big difference between se1.0x and 1.1 se1.0x was syncing oscillators automaticly if you hit play on your daw. but this was the reason fl-studio was stuttering with complex synthedit sequencing stuff. in se1.1 they are running free and you have to sync all manually. so its more difficult to build proper syncing stuff with se1.1 but there is no more stuttering in fl-studio and about writing my own clock... im not able to do that. i was trying to, but if i was looking into the bpm3-code it was like reading chinese. also i found a bug. im not sure it is fl-studio or the se-module. wich DAW do you use for testing vst-plugins?
|
|
setronic
Synthedit regular
Posts: 293
Posts: 293
|
Post by setronic on Mar 5, 2014 23:04:01 GMT 1
Orion
|
|
setronic
Synthedit regular
Posts: 293
Posts: 293
|
Post by setronic on Mar 16, 2014 9:30:57 GMT 1
am i right in thinking a 17th step is needed for resets and correct led display? i'm using in this case RH stepcounterB which has fixed 16 steps outputs -not autoduplicating- and GUIList #steps selection etc.(very nice, wish it had autodup' pins -oh and combine the float, or do different versions, for all GUI pin types ) (32/64bit wishlist) so: if i select length =4 it counts on the leds 1,2,3, then flashes 16. mmm, maybe i just need to offset my leds, right,(doh) ;-) that's gotta be it. other train of thought was have a rest position step, is: 17 so, on 'run', it comes in from #17, does the # of steps selected, resetting at last selected step+1 ... ?? so that would be: off=17,(ie:)1,2,3,4,reset, and skip the off step somehow... how's that? not tried it yet. so 'run' could operate a DH bool 1 to many, including/excluding the 17 step from the count. :/ ?
|
|
setronic
Synthedit regular
Posts: 293
Posts: 293
|
Post by setronic on Mar 16, 2014 13:16:08 GMT 1
tried inverting LED order, no good. basically, when it is off/stopped, no LED is(should) on, LED1 should light with step1. but it is missing, and doesn't light until step 2. it does this in normal order, and with steps offet-1 compensated with LED offset-1. if no LED offset, steps show 1,2,3, then 16,1,2,3, eg: if you do a 4 step loop. it only runs to 3 with 4 steps seleted. so i'm thinking about a suggestion once involving end steps: a 'rest' step/0 step, for stop, but resetting to the first step used. so...that 0 step would? go on the end? with steps and LEDs offset. :/
|
|
setronic
Synthedit regular
Posts: 293
Posts: 293
|
Post by setronic on Mar 17, 2014 7:23:10 GMT 1
here is an example prefab (1.017) showing RHstepcounterB driving LEDs, one regular order, the other offset-1, with #steps selection: stepcount order test2 versions of each: one has added AND stage connected to 'run'. -'runAND/normal order' runs correct length, but 1st LED doesn't come on with Run -'normal order'; same but no AND stage, you see where the LED is -'runAND/offset-1' does the right sort of thing, but only shows 1,2,3, then skips back to 16 -'offset-1' shows it without AND this is a bools version of what max suggests, unless i've misunderstood. i'm confused. i'd be inclined to do normal order -but i still can't get the LED to light on the 1st step with the AND+Run. surely, if you offset-1, you have to wait for that first stepcount to elapse before getting your first actual count? what's the right way to do this?
|
|
|
Post by KlangManipulation on Mar 18, 2014 17:55:53 GMT 1
i am sorry to say that but...... GUI-Counters are to slow because of the individual refreshrate of DAW´s. its better to use DSP for realtime changes ! i was trying the dh-array sequenzer and it was always 1 step too late. you can do a workaround and exchange first and last step.
|
|
setronic
Synthedit regular
Posts: 293
Posts: 293
|
Post by setronic on Mar 19, 2014 22:04:16 GMT 1
yeah i was coming to that conclusion. the 1st step isn't hitting on the Lut noD bcs it is too slow. dsp is at sample rate. i also found my text array attempt with DH array sequencer to be 1 step out, very weird. do you mean you can't use text array to sequence? i think sirsicksik has done it. the prefab is a test for the 1st/last step offet. how should i do sequenced accents? just 1 level, on/off. i've been getting all steps playing same accent level before, i did step x accent, but i don't think that is it.
|
|
|
Post by KlangManipulation on Mar 20, 2014 22:31:41 GMT 1
uhhm.. sorry but i dont understand sequenced accents?... just 1 level? accent = volume? just 1 volume level? what is the problem and what are you trying to do?
EDIT: the counter must be a dsp-module! thats why i gave you the "sequenzecontrol" prefab.
|
|
setronic
Synthedit regular
Posts: 293
Posts: 293
|
Post by setronic on Mar 21, 2014 19:51:15 GMT 1
yes of course - i've never used the bools stepper, it is just for (fun) really - i always use a bpmclock2 and stepcounter.
accents: yeah i said that wrong: accents is accents i'm not completely happy with what i've used before.
steps can be just accented or not accented, global accent level for all voices. or: individual accent level accent on/off, or different levels - eg: Off, +1,+2,...+7 and minus values too. gotta go out, back in a bit ;-)
|
|
setronic
Synthedit regular
Posts: 293
Posts: 293
|
Post by setronic on Mar 22, 2014 8:44:08 GMT 1
i've been looking at GUIBools/List/Text/float/ for a while, figuring it out - very cool, it all runs with SE transport 'off' - and avoiding the dsp stage until i get to the essential parts. don't think it needs to be taken too far though it doesn't have to be bools stepcounter, it's just for trying out, eliminates volt>GUIbool conversion, though..i have a few little side prefabs trying out/learning logic connections. you get a few sites online that demonstrate boolean logic with java or flash or whatever. synthedit is a lot more fun to try that sort of thing out with, logic-lab thing. it suddenly slows down a lot though when things get big. and not really *that* big, 16x gate+time/accent/porta/semi+oct/ and some LED display stuff, with what i've got for patterns (which could be vastly streamlined with arrays, all of it could), copy all this a couple of times, and it slows up on 2ghz budget thing with onboard graphics.(what would speed redraw up? a decent graphics cache?/card?)
|
|