Just for fun I quickly benchmarked (very non-scientifically) single patch per instance vs. multitimbral on my AMD Ryzen Threadripper 3970X (32 cores, SMT disabled), using Omnisphere 2 in Nuendo 10.2.20. I picked Omnisphere, because anecdotally it seems like it has one of the highest per process overheads out of all my VST instruments, so it should provide a "worst case scenario" baseline. I loaded the first eight patches in the Omnisphere 2 default list and had them play a scale with 1/8s at 120. I kept duplicating the tracks until the machine would play no more, with ASIO Guard at high and my RME HSDPe AIO set to 512 samples. Results: multimbral Omnisphere capped at 136 patches. Single instances at 152. That's a 12% improvement in the single instances favor, which, while not huge, is still significant. However, I was expecting actually a bit more on my machine, given how much the CPU favors parallel workflows. Clearly the multitimbral Omnisphere utilizes multithreading quite well already. RAM usage doubled from 10GB to 20GB, so RAM is the currency with which you pay for this improved parallelization (and routing convenience). Calculating the RAM difference per instance here, there's roughly a 81MB RAM cost per instance of Omnisphere, so for each eight patches, you use an extra 567MB. That's quite high, so it's something to keep in mind. Maximum RAM is going up way faster than single thread perfomance in CPUs, so I agree with the idea that it's definitely time to consider leaving multitimbral workflows behind. The day to do that won't be here for every system yet, especially if you have a CPU with a lower core count and less RAM. But it's coming for sure.
No one of us loads 152 instance of Omnisphere. So multitimbral has still his place....as he told in the video for instance for Groove Agent or multilayered Sounds in Halion Sonic
Don't forget... if one sets up their Hardware as a Cubase "External Instrument", there's no need to use Midi tracks at all. The Hardware will load as an instrument track, Midi and Audio combined, in realtime (if you set it up right).
@@vegsozoltan5283 Cubase "External Instruments" works for multitimbral instruments too, but yes, you'd need midi tracks for the extra midi channels to route to the "master" instrument track for the external hardware.
If you use MIDI tracks, it is much more simple to change the entire instrument afterward. Your MIDI track is then assigned to an instrument instead of connected to it and fully integrated. Change an entire instrument on an instrument track is much more complex than with a MIDI track assigned to a Rack Instrument. So that is for me an important pro-MIDI track argument.
agreed , unfortunately there is something of an issue here if you have not made up your mind what instrument to use then if you change to something even relatively close but not the same then the way you play it often needs to change. For example a bass track. Many different acoustic or electric basses you need to play differently because of the way each responds to velocity or note duration these effect the way it sounds and often needs you to replay it again. So its really good to decide first what exact sound you are going to use so your MIDI is played in in a way , for example the piano you choose of bass responds. Of course you can do this later but rerecording a track. its small but important thing to keep in mind to help your work flow be as efficient as possible, Choice of sounds is extremally import in the final mix. planning as much as possible before can help. of course sometimes we may like to just wing completely but in that case at least having a good grasp of what sounds fit well in the mix can help so you grab good ones
I totally agree with you. But since you asked about ways I use instrument channels: I pretty much use one instrument track per patch/sound. But I have found that some libraries, like for example Hollywood orchestra, where you need several patches in one instrument (for example violins - could be up to 10 patches), I use expression maps in order to trigger various channels (one patch per midi channel). That is, only one instrument track and no midi channels in cubase, but the sampler (in this example "play") have several patches loaded, and each patch is dedicated to a midi channel. The old way was to then have several midi channels in cubase, connected to the instrument channel/rack, but now the map is instead controlling what sound is heard. I can either change the sound with keyswitches/controller, or I can change it in the key editor. I recently came across this trick and it is a massive time saver and workflow enhancer. For example, the violins could go from pizzicato to legato to marcato etc, and it is all controlled in the key editor with "articulations". Keyswitches are actually eliminated this way, it is faster, and no need for alignment and having midi notes in the score who are just there for the switching, but I can still control what sounds I want to hear with a controller if needed. This can be done with any sampler, and is very useful for any patch with no keyswitches built in. But so far this can only be done in Cubase as far as I know.
Instrument tracks are faster but the rack is still useful for layering, to give one example. I've also had issues with multi-out instruments like Superior Drummer when not using the rack, so I just use it for all multi-out instruments. You can do some interesting things with midi tracks like use them for patterns using the step sequencer or as events that hold automation that can then be moved around the timeline easily. It can also be conceptually useful to think of a part as notes rather than locked into a sound. I like that Cubase offers such flexibility.
I am totally with you. But I really miss that there is no easy way to convert them into each other (MIDI instrument). This is completely incomprehensible to me.
when you load one instance of VST instrument, Cubase will instantiate that instrument and reserver like 1.5GB memory. When you instantiate second VST instrument, it will reserver more memory. If you load one VST instrument, it will only instantiate memory once. Basically if you have tons of ram, you can just always instantiate new VST instrument, no need just basic midi tracks.
I use midi channels for layering. For example I may create low and high layers of sound on synthesizers and then send midi channel to them so I don't need to always input midi on two channels.
Greetings from Vegas Dom! There is great value in having both options - I use both MIDI and Instrument Tracks. So many Cubase users are doing cool creative projects - I do jingles, and often I have to work fast. In my situation, the drawbacks of using MIDI tracks with a frequently-used Multi-timbral synth (Halion Sonic, etc.) are overcome with preset templates. In other words - the template loads preset MIDI tracks, and audio pre-routed to preset Group Channels. (In my case - 4, labeled "drums", "bass", "keys", "guitar". I have 4 tracks in Halion preset and prerouted accordingly. Make sense? If i'm on a deadline, I know where to choose patches, and they are already routed. All that said - these days - there are SO many cool instruments out there, it's completely reasonable as you're saying in the video to use Instrument tracks. You're not going to use them on every project, and you're not likely to have them pre-loaded in your template. I think the bottom line is to be aware of both types, and be willing to use either depending on your workflow.
I build large orchestral templates. I halion\kontakt per track. I agree with you. It would also be so good if the track name would take the name of the preset it allocated to by default. This could be an option in preferences “default track name takes preset name” . Also additionally the actual sound being played I.e. the name of the preset or even its full articulation name could be seen in the track., not just in the tiny expression map panel but clealrly in the project window so when the track is stopped the cursor position might show “staccato” . So much time is spent typing in track names simply calling them by the preset names in most cases.
One crucial piece of info here is that for those who load lots of tracks in templates - the disable track feature is useful. Nothing is active but all tracks can be ready at your fingertips without having to add tracks repeatedly in every big project.
Man how do I insert a vsti if I import a midi track i create with other software. Bcs in cubase it is created as midi track, since you need instrument track to play the midi
@@firmans12 Two ways I can think of would be - select "Destination" to be instrument tracks in the import options. See the help here for more info: steinberg.help/cubase_ai/v10/en/cubase_nuendo/topics/importing_audio_and_midi/importing_audio_and_midi_import_options_for_midi_files_r.html Otherwise, just create instrument track, and then drag the midi events you imported over to that track from the midi track and delete the midi track afterwards. Hope that helps.
You don’t have to load instruments into the rack for multi out instruments anymore. Since about Cubase 8 or 9 the instrument tracks were overhauled to handle multi timbral instruments. You can add additional outputs to an instrument track now & just add MIDI tracks to route MIDI to the additional instruments in Halion or whatever sampler you are using
You can load a HALion (Sonic) full of the same sound in several slots with variation you've made and then use an expression map to switch between MIDI channels in one instance you have on screen all the time. You can have one big phat Anima (or Trium or SkyLab or ... n), or one thin one with some hi pass filter, one with some flanger or chorus super stereo etc.. It's less of a hassle to my brain to have the variations in one plugin unit but everybody's different. In the same manner you can save a little orchestra with leads, pads and basses as a template if you save the sounds as a bank. Then you're ready to go in a similar project. I don't like the mess in the inspector with both MIDI and audio inserts sends and whathaveyou. It's hard to teach an old dog to shit! :D There are benefits of both ways so I use both. Monotimbral VSTi:s I sometimes use Intrument Tracks for like most bass tracks only need one. Maybe I should have more of a method to my madness? 🙃😂
Simple answer - create templates - the number of steps (midi tracks) is done once. I use both because sometimes you need midi only e.g. Original groove agent or having visual cues on the arrangement page (not as nice if embedded within several takes of single instrument track)
I am new to Cubase and there is such a steep learning curve but you are so good at simplifying and condensing the information. This video completely demystified the concept of midi track vs instrument track vs rack instrument. Bravo
I use midi tracks to automate some elements in fabfilter plugins. I also use separate midi channels for some other data besides notes. A lot of times when creating, it's very convenient to have some extra track that only does the cutoff frequency or some other parameter. I use a lot of outboard gear. So in the end, tracks are merged. The cpr is exported as a midi file and imported in the MPC. It plays all synths and modules and sets the effect processor to the desired program. In a few cases I exported some audio as well and used that in the sampler, but sinds it's the intention to play and sing along live, I only use that for drum loops and strange effect samples. You don't need to activate the audiotracks per se, if you assign outputs in the plugin, they will be activated automatically.
I have been a Cubase user for 6 years and always thought that Groove Agent would be pointless for me, so thank you for that output tip, I can now see it becoming a part of my work flow
I'm a little bit late to the party, but it really makes little sense to use complex MIDI/Output routing with the same plugin, because this is how modern OSes work. Anything that can run-executable files, plugins etc-internally consists of _sections._ A section may contain executable code, read-only (RO) data, or read-write (RW) data. The read-write data is not stored in file, rather declared only (I'm omitting unnecessary details). Now, executable code is always read-only, it's just like RO memory: it never changes, it what was put in there by the compiler, the tool that programmer used to create the plugin file. Since there's no difference between code and RO data for our purposes, I'll be speaking of RO code only. Modern OSes use _virtual memory:_ a section loaded from the file is _mapped_ into the host's _virtual_ RAM, and this mapping is separate from the occupied physical address. Cubase is even unaware of that, the OS hides this unneeded info from the program. This is in fact done with the help of the CPU memory manager. Each program has access to insane virtual memory, about 10 quintillion bytes, and each byte is numbered from 0 to 10 quintillion. But it's linear, just like the timeline of a song has a beat number. Let's better measure memory in beats-why not? Just like a DAW's track, virtual memory is empty, and the OS adds _(maps)_ continuous "events" so many beats long here and there-exactly like audio events in a track, which may have gaps (but no overlaps). You can think that each program is a separate track; a Web browser has its own "track" 10 quintillion beats long. But we'll focus on one, single Cubase "track", its virtual memory. When you load an empty Halion, its executable code and RO data (which contains various baked-in DSP tables, for example, also baked in when the plugin was created, strings for messages etc) is loaded from the Halion.vst3 file into the main physical memory (RAM). Of course, it creates some RW data to keep track of what programs are loaded, the controls positions, connections with the host (e.g., Cubase) etc. But these tables are small when empty, and 16 or 64 empty slots take very little RAM. Now, Halion's RO code is loaded from the file Halion.vst3 into physical memory and _mapped_ into Cubase's insane address space. Suppose Halion's code got loaded at the beat 100, and is 400 beats long. Some RW data, private to this Halion instance, is also placed somewhere into its own "event" somewhere in this "track". Now, imagine you load a second Halion. Now something interesting happens. Program code (RO) is _never loaded the second time._ The OS remembers where it's in physical memory. Even better, you load it into the same host. The OS figures out: RO code from Halion.vst3 is already loaded into this copy of Cubase, so let's just reuse it. It's already in the beats 100 to 500 in its memory. Only a small chunk of private RW data for the second Halion is created, because it has its own 64 slots, an audio connection with the host, etc. The executable code never changes, so that totally fine to do: unlike a song track that also _plays_ from lower to higher beat numbers, memory is only accessed by address, our analogy of the beat number, but in random order. Does the overhead of the little RW memory chunk, individual to each Halion (say, the first one gets private "event" in beats 700 to 800, and the second in beats 900 to 1000) matter? I'd say no, because for a multi-timbral instrument Cubase also creates extra data for each used slot: a new MIDI track and a new output. These also unpredictable in advance, and needs allocating private RW memory as needed. Dom is totally right: it will save you no memory in the end. When you load a sampler… now that's where memory-heavy stuff really begins. Halion reads samples from disk when playing a track, or even preloads then if you play live. _This_ is what eats enormous amount of memory for high-quality library. Synths, like analogue or FM zone, take much less space: they need to store a hundred or two of numbers for knob positions, and some more to track every voice phase, LFO phases, envelope time position etc. But this is nothing compared to a sample library with 5 articulations, 5 expressions and 6 levels of loudness for each group of 3‒4 notes. This dwarfs the extra overhead, whether you use a Halion per track or multiple inputs or outputs. Since Cubase virtual memory ("track length", including empty space and "events") is limited to quintillions of bytes, the first limit you hit is the amount of allocated physical memory, say, 64GB ≈ 70 billion bytes, of which only 55 is available to _all_ user programs. This is why the number of Halions playing at once is not that important: they share all memory that never changes, which is very economical if your instruments don't depend on samples (Analogue, FM, Organ zones) or use small samples (loops, sampler tracks, Wavetable, Granular and Spectral zones). This is where sharing code between all instances of the same VSTi saves a lot of physical RAM. With high-quality sample libraries, even this optimisation drowns in the ocean of memory required for keeping samples in RAM. BTW, samplers may or may not share samples between the same instrument loaded into multiple tracks. Good programming techniques permit mapping sample files directly into memory, and the OS plays the same trick: gives each use of the sample a separate virtual space (range of beats) while using one shared copy of the sample, as it never changes, too. Even more interesting, the same physical memory is mapped into _different_ programs! If you open another program that loads Halion, the same physical memory will be mapped at zero cost into other program's virtual memory, as if into its own separate track (compare with Cubase's shared events, where you may clone events to different tracks without making a copy of an audio file on disk-this is a nearly exact analogy). I can't think of using two DAWs at once, but if you open, say, Cubase and WaveLab and load the same effect into both, its physical copy of executable code will be loaded only once, and shared between the two programs. In fact, parts of the OS are loaded using the same mechanism into a hundred or so of running processes at once. The only thing to be aware of is VST plugin graphics windows. Loading 100 Halions and making them play at once is totally fine and results in little memory overhead, but don't open the windows of all 100 of them at once: you'll deprive the system of graphics resources, as they are expensive, each window needs much more private RW memory than 63 of 64 empty slots in each instance, and the system needs even more for each open window. But I'd say if you have 64GB, opening 5-10 Halion windows is fine. You'd rarely need even 5.
I have a sample server, where I use Vienna Ensemble Pro. In Pro Tools, they allow you to route Instrument Tracks to other instrument tracks' instruments. (In Cubase, this would be routing them to a Rack or Track Instrument.) Boy, this functionality would be so handy in Cubase, for those of use with a similar setup. Sure, I _could_ just use a single box and do without the VEP, but... the sample box is kept extremely cool, which means its super noisy, and I prefer having it in a different room so it doesn't disrupt the composition process.
Makes a lot of sense man! :-) There is a lot of "suggestions" in the production world that seems to just be information left over from the legacy days.
Excellent video Dom, thank you. I would like to add 2 additional reasons to use midi tracks: 1. when we need to control plugins like PitchCorrect If you want to drive the voice with a midi keyboard you need to set pitchcorrect to external midi input and add a midi track which outputs to inserts>pitch correct and voila. 2. When you want to use midi sends to create a layer of different instruments without creating several instrument tracks and copying the same midi part several times in the project window. Instrument tracks do not have midi sends. Hope this helps Cheers
One (bug?) I've noticed about using instrument tracks, certainly with EZ Drummer, is that the instrument track name is always the same as output 1, changing one changes the other and vice versa. This means you can't meaningfully name your track if you also want to meaningfully name output 1. For example, output 1 is Kick - you whole drum track is now called Kick. This is why I stick with using the VST Rack 😐
I use both types of tracks at the same time. That way I can use diffrent sequences on the same instrument if it is playing sevral things, like I will still print the audio in difrent passes, but if I just use one generic synth to compose diffrent elements to get a feel of the composition, I do not have to pull up new VSTs and tweak the settings every time. I could probebly set it up simpler, but it is the way I have gotten used to from 2.0 and 3.0 SX. I do have a full outboard rig, so usually I will use VST instuments when composing, then later route them to the outboard gear. That way I can really focus on the composing when I am doing that, and the sequence will be superversitile when it is finished because it can be routed to just about anything.
It's really convenient to have Groove Agent as a VST instrument and export a kit's mix to Cubase. One drawback that I have found is that the audio channel for the kick, for example, cannot be "ungrouped" or moved out from under the VST instrument. This is inconvenient in the mixer if you have downstream group channels or FX channels associated with it and you would like to group them together in the mixer view. For example, say you add an FX channel with a sine wave generator to add a bit of energy to your kick, and then say you also add some parallel compression after that, then those extra channels cannot be next to the original kick audio channel in the mixer which could lead to a lot of scrolling or navigation in the channel settings window.
I use MIDI channels for arrangement purposes. Like, I like having my hit hats, cymbals and toms in separate MIDI tracks from kick and snare, but all part of the same instrument/instance of the plugin so it is an in-between approach. I use multiple instruments if the instruments themselves are different, so same workflow you described.
Thanks. I used to argued with Cubase user friends and they always told me that "is all about saving CPU power". In my case, I love Instrument tracks because they save me "time" and my CPU have never complained. But the most important thing is that I can pretend that I have another audio channel with the ability to sneak into the midi parts if it's necessary. That visually clean my session and that is really cool. 😎
I had difficulty using midi tracks probably because I didn't understand how to use it. I only use instrument tracks exclusively... Thanks for sharing your expertise.
Hmmm, yes Instrument track is much easier and I use that every time I need one track / one sound But when doing orchestral composition, one trick to get beautiful strings sound is to layer several libraries. And in such case, it's much easier to manage one midi track / several instruments. I don't know if it's possible in any other way in Cubase.
Addressing Dom's question here I actually stumbled on one situation where I'm finding it useful to run one Instrument Track together with up to 15 MIDI Tracks. It's when using a Kontakt Multi instance loaded with several instruments of a typical orchestral sample library. The scenario is when I spread out the Kontakt sampler CPU load between Cubase and Vienna Ensemble Pro streaming server software. The way, in Cubase, to address VEP-hosted Kontakt instances and their included instruments then is by MIDI channel assignment. The reason for such a setup is that VEPro is very fast at streaming samples from SSD drives so I can actually use a template four times bigger than Cubase could handle. VEP keeps shuffling samples in real-time and Cubase stays totally snappy, never feels bogged down by heavy CPU load. And it's fast when you want to try out layering instruments, just drop them into a Kontakt instance and assign them to the same MIDI channel.
I use Vienna Ensemble Pro for my orchestral projects. So, I use MIDI tracks. Thats the standard. However, when I'm not doing orchestral, instrument tracks all the way. Thus was an awesome video. Good confirmation of how I use all my DAWs. Now, you should check out Digital Performer. Lol...beast of a DAW!
...dear Dom, informative as always...but I must confess, I am more a midi track user, especially for plug-ins as Kontakt or groove agent with many midi tracks going into one vst rack, as I do not need the audio functions that often...I first render the midi into audio and then use inserts, sends, etc.
I came here to see if maybe I was missing out on some trick by using Instrument tracks instead of MIDI tracks - turns out I've been using them correctly! I literally have a Groove Agent/Halion template (among others) set up in just this manner! This is literally my "go to" channel for all things Cubase! I've used many DAWS, but Cubase trumps them all IMO!
Hi Dom, great advice as always. I'm basically a novice. I use outboard gear and soft synths but usually only one patch per synth so instrument tracks are fine. Where I do vary is with drum machines like Battery. I'll have one instance of Battery and one midi track for each drum sound, then when I'm happy with the arrangement I bounce to audio. I'm sure I'm probably missing a trick and could simplify but I do like each drum sound in its own midi track.
SUPERB !! Many Thanks Dom. Sooooo many tutorials still opening midi tracks & racks, then routing, then grouping mix channels, really confusing when getting started. Very clear, very concise Dom. As always !!
I only really use MIDI tracks to trigger VST FX with midi inputs. Like it can be useful for retriggering LFO tool at specified points. Aside from that it’s instrument tracks for all of the things. I really hope Cubase 11 finally delivers on long awaited native modulation tools that can be assigned anywhere in the DAW, then I’d have little use for LFO tool. The MIDI inserts need replacing with a better modulation system & quick controls need to be assignable to multiple parameters if I’m going to upgrade again. Otherwise Bitwig is looking a lot better for my needs.
I'm never disappointed with these videos, there is always something to be learned! Now I understand the concept of "rack mounted" instruments (it seemed like an odd option). It makes far more sense with its historical context as "genuine virtual outboard midi gear" .
I still use midi tracks and rack-instruments because I am limited in the Elements Version to 24 Instrument Tracks..and with the additional 16 Rack-Insruments I have enough for my projects.
I use instrument tracks but if I use multitimbral instruments I'll add midi channels to use the other instrument channels. I haven't tested it myself but it should use less resources. I saw a video which said if you load several instances of the same instrument in one Kontakt instance it uses the needed Ram only once. The whole VST instrument situation can become difficult quickly. What I would really like to see is a video (series) covering the whole difficult situation of creating VST drums...Creating a groove (midi files, finger drumming, helper tools like Realidrums and Jamstix), triggering different drum vst's (problem midi mapping and drum maps), creating mix ready audio of the seperate drum instrument channels. That is a topic which seems to be an endless journey to me, finding one solution and stepping into the next problem at the same time. I have several drum VST's and as nothing is perfect (I like Grovve Agents integration in Cubase, SSD5's versatility and the sound of Abbey Road Drums in a messy Kontakt) but I am still considering buying more although I fear I will never be satisfied.
Thanks again, Dom! This was the clearest explanation I've seen. Having used Cubase all the way back to Steinberg VST on an Atari, I have some long ingrained habits and have been averse to switching from rack instruments to instrument tracks. You have led me to see the light!
Snap! Though I changed about 2 weeks ago. Haven't seen any CPU issues and the flexibility is so much better. Goodbye MIDI track you served me well :-) Actually MIDI tracks can be useful for sending controller commands to multiple instruments using MIDI sends.
Yeah I would never use rack instruments like that now; namely because the best use of those additional midi channels in Halion Sonic is for layering different voices onto 1 instrument so it's like you're creating a new instrument. If you want to layer on non-Halion Sonic instruments it's honestly still easier to use a different Instrument track and just use a shared copy of the MIDI event so it's always in sync. You can use a group to apply effects or a VCA for volume and it's all still much easier than the rack.
Dom can you also create a tutorial to show us how to have multiple virtual instruments tracks with different midi channels that way we can play each instrument part simultaneously on different channels.
I'm migrating from Ableton, and it's pretty easy to route the Midi out of my Interface to the Midi in on my keyboard, I love the complexity of Cubase but damn it's crazy being a beginner with a DAW again. The struggle is worth it, but I definitely need an in depth tutorial on routing Midi data in Cubase.
I am on the same this days man. Personally i am going to open manuals pdf with cubase at the same time i think it's going to be so useful , videos of dom of course too useful haha.
Hi Dom! In the main, I prefer to use Instruments as Tracks, but the whole 'Track v Rack' approach depends on individual workflow preferences. I know that for opening old projects and multi-timbral MIDI files (especially old GM Atari ones) you would be faster using a rack instrument to load. MIDI tracks by themselves can be used to record Sysex data dumps from external hardware, to control external FX units with MIDI (such as Reverbs and Delays) and for triggering Plug-ins like Stutter Edit, Vocal synth and Break Tweaker from Izotope (to name a few). In addition to all of this, I understand that you can even control a lighting rig with MIDI! Thanks for the video :-)
Thanks Again Dom! Nothing beats the access to all the plugin power of an audio track over a midi track. Greetings from Oz! I just went back over this a day later & assigning the kit to individual outs is a BONUS, thanks again!
I'm not used yet to Rack instruments. I must say it's a bit weird to me, but for now there is another reason I would use them above classical instrument tracks. Simply editing in groove agent seems nicer for what I can tell. There is a quick tip from Steinberg that demonstrate the difference between using an instrument track and a rack instrument when dissolving the part. Instrument tracks dissolve the parts but duplicate the instrument. Looks like a waste of time and resources. Rack instruments create only the routed MIDI tracks which is what you'd expect. Maybe later on Steinberg would get rid of rack instruments or at least have the same behaviour when dissolving the patterns with instrument tracks. Then I would not take racks as an option.
The only thing that is sometimes "frustrating" with instrument track is that when we mute it, it's for MIDI and audio. So, if I just want to live play to test something without hearing what was recorded before, I have to mute events ... with MIDI track linked to an instrument track, I can select the MIDI track, mute it and continue to play on this MIDI track, it works !
Wow. I had not even thought about those differences. I mainly used instrument tracks so far. I really like the tip about routing Groove Agent to different tracks. Thx Dom.
Heads up: - for those who like to route the Groove Agent individual sounds to different output tracks (as shown in the video) ... This only works when you use drumsets from "The Kit Se". - If you use a different kit (for example Beat Agent), then the "Export mixer to Cubase" will be missing from the menu.
Great video...it cleared up a lot. I tend to use the rack feature only when I'm starting a project and I'm 'sketching' an initial arrangement and just want a Halion instrument with a preset containing basic instruments; Drums, bass, piano, etc.. I then put the drums into Groove Agent and select different instruments and put them on instrument tracks. So I don't really use the rack feature any longer. The one thing I was not aware of was the quick way you assigned the drums, with names, to each output. I always did the slow, tedious way. It made the whole video worth the watch. Since I've been using Cubase since VST32, I tend to miss some subtle improvements that are great time savers. Thx.
One thing Steinberg should change: Mute a MID-track and you still can play the instrument. Mute an instrument track and you can't. I often need this to just try a small change in a hook-line or checking out a different melody.
Great work Dom! I have used instrument tracks from the time I started using Cubase. As yet I have not had a problem with overloading the CPU at all. It is so much easier to use multiple instances of Halion or Halion Sonic or Groove Agent. There is too much to think about if you use midi tracks!
Great explanation and video. I still use Rack Instruments when using Vienna Ensemble Pro because I like to freely move the outputs next to each one of my midi channels in the arrange window. The Instrument track locks the outputs all to it so the additional midi tracks (channels 2-16) can not have the outputs next to them (midi track 2, output 2, etc). I like to have them next to each midi track so I can do automation in the arrange window and such. I think when one just wants a single instrument, instrument tracks are great because the midi and audio are together as a hybrid but for VE PRO environments with multi-timbral it doesn't quite work as well for me.
Thanks! Since I started with Cubase many years ago I was using one instance of HAlion Sonic. Now based on your advice I' changing my project to have one per track. I also was using MIDI tracks and now using Instrument tracks. Great advice especially for Groove Agent SE outputs.
Dom, you’ve done it again. I know all about instrument versus midi track but..........reminding me of the GA mixer to Cubase that I had forgotten about. I gave myself a slap. So many years as a semi pro dabbler and never really spending enough time on individual beneficial tweaks. At last I learn the right way. Cheers
Maybe years ago, our laptops did not have the processing power to handle all the info for each instrument. [It happened, and it was disastrous. Even if instrument tracks were muted, they still used processor power.] I have a hoarding problem - in the house, or on my project page. I keep too many incomplete 'versions' muted, so now I have several pieces of one 'song' over several names, and have to drag them across from one project to another. I have severe ADD etc. and have to move on to a new verse, chorus, etc. before fixing the last one. There must be a better way to file things away from cluttering up my project page. It's probably a matter of not having got to a stage of using a regular method to build with. It's also probably that I'm not a live 'muso' with a repertoire of styles, rhythms or instrument set-ups, etc. So many tutorials to get through. I just saw your one for vocal harmonies - but I still have not even had time to use the new mike and work out how to record vocals without a sound-proof room, etc. Our local builders were supposed to build my studio months ago, but we now have a 'covid change' - a rush of people moving out of cities, making tradespeople scarce here. So, it looks like I have no choice - there's no room to MIDI up all the old keyboards, etc. I have to stay in bed with a light MIDI keyboard controller and laptop! We've extreme 'social distancing' - but not one case of Covid locally. One step closer to a cyber-matrix world.
One of the things thats good about pro tools, not many things (for composition), is that an Instrument track can have audio and midi from whichever track routed to that same track.
For me the "render in place" method is number one. But has anyone solved how to setup multitimbral hardware synthesizers as "external instrument tracks" and use render in place on the other parts than just the external instrument main track itself?
Thanks so much!........I need to watch this again.......and then again to absorb it all. I'm a real newbie to virtual instruments and midi. So much to learn!!
I started with Cubase with 8 and by that time the rack intsrument thing was already history really. I've always used instrument tracks and never had a problem so I've never used MIDI tracks really. I have 12 logical cores and rarely do they ever go over about 50% each, usually about 6-8BG RAM use out of 16. So no system resource issues. Where I do sometimes have a problem is with the real time processing meter hitting the buffers when I have a VST instrument track armed for recording. This is because I can't afford a stripped down dedicated separate machine for music production yet so there's often stuff running in the background that interrupts the CPU which then interrupts the ASIO buffer stream transfer. Overall system load is fine, maybe 30-40% but RTP hits the buffers. I can usually fix it by restarting my session from a cold boot, but this is a nuisance. If you got any words of wisdom on how to improve RTP, Dom, from my experience I'd say that's the biggest system resource problem running Cubase on today's computers. PC anyway, dunno about Macs. Also some VSTs seem cause much less load on RTP than others. All the Cubase stock instruments like Retrologue are super efficient. Some of the Arturia Collection synths are the worst. Serum and Omnisphere in the middle.
I know you're right on this, it just feels wrong. Example, i use xpand 2 to create a 4 part choir harmony. I'd much rather use 4 midi tracks, one instrument, and put them in a folder than use 4 instrument tracks. The biggest benefit of instrument tracks for me though is just not having to worry about midi channel assignments.
If I need to choose an instrument I add a MIDI track, and different instruments or samplers. Then I can send my MIDI-track to any instrument and choose what I need.
Best 12 minutes ever!! Yup, I arrived at the same conclusion (instrument track), but what you really clarified was how the computer processing was great for that decision now. Ahhhhh.
brilliant video Dom. i've always wondwered about this. i use instrument tracks most of the time, and in fact midi tracks very rarely nowadays. Usually only use midi tracks if i'm importing a midi drum groove form a 3rd party developer - although not exclusively.
Useful vid, thanks. I use Cubase Elements for VSTi projects and use individual instrument tracks. I don’t really need any more than the 24 tracks included but could always use the midi track/rack instrument method if needed.
Hey Dom - I actually do use MIDI tracks for the reason of separation .. Let me explain, when I use a drum kit - I like to put each triggered sound on it's own channel so I can find it easily when making modifications later. I also use them to break up sections that might use the same instrument in some case in a different way; for example... If I was working on an all Piano track... I might separate my break it into parts where I would play the bass, and lead parts on different tracks. That's just me - maybe I will do a partner video to show what I mean and mention your video. If that is ok with you.
I only use multiple outputs when I use groove agent easy to keep the drums separated, still nice to know what and why Didnt know about the activate menu I was renaming them manualy! thanks for that
Thanks Dom. I do use Inst track with software sinths except when i m doing arrangements and final instrument is not finally defined. It is easy for an A /B comparison.
I have not used external instruments since I got HALion back in 2001. I don't use midi tracks since "instrument tracks" were intruduced. Always hated the midi tracks getting in the way and had to hide them all the time in the mixer. I got to talk to you Dom...
Thank you for this, Dom. I have had that question in the back of my mind about using MIDI tracks and since I don't use outboard instruments besides guitars, I now know I don't need to worry about it!
Hi Dom , thanks for the great advice . I just realized that I uselessly use multiple rack instruments and probably waste cpu power. I am switching over to instrument . All you videos are really helpful. I use your tricks daily,. Thank you.
Makes sense. Thanks, Dom. However, as an improvising composer, I often take my separate MIDI track and create a score that I can edit. This is normally done instrument by instrument, of course. Perhaps I just don't understand how to get access to the Instrument Track MIDI?
I used to use midi channels back in the Cubase 6 and 7 days. I still have projects like that and a couple of them have lost the sounds they were assigned to. I didn't know about the trick with Groove Agent se. I had been using loops then doing the resolved part thing.
That is the most amazing intro track I have ever heard dom! It’s so exciting everytime I hear it 😍. I’ve just gone back to Cubase after years and years of neglecting it. I’ve never been able to fully get my head around it but I must say you have helped re-enthuse me. I loved your cc121 video and I was torn between that and a keyboard controller. I ended up getting a nektar P6 and it’s taking some getting used to.....but I think it’s going to be really good. I have about three albums worth of unfinished stuff on nuendo 2 (yes lol) that I was going to re-do on an akai force! Instead I’ve managed to get Cubase 5 up n running so I’m bringing them all into that 😍
Don't know about anyone else but I revisited a session where I had used Lots of different midi channels in around 4 instances of Kontakt with lots of automation on the midi channels and it was a complete mess .... Although it maybe more efficient this way, its a lot harder to work with., I prefer to keep 1 instrument track (instance) per channel strip.
Instrument track!! 💯💯💯 “Keep your life simpler”
Just for fun I quickly benchmarked (very non-scientifically) single patch per instance vs. multitimbral on my AMD Ryzen Threadripper 3970X (32 cores, SMT disabled), using Omnisphere 2 in Nuendo 10.2.20. I picked Omnisphere, because anecdotally it seems like it has one of the highest per process overheads out of all my VST instruments, so it should provide a "worst case scenario" baseline.
I loaded the first eight patches in the Omnisphere 2 default list and had them play a scale with 1/8s at 120. I kept duplicating the tracks until the machine would play no more, with ASIO Guard at high and my RME HSDPe AIO set to 512 samples.
Results: multimbral Omnisphere capped at 136 patches. Single instances at 152. That's a 12% improvement in the single instances favor, which, while not huge, is still significant. However, I was expecting actually a bit more on my machine, given how much the CPU favors parallel workflows. Clearly the multitimbral Omnisphere utilizes multithreading quite well already.
RAM usage doubled from 10GB to 20GB, so RAM is the currency with which you pay for this improved parallelization (and routing convenience). Calculating the RAM difference per instance here, there's roughly a 81MB RAM cost per instance of Omnisphere, so for each eight patches, you use an extra 567MB. That's quite high, so it's something to keep in mind.
Maximum RAM is going up way faster than single thread perfomance in CPUs, so I agree with the idea that it's definitely time to consider leaving multitimbral workflows behind. The day to do that won't be here for every system yet, especially if you have a CPU with a lower core count and less RAM. But it's coming for sure.
Im happy with my omnisphere. Really happy.
Thank you for this test, that's informative.
No one of us loads 152 instance of Omnisphere. So multitimbral has still his place....as he told in the video for instance for Groove Agent or multilayered Sounds in Halion Sonic
Going by your test, the best would be to have 2 or 3 instances and then do everything internally. Just a theory.
Don't forget... if one sets up their Hardware as a Cubase "External Instrument", there's no need to use Midi tracks at all. The Hardware will load as an instrument track, Midi and Audio combined, in realtime (if you set it up right).
dont worry he don’t know what you talking about , cheers 🥂
How do you use the external instrument as a Cubase Instrument if that is a multi timbral instrument? In this case you should use MIDI channels.
@@vegsozoltan5283 Cubase "External Instruments" works for multitimbral instruments too, but yes, you'd need midi tracks for the extra midi channels to route to the "master" instrument track for the external hardware.
If you use MIDI tracks, it is much more simple to change the entire instrument afterward. Your MIDI track is then assigned to an instrument instead of connected to it and fully integrated. Change an entire instrument on an instrument track is much more complex than with a MIDI track assigned to a Rack Instrument. So that is for me an important pro-MIDI track argument.
agreed , unfortunately there is something of an issue here if you have not made up your mind what instrument to use then if you change to something even relatively close but not the same then the way you play it often needs to change. For example a bass track. Many different acoustic or electric basses you need to play differently because of the way each responds to velocity or note duration these effect the way it sounds and often needs you to replay it again. So its really good to decide first what exact sound you are going to use so your MIDI is played in in a way , for example the piano you choose of bass responds. Of course you can do this later but rerecording a track. its small but important thing to keep in mind to help your work flow be as efficient as possible, Choice of sounds is extremally import in the final mix. planning as much as possible before can help. of course sometimes we may like to just wing completely but in that case at least having a good grasp of what sounds fit well in the mix can help so you grab good ones
I totally agree with you. But since you asked about ways I use instrument channels:
I pretty much use one instrument track per patch/sound. But I have found that some libraries, like for example Hollywood orchestra, where you need several patches in one instrument (for example violins - could be up to 10 patches), I use expression maps in order to trigger various channels (one patch per midi channel).
That is, only one instrument track and no midi channels in cubase, but the sampler (in this example "play") have several patches loaded, and each patch is dedicated to a midi channel. The old way was to then have several midi channels in cubase, connected to the instrument channel/rack, but now the map is instead controlling what sound is heard.
I can either change the sound with keyswitches/controller, or I can change it in the key editor.
I recently came across this trick and it is a massive time saver and workflow enhancer. For example, the violins could go from pizzicato to legato to marcato etc, and it is all controlled in the key editor with "articulations". Keyswitches are actually eliminated this way, it is faster, and no need for alignment and having midi notes in the score who are just there for the switching, but I can still control what sounds I want to hear with a controller if needed.
This can be done with any sampler, and is very useful for any patch with no keyswitches built in. But so far this can only be done in Cubase as far as I know.
You became my number one cubase tutorial guy really fast and helped me tons already. Just wanted to say thank you :)
Greg is still the best!!!
Totally agree.
Instrument tracks are faster but the rack is still useful for layering, to give one example. I've also had issues with multi-out instruments like Superior Drummer when not using the rack, so I just use it for all multi-out instruments. You can do some interesting things with midi tracks like use them for patterns using the step sequencer or as events that hold automation that can then be moved around the timeline easily. It can also be conceptually useful to think of a part as notes rather than locked into a sound. I like that Cubase offers such flexibility.
You can multi output instrument tracks too
I am totally with you. But I really miss that there is no easy way to convert them into each other (MIDI instrument). This is completely incomprehensible to me.
when you load one instance of VST instrument, Cubase will instantiate that instrument and reserver like 1.5GB memory. When you instantiate second VST instrument, it will reserver more memory. If you load one VST instrument, it will only instantiate memory once. Basically if you have tons of ram, you can just always instantiate new VST instrument, no need just basic midi tracks.
I use midi channels for layering. For example I may create low and high layers of sound on synthesizers and then send midi channel to them so I don't need to always input midi on two channels.
Greetings from Vegas Dom! There is great value in having both options - I use both MIDI and Instrument Tracks. So many Cubase users are doing cool creative projects - I do jingles, and often I have to work fast. In my situation, the drawbacks of using MIDI tracks with a frequently-used Multi-timbral synth (Halion Sonic, etc.) are overcome with preset templates. In other words - the template loads preset MIDI tracks, and audio pre-routed to preset Group Channels. (In my case - 4, labeled "drums", "bass", "keys", "guitar". I have 4 tracks in Halion preset and prerouted accordingly. Make sense? If i'm on a deadline, I know where to choose patches, and they are already routed.
All that said - these days - there are SO many cool instruments out there, it's completely reasonable as you're saying in the video to use Instrument tracks. You're not going to use them on every project, and you're not likely to have them pre-loaded in your template.
I think the bottom line is to be aware of both types, and be willing to use either depending on your workflow.
This topic had been confusing me for years - makes perfect sense now. Thanks, man.
I build large orchestral templates. I halion\kontakt per track. I agree with you. It would also be so good if the track name would take the name of the preset it allocated to by default. This could be an option in preferences “default track name takes preset name” . Also additionally the actual sound being played I.e. the name of the preset or even its full articulation name could be seen in the track., not just in the tiny expression map panel but clealrly in the project window so when the track is stopped the cursor position might show “staccato” . So much time is spent typing in track names simply calling them by the preset names in most cases.
One crucial piece of info here is that for those who load lots of tracks in templates - the disable track feature is useful. Nothing is active but all tracks can be ready at your fingertips without having to add tracks repeatedly in every big project.
Man how do I insert a vsti if I import a midi track i create with other software.
Bcs in cubase it is created as midi track, since you need instrument track to play the midi
@@firmans12 Two ways I can think of would be - select "Destination" to be instrument tracks in the import options. See the help here for more info:
steinberg.help/cubase_ai/v10/en/cubase_nuendo/topics/importing_audio_and_midi/importing_audio_and_midi_import_options_for_midi_files_r.html
Otherwise, just create instrument track, and then drag the midi events you imported over to that track from the midi track and delete the midi track afterwards. Hope that helps.
You don’t have to load instruments into the rack for multi out instruments anymore.
Since about Cubase 8 or 9 the instrument tracks were overhauled to handle multi timbral instruments. You can add additional outputs to an instrument track now & just add MIDI tracks to route MIDI to the additional instruments in Halion or whatever sampler you are using
You can load a HALion (Sonic) full of the same sound in several slots with variation you've made and then use an expression map to switch between MIDI channels in one instance you have on screen all the time. You can have one big phat Anima (or Trium or SkyLab or ... n), or one thin one with some hi pass filter, one with some flanger or chorus super stereo etc.. It's less of a hassle to my brain to have the variations in one plugin unit but everybody's different.
In the same manner you can save a little orchestra with leads, pads and basses as a template if you save the sounds as a bank. Then you're ready to go in a similar project.
I don't like the mess in the inspector with both MIDI and audio inserts sends and whathaveyou. It's hard to teach an old dog to shit! :D
There are benefits of both ways so I use both. Monotimbral VSTi:s I sometimes use Intrument Tracks for like most bass tracks only need one.
Maybe I should have more of a method to my madness? 🙃😂
Simple answer - create templates - the number of steps (midi tracks) is done once. I use both because sometimes you need midi only e.g. Original groove agent or having visual cues on the arrangement page (not as nice if embedded within several takes of single instrument track)
I am new to Cubase and there is such a steep learning curve but you are so good at simplifying and condensing the information. This video completely demystified the concept of midi track vs instrument track vs rack instrument. Bravo
I use midi tracks to automate some elements in fabfilter plugins.
I also use separate midi channels for some other data besides notes. A lot of times when creating, it's very convenient to have some extra track that only does the cutoff frequency or some other parameter. I use a lot of outboard gear. So in the end, tracks are merged. The cpr is exported as a midi file and imported in the MPC. It plays all synths and modules and sets the effect processor to the desired program. In a few cases I exported some audio as well and used that in the sampler, but sinds it's the intention to play and sing along live, I only use that for drum loops and strange effect samples.
You don't need to activate the audiotracks per se, if you assign outputs in the plugin, they will be activated automatically.
I have been a Cubase user for 6 years and always thought that Groove Agent would be pointless for me, so thank you for that output tip, I can now see it becoming a part of my work flow
Your channel could easily become an entire grad school music production program. Thank you very much for sharing your knowledge!
I'm a little bit late to the party, but it really makes little sense to use complex MIDI/Output routing with the same plugin, because this is how modern OSes work. Anything that can run-executable files, plugins etc-internally consists of _sections._ A section may contain executable code, read-only (RO) data, or read-write (RW) data. The read-write data is not stored in file, rather declared only (I'm omitting unnecessary details). Now, executable code is always read-only, it's just like RO memory: it never changes, it what was put in there by the compiler, the tool that programmer used to create the plugin file. Since there's no difference between code and RO data for our purposes, I'll be speaking of RO code only.
Modern OSes use _virtual memory:_ a section loaded from the file is _mapped_ into the host's _virtual_ RAM, and this mapping is separate from the occupied physical address. Cubase is even unaware of that, the OS hides this unneeded info from the program. This is in fact done with the help of the CPU memory manager. Each program has access to insane virtual memory, about 10 quintillion bytes, and each byte is numbered from 0 to 10 quintillion. But it's linear, just like the timeline of a song has a beat number. Let's better measure memory in beats-why not? Just like a DAW's track, virtual memory is empty, and the OS adds _(maps)_ continuous "events" so many beats long here and there-exactly like audio events in a track, which may have gaps (but no overlaps). You can think that each program is a separate track; a Web browser has its own "track" 10 quintillion beats long. But we'll focus on one, single Cubase "track", its virtual memory.
When you load an empty Halion, its executable code and RO data (which contains various baked-in DSP tables, for example, also baked in when the plugin was created, strings for messages etc) is loaded from the Halion.vst3 file into the main physical memory (RAM). Of course, it creates some RW data to keep track of what programs are loaded, the controls positions, connections with the host (e.g., Cubase) etc. But these tables are small when empty, and 16 or 64 empty slots take very little RAM. Now, Halion's RO code is loaded from the file Halion.vst3 into physical memory and _mapped_ into Cubase's insane address space. Suppose Halion's code got loaded at the beat 100, and is 400 beats long. Some RW data, private to this Halion instance, is also placed somewhere into its own "event" somewhere in this "track". Now, imagine you load a second Halion. Now something interesting happens. Program code (RO) is _never loaded the second time._ The OS remembers where it's in physical memory. Even better, you load it into the same host. The OS figures out: RO code from Halion.vst3 is already loaded into this copy of Cubase, so let's just reuse it. It's already in the beats 100 to 500 in its memory. Only a small chunk of private RW data for the second Halion is created, because it has its own 64 slots, an audio connection with the host, etc. The executable code never changes, so that totally fine to do: unlike a song track that also _plays_ from lower to higher beat numbers, memory is only accessed by address, our analogy of the beat number, but in random order.
Does the overhead of the little RW memory chunk, individual to each Halion (say, the first one gets private "event" in beats 700 to 800, and the second in beats 900 to 1000) matter? I'd say no, because for a multi-timbral instrument Cubase also creates extra data for each used slot: a new MIDI track and a new output. These also unpredictable in advance, and needs allocating private RW memory as needed. Dom is totally right: it will save you no memory in the end.
When you load a sampler… now that's where memory-heavy stuff really begins. Halion reads samples from disk when playing a track, or even preloads then if you play live. _This_ is what eats enormous amount of memory for high-quality library. Synths, like analogue or FM zone, take much less space: they need to store a hundred or two of numbers for knob positions, and some more to track every voice phase, LFO phases, envelope time position etc. But this is nothing compared to a sample library with 5 articulations, 5 expressions and 6 levels of loudness for each group of 3‒4 notes. This dwarfs the extra overhead, whether you use a Halion per track or multiple inputs or outputs. Since Cubase virtual memory ("track length", including empty space and "events") is limited to quintillions of bytes, the first limit you hit is the amount of allocated physical memory, say, 64GB ≈ 70 billion bytes, of which only 55 is available to _all_ user programs. This is why the number of Halions playing at once is not that important: they share all memory that never changes, which is very economical if your instruments don't depend on samples (Analogue, FM, Organ zones) or use small samples (loops, sampler tracks, Wavetable, Granular and Spectral zones). This is where sharing code between all instances of the same VSTi saves a lot of physical RAM. With high-quality sample libraries, even this optimisation drowns in the ocean of memory required for keeping samples in RAM. BTW, samplers may or may not share samples between the same instrument loaded into multiple tracks. Good programming techniques permit mapping sample files directly into memory, and the OS plays the same trick: gives each use of the sample a separate virtual space (range of beats) while using one shared copy of the sample, as it never changes, too.
Even more interesting, the same physical memory is mapped into _different_ programs! If you open another program that loads Halion, the same physical memory will be mapped at zero cost into other program's virtual memory, as if into its own separate track (compare with Cubase's shared events, where you may clone events to different tracks without making a copy of an audio file on disk-this is a nearly exact analogy). I can't think of using two DAWs at once, but if you open, say, Cubase and WaveLab and load the same effect into both, its physical copy of executable code will be loaded only once, and shared between the two programs. In fact, parts of the OS are loaded using the same mechanism into a hundred or so of running processes at once.
The only thing to be aware of is VST plugin graphics windows. Loading 100 Halions and making them play at once is totally fine and results in little memory overhead, but don't open the windows of all 100 of them at once: you'll deprive the system of graphics resources, as they are expensive, each window needs much more private RW memory than 63 of 64 empty slots in each instance, and the system needs even more for each open window. But I'd say if you have 64GB, opening 5-10 Halion windows is fine. You'd rarely need even 5.
I have a sample server, where I use Vienna Ensemble Pro. In Pro Tools, they allow you to route Instrument Tracks to other instrument tracks' instruments. (In Cubase, this would be routing them to a Rack or Track Instrument.) Boy, this functionality would be so handy in Cubase, for those of use with a similar setup.
Sure, I _could_ just use a single box and do without the VEP, but... the sample box is kept extremely cool, which means its super noisy, and I prefer having it in a different room so it doesn't disrupt the composition process.
Makes a lot of sense man! :-) There is a lot of "suggestions" in the production world that seems to just be information left over from the legacy days.
Excellent video Dom, thank you. I would like to add 2 additional reasons to use midi tracks:
1. when we need to control plugins like PitchCorrect If you want to drive the voice with a midi keyboard you need to set pitchcorrect to external midi input and add a midi track which outputs to inserts>pitch correct and voila.
2. When you want to use midi sends to create a layer of different instruments without creating several instrument tracks and copying the same midi part several times in the project window. Instrument tracks do not have midi sends.
Hope this helps
Cheers
One (bug?) I've noticed about using instrument tracks, certainly with EZ Drummer, is that the instrument track name is always the same as output 1, changing one changes the other and vice versa. This means you can't meaningfully name your track if you also want to meaningfully name output 1. For example, output 1 is Kick - you whole drum track is now called Kick. This is why I stick with using the VST Rack 😐
I use both types of tracks at the same time. That way I can use diffrent sequences on the same instrument if it is playing sevral things, like I will still print the audio in difrent passes, but if I just use one generic synth to compose diffrent elements to get a feel of the composition, I do not have to pull up new VSTs and tweak the settings every time. I could probebly set it up simpler, but it is the way I have gotten used to from 2.0 and 3.0 SX. I do have a full outboard rig, so usually I will use VST instuments when composing, then later route them to the outboard gear. That way I can really focus on the composing when I am doing that, and the sequence will be superversitile when it is finished because it can be routed to just about anything.
After almost 20 years of screwing around with FL, I'm finally here 😊 really helpful this guy. Love his enthusiasm
Welcome!
It's really convenient to have Groove Agent as a VST instrument and export a kit's mix to Cubase. One drawback that I have found is that the audio channel for the kick, for example, cannot be "ungrouped" or moved out from under the VST instrument. This is inconvenient in the mixer if you have downstream group channels or FX channels associated with it and you would like to group them together in the mixer view. For example, say you add an FX channel with a sine wave generator to add a bit of energy to your kick, and then say you also add some parallel compression after that, then those extra channels cannot be next to the original kick audio channel in the mixer which could lead to a lot of scrolling or navigation in the channel settings window.
I use MIDI channels for arrangement purposes.
Like, I like having my hit hats, cymbals and toms in separate MIDI tracks from kick and snare, but all part of the same instrument/instance of the plugin so it is an in-between approach.
I use multiple instruments if the instruments themselves are different, so same workflow you described.
Thanks. I used to argued with Cubase user friends and they always told me that "is all about saving CPU power". In my case, I love Instrument tracks because they save me "time" and my CPU have never complained.
But the most important thing is that I can pretend that I have another audio channel with the ability to sneak into the midi parts if it's necessary. That visually clean my session and that is really cool. 😎
I had difficulty using midi tracks probably because I didn't understand how to use it.
I only use instrument tracks exclusively...
Thanks for sharing your expertise.
Hmmm, yes Instrument track is much easier and I use that every time I need one track / one sound
But when doing orchestral composition, one trick to get beautiful strings sound is to layer several libraries. And in such case, it's much easier to manage one midi track / several instruments. I don't know if it's possible in any other way in Cubase.
Addressing Dom's question here I actually stumbled on one situation where I'm finding it useful to run one Instrument Track together with up to 15 MIDI Tracks. It's when using a Kontakt Multi instance loaded with several instruments of a typical orchestral sample library. The scenario is when I spread out the Kontakt sampler CPU load between Cubase and Vienna Ensemble Pro streaming server software. The way, in Cubase, to address VEP-hosted Kontakt instances and their included instruments then is by MIDI channel assignment.
The reason for such a setup is that VEPro is very fast at streaming samples from SSD drives so I can actually use a template four times bigger than Cubase could handle. VEP keeps shuffling samples in real-time and Cubase stays totally snappy, never feels bogged down by heavy CPU load. And it's fast when you want to try out layering instruments, just drop them into a Kontakt instance and assign them to the same MIDI channel.
I use Vienna Ensemble Pro for my orchestral projects. So, I use MIDI tracks. Thats the standard. However, when I'm not doing orchestral, instrument tracks all the way. Thus was an awesome video. Good confirmation of how I use all my DAWs. Now, you should check out Digital Performer. Lol...beast of a DAW!
...dear Dom, informative as always...but I must confess, I am more a midi track user, especially for plug-ins as Kontakt or groove agent with many midi tracks going into one vst rack, as I do not need the audio functions that often...I first render the midi into audio and then use inserts, sends, etc.
I came here to see if maybe I was missing out on some trick by using Instrument tracks instead of MIDI tracks - turns out I've been using them correctly! I literally have a Groove Agent/Halion template (among others) set up in just this manner!
This is literally my "go to" channel for all things Cubase! I've used many DAWS, but Cubase trumps them all IMO!
Hi Dom, great advice as always. I'm basically a novice. I use outboard gear and soft synths but usually only one patch per synth so instrument tracks are fine. Where I do vary is with drum machines like Battery. I'll have one instance of Battery and one midi track for each drum sound, then when I'm happy with the arrangement I bounce to audio. I'm sure I'm probably missing a trick and could simplify but I do like each drum sound in its own midi track.
SUPERB !! Many Thanks Dom. Sooooo many tutorials still opening midi tracks & racks, then routing, then grouping mix channels, really confusing when getting started. Very clear, very concise Dom. As always !!
I only really use MIDI tracks to trigger VST FX with midi inputs. Like it can be useful for retriggering LFO tool at specified points. Aside from that it’s instrument tracks for all of the things.
I really hope Cubase 11 finally delivers on long awaited native modulation tools that can be assigned anywhere in the DAW, then I’d have little use for LFO tool. The MIDI inserts need replacing with a better modulation system & quick controls need to be assignable to multiple parameters if I’m going to upgrade again. Otherwise Bitwig is looking a lot better for my needs.
I'm never disappointed with these videos, there is always something to be learned! Now I understand the concept of "rack mounted" instruments (it seemed like an odd option). It makes far more sense with its historical context as "genuine virtual outboard midi gear" .
I still use midi tracks and rack-instruments because I am limited in the Elements Version to 24 Instrument Tracks..and with the additional 16 Rack-Insruments I have enough for my projects.
I use instrument tracks but if I use multitimbral instruments I'll add midi channels to use the other instrument channels. I haven't tested it myself but it should use less resources. I saw a video which said if you load several instances of the same instrument in one Kontakt instance it uses the needed Ram only once.
The whole VST instrument situation can become difficult quickly.
What I would really like to see is a video (series) covering the whole difficult situation of creating VST drums...Creating a groove (midi files, finger drumming, helper tools like Realidrums and Jamstix), triggering different drum vst's (problem midi mapping and drum maps), creating mix ready audio of the seperate drum instrument channels.
That is a topic which seems to be an endless journey to me, finding one solution and stepping into the next problem at the same time.
I have several drum VST's and as nothing is perfect (I like Grovve Agents integration in Cubase, SSD5's versatility and the sound of Abbey Road Drums in a messy Kontakt) but I am still considering buying more although I fear I will never be satisfied.
Thanks again, Dom! This was the clearest explanation I've seen. Having used Cubase all the way back to Steinberg VST on an Atari, I have some long ingrained habits and have been averse to switching from rack instruments to instrument tracks. You have led me to see the light!
Same here. Exactly.
Snap! Though I changed about 2 weeks ago. Haven't seen any CPU issues and the flexibility is so much better. Goodbye MIDI track you served me well :-) Actually MIDI tracks can be useful for sending controller commands to multiple instruments using MIDI sends.
Thanks Dom. Just learning Cubase and was wondering about this.
Yeah I would never use rack instruments like that now; namely because the best use of those additional midi channels in Halion Sonic is for layering different voices onto 1 instrument so it's like you're creating a new instrument. If you want to layer on non-Halion Sonic instruments it's honestly still easier to use a different Instrument track and just use a shared copy of the MIDI event so it's always in sync. You can use a group to apply effects or a VCA for volume and it's all still much easier than the rack.
In fact, a lot of users do not know the difference, thanks for your explanation.
I always battle myself with Groove Agent. I often debate whether I should record the midi on different tracks for organizational purposes.
Dom can you also create a tutorial to show us how to have multiple virtual instruments tracks with different midi channels that way we can play each instrument part simultaneously on different channels.
I'm migrating from Ableton, and it's pretty easy to route the Midi out of my Interface to the Midi in on my keyboard, I love the complexity of Cubase but damn it's crazy being a beginner with a DAW again. The struggle is worth it, but I definitely need an in depth tutorial on routing Midi data in Cubase.
I am on the same this days man. Personally i am going to open manuals pdf with cubase at the same time i think it's going to be so useful , videos of dom of course too useful haha.
Hi Dom!
In the main, I prefer to use Instruments as Tracks, but the whole 'Track v Rack' approach depends on individual workflow preferences. I know that for opening old projects and multi-timbral MIDI files (especially old GM Atari ones) you would be faster using a rack instrument to load.
MIDI tracks by themselves can be used to record Sysex data dumps from external hardware, to control external FX units with MIDI (such as Reverbs and Delays) and for triggering Plug-ins like Stutter Edit, Vocal synth and Break Tweaker from Izotope (to name a few). In addition to all of this, I understand that you can even control a lighting rig with MIDI!
Thanks for the video :-)
I have used midi track for triggering loopmash FX on audio or instrument tracks
Thanks Again Dom! Nothing beats the access to all the plugin power of an audio track over a midi track. Greetings from Oz!
I just went back over this a day later & assigning the kit to individual outs is a BONUS, thanks again!
I'm not used yet to Rack instruments. I must say it's a bit weird to me, but for now there is another reason I would use them above classical instrument tracks.
Simply editing in groove agent seems nicer for what I can tell.
There is a quick tip from Steinberg that demonstrate the difference between using an instrument track and a rack instrument when dissolving the part.
Instrument tracks dissolve the parts but duplicate the instrument. Looks like a waste of time and resources.
Rack instruments create only the routed MIDI tracks which is what you'd expect.
Maybe later on Steinberg would get rid of rack instruments or at least have the same behaviour when dissolving the patterns with instrument tracks.
Then I would not take racks as an option.
The only thing that is sometimes "frustrating" with instrument track is that when we mute it, it's for MIDI and audio. So, if I just want to live play to test something without hearing what was recorded before, I have to mute events ... with MIDI track linked to an instrument track, I can select the MIDI track, mute it and continue to play on this MIDI track, it works !
Definitely enjoy working with Instrument tracks more than MIDI for the same reasons you highlighted!
Wow. I had not even thought about those differences. I mainly used instrument tracks so far. I really like the tip about routing Groove Agent to different tracks. Thx Dom.
Heads up:
- for those who like to route the Groove Agent individual sounds to different output tracks (as shown in the video) ... This only works when you use drumsets from "The Kit Se".
- If you use a different kit (for example Beat Agent), then the "Export mixer to Cubase" will be missing from the menu.
Great video...it cleared up a lot. I tend to use the rack feature only when I'm starting a project and I'm 'sketching' an initial arrangement and just want a Halion instrument with a preset containing basic instruments; Drums, bass, piano, etc.. I then put the drums into Groove Agent and select different instruments and put them on instrument tracks. So I don't really use the rack feature any longer.
The one thing I was not aware of was the quick way you assigned the drums, with names, to each output. I always did the slow, tedious way. It made the whole video worth the watch. Since I've been using Cubase since VST32, I tend to miss some subtle improvements that are great time savers. Thx.
One thing Steinberg should change: Mute a MID-track and you still can play the instrument. Mute an instrument track and you can't. I often need this to just try a small change in a hook-line or checking out a different melody.
I have seen many videos on this subject, but yours is the best by far.
Great work Dom! I have used instrument tracks from the time I started using Cubase. As yet I have not had a problem with overloading the CPU at all. It is so much easier to use multiple instances of Halion or Halion Sonic or Groove Agent. There is too much to think about if you use midi tracks!
Thank you for this very interesting video. But I still don't understand the diff. between instrument track and rack...
Great explanation and video. I still use Rack Instruments when using Vienna Ensemble Pro because I like to freely move the outputs next to each one of my midi channels in the arrange window. The Instrument track locks the outputs all to it so the additional midi tracks (channels 2-16) can not have the outputs next to them (midi track 2, output 2, etc). I like to have them next to each midi track so I can do automation in the arrange window and such. I think when one just wants a single instrument, instrument tracks are great because the midi and audio are together as a hybrid but for VE PRO environments with multi-timbral it doesn't quite work as well for me.
I used to use midi chanels as an old user of Cubase. Thaks for the update!
Thanks! Since I started with Cubase many years ago I was using one instance of HAlion Sonic. Now based on your advice I' changing my project to have one per track. I also was using MIDI tracks and now using Instrument tracks. Great advice especially for Groove Agent SE outputs.
Dom, you’ve done it again. I know all about instrument versus midi track but..........reminding me of the GA mixer to Cubase that I had forgotten about. I gave myself a slap.
So many years as a semi pro dabbler and never really spending enough time on individual beneficial tweaks. At last I learn the right way. Cheers
Maybe years ago, our laptops did not have the processing power to handle all the info for each instrument. [It happened, and it was disastrous. Even if instrument tracks were muted, they still used processor power.]
I have a hoarding problem - in the house, or on my project page. I keep too many incomplete 'versions' muted, so now I have several pieces of one 'song' over several names, and have to drag them across from one project to another. I have severe ADD etc. and have to move on to a new verse, chorus, etc. before fixing the last one. There must be a better way to file things away from cluttering up my project page.
It's probably a matter of not having got to a stage of using a regular method to build with. It's also probably that I'm not a live 'muso' with a repertoire of styles, rhythms or instrument set-ups, etc.
So many tutorials to get through. I just saw your one for vocal harmonies - but I still have not even had time to use the new mike and work out how to record vocals without a sound-proof room, etc. Our local builders were supposed to build my studio months ago, but we now have a 'covid change' - a rush of people moving out of cities, making tradespeople scarce here.
So, it looks like I have no choice - there's no room to MIDI up all the old keyboards, etc. I have to stay in bed with a light MIDI keyboard controller and laptop! We've extreme 'social distancing' - but not one case of Covid locally. One step closer to a cyber-matrix world.
thank you! Been using Cubase for 20 years didn't realize the new way was better
One of the things thats good about pro tools, not many things (for composition), is that an Instrument track can have audio and midi from whichever track routed to that same track.
For me the "render in place" method is number one. But has anyone solved how to setup multitimbral hardware synthesizers as "external instrument tracks" and use render in place on the other parts than just the external instrument main track itself?
Thanks so much!........I need to watch this again.......and then again to absorb it all. I'm a real newbie to virtual instruments and midi. So much to learn!!
Love those little explanation bubbles
I started with Cubase with 8 and by that time the rack intsrument thing was already history really. I've always used instrument tracks and never had a problem so I've never used MIDI tracks really. I have 12 logical cores and rarely do they ever go over about 50% each, usually about 6-8BG RAM use out of 16. So no system resource issues. Where I do sometimes have a problem is with the real time processing meter hitting the buffers when I have a VST instrument track armed for recording. This is because I can't afford a stripped down dedicated separate machine for music production yet so there's often stuff running in the background that interrupts the CPU which then interrupts the ASIO buffer stream transfer. Overall system load is fine, maybe 30-40% but RTP hits the buffers. I can usually fix it by restarting my session from a cold boot, but this is a nuisance. If you got any words of wisdom on how to improve RTP, Dom, from my experience I'd say that's the biggest system resource problem running Cubase on today's computers. PC anyway, dunno about Macs. Also some VSTs seem cause much less load on RTP than others. All the Cubase stock instruments like Retrologue are super efficient. Some of the Arturia Collection synths are the worst. Serum and Omnisphere in the middle.
I know you're right on this, it just feels wrong. Example, i use xpand 2 to create a 4 part choir harmony. I'd much rather use 4 midi tracks, one instrument, and put them in a folder than use 4 instrument tracks. The biggest benefit of instrument tracks for me though is just not having to worry about midi channel assignments.
You can also route MIDI tracks to an instrument track. 😉
Very useful discussion on separating the output channels. Loved the Groove Agent Trick! Gold level content once again!!
Thanks for taking the time Dom!
If I need to choose an instrument I add a MIDI track, and different instruments or samplers. Then I can send my MIDI-track to any instrument and choose what I need.
Best 12 minutes ever!! Yup, I arrived at the same conclusion (instrument track), but what you really clarified was how the computer processing was great for that decision now. Ahhhhh.
brilliant video Dom. i've always wondwered about this. i use instrument tracks most of the time, and in fact midi tracks very rarely nowadays. Usually only use midi tracks if i'm importing a midi drum groove form a 3rd party developer - although not exclusively.
Useful vid, thanks. I use Cubase Elements for VSTi projects and use individual instrument tracks. I don’t really need any more than the 24 tracks included but could always use the midi track/rack instrument method if needed.
Hey Dom - I actually do use MIDI tracks for the reason of separation .. Let me explain, when I use a drum kit - I like to put each triggered sound on it's own channel so I can find it easily when making modifications later. I also use them to break up sections that might use the same instrument in some case in a different way; for example... If I was working on an all Piano track... I might separate my break it into parts where I would play the bass, and lead parts on different tracks. That's just me - maybe I will do a partner video to show what I mean and mention your video. If that is ok with you.
I only use multiple outputs when I use groove agent easy to keep the drums separated, still nice to know what and why Didnt know about the activate menu I was renaming them manualy! thanks for that
Cheers my man,always easy to understand.
Thanks Dom. I do use Inst track with software sinths except when i m doing arrangements and final instrument is not finally defined. It is easy for an A /B comparison.
I have not used external instruments since I got HALion back in 2001. I don't use midi tracks since "instrument tracks" were intruduced. Always hated the midi tracks getting in the way and had to hide them all the time in the mixer.
I got to talk to you Dom...
Thank you for this, Dom. I have had that question in the back of my mind about using MIDI tracks and since I don't use outboard instruments besides guitars, I now know I don't need to worry about it!
Hey thanks! Very useful and I enjoyed your style of teaching, and, like you, I'm gonna continue to use Instrument Tracks a lot!!
Always use instrument tracks, it is just most convenient for sound design and making changes in the mixing process.
Hi Dom , thanks for the great advice . I just realized that I uselessly use multiple rack instruments and probably waste cpu power. I am switching over to instrument . All you videos are really helpful. I use your tricks daily,. Thank you.
Makes sense. Thanks, Dom. However, as an improvising composer, I often take my separate MIDI track and create a score that I can edit. This is normally done instrument by instrument, of course. Perhaps I just don't understand how to get access to the Instrument Track MIDI?
This channel is so promising with good content. Intro gets me pumped up. Wish you good luck for your channel.
I used to use midi channels back in the Cubase 6 and 7 days. I still have projects like that and a couple of them have lost the sounds they were assigned to. I didn't know about the trick with Groove Agent se. I had been using loops then doing the resolved part thing.
The only case I use midi tracks for VST is when I load multiple kits in Groove Agent.
I send each midi tracks to it's own agent into Groove agent.
Fantastic Dom. Always learning new features and concepts with your videos and comments. Olé.
Really helpful tutorial thank you very much
That is the most amazing intro track I have ever heard dom! It’s so exciting everytime I hear it 😍. I’ve just gone back to Cubase after years and years of neglecting it. I’ve never been able to fully get my head around it but I must say you have helped re-enthuse me. I loved your cc121 video and I was torn between that and a keyboard controller. I ended up getting a nektar P6 and it’s taking some getting used to.....but I think it’s going to be really good. I have about three albums worth of unfinished stuff on nuendo 2 (yes lol) that I was going to re-do on an akai force! Instead I’ve managed to get Cubase 5 up n running so I’m bringing them all into that 😍
Really helpful and concisely explained. Thanks!
Don't know about anyone else but I revisited a session where I had used Lots of different midi channels in around 4 instances of Kontakt with lots of automation on the midi channels and it was a complete mess .... Although it maybe more efficient this way, its a lot harder to work with., I prefer to keep 1 instrument track (instance) per channel strip.
Thank you so much, Dom! Will be sticking to instrument channel. Seems the most logical for me at this moment.