Pre-Release Feedback: Factories

The sound from my choosen figure disappear, when I go to the factory. How does I keep the sound when use the factory?

The same goes for the harmony that comes with the phrase/ figure. It would be nice if you could lock the hamony from the phrase you want to make a varity of.

It would also be nice if you had the ability to lock / preserve part of the Phrase/figure, so you got a variation simular to the starting point. Just an idea.

The same goes for the harmony that comes with the phrase/ figure. It would be nice if you could lock the hamony from the phrase you want to make a varity of.

It's been discussed in this thread, you have to use the Preview parameter. Load or drop an harmony in there and it will become yellow, meaning it applies to the whole pool. If you then want a particular phrase to have his own harmony you can use the Harmony parameter (green).

The sound from my choosen figure disappear, when I go to the factory. How does I keep the sound when use the factory?

to choose the instrument, or:

Could not find [image 2.jpg]

to use the currently selected instrument, harmony, tempo, etc.

It would also be nice if you had the ability to lock / preserve part of the Phrase/figure, so you got a variation simular to the starting point. Just an idea.

I'm pretty sure that you can because I've seen it in action in a factory example, however I'm not sure how it's done.

Kept playing with the factory a bit. It's even too many parameters XD On the other hand once you've found something that fits your goals, you can create tons of variations of it. So it looks like the workflow is making a lot of pools/factories that fits your genre/music, which will take a lot of time, but then with a single click you can create tons of phrases and just fine-tune some settings. And the part where you set up everything is the part where there is the highest human-input, which is better than e.g. full-A.I.-generated stuff.

I have a question about the "Glissando" type: it doesn't seem to do anything at all; in fact if you put it at 100%, nothing is generated.

Another question, what's the difference between e.g. Line.Bass -> Bass (Single) and Lines.Melody -> Generic Melody (Single)? They seem to do the exact same thing. If there is something internal that is done differently, it would be nice to have at least a summary of what that is.

I have played around with the chord master from parameters.harmony. Any progression I generate with it seems to use chords outside the scale. I thought the purpose of this factory would be to generate chord progressions, but none of them sound "good". I can edit the progression in the palette tab, but that defeats the purpose of using the factory. Do I just misunderstand the goal of this factory?

Hi. Synfire Kim seems locked on the following progression when using factories. How do I change this progression?

PS. The green switch to use the current arrangement is only available when using the arrangement library tab - rather than library view.

By using the "Preview" parameter, which applies to the whole pool. It behaves otherwise like the Harmony parameter.

Scriabiner, you're a star!! That works! Thank you.

It was actually explained to me in this very thread. ^ ^

I would like to start with some praise of the Synfire team: I feel this software nicely balances ease of use and expressiveness (possible range of musical results). It is not the most easy to use composition system, and neither is it the most flexible one, but I feel its design does a great job at balancing these two demands (which commonly contradict each other). Thank you for this! 

While playing around with figures, I found that I felt particularly limited by the current rhythm generation, and the bias of the system towards a certain range of rhythms. Here is an idea that could allow users to create a wider range of rhythms and also offers more control.

For generating the rhythm of individual phrases and beyond (step parameter), many factories offer two approaches that are at opposite sides of the spectrum of computer control vs manual control. Users can either specify the likelyhood of individual note values (1/4, 1/8...), and leave it to the computer to choose how sequences of note values form rhythmic ideas. Alternatively, users can have a very high degree of manual control over the rhythm by specifying a pre-exiting step parameter for all notes of the result to generate instead.

I think it would musically be very useful, if there would additionally be an option in between these two extremes. How about users could manually specify sub-sequences of rhythmic values (e.g., rhythmic phrases), which are then used as building blocks for composing the step parameter? If users can specify the rhythm of individual phrases, then users have more control over the musical result, e.g., by specifying one or more characteristic (or not so characteristic) rhythmic motifs. They could also group rhythmic values they want to stay together in a musical phrase, e.g., by specifying some rhythmic motif consisting mainly of short note values (e.g., for runs, arpeggios and similar figures), and another that also uses longer note values. This way, users could create and control rhythmic contrast in their results. With such an approach, users could then also freely use tuplets (not supported at all in the current step factories -- except when completely manually setting the step parameter), because it would be the responsibility of the user to ensure that these are complete.

In terms of interface, the sequence generation already supported for various other musical parameters in the factories could be used here as well. 
 

EDIT: the Step parameter sequencer is already close to what I suggest here, but instead of specifying multiple rhythmic motif (which could be contrasting), it generates a longish sequence from a single spec. But I could use this sequencer for generating pretty short figures and then arrange these manually.

I already collected more feedback and ideas on figures locally that I am happy to share later, but the point above is particularly important for me.
 

"How about users could manually specify sub-sequences of rhythmic values "  .   This is exactly what we need. 

> ensure that generated stuff is actually useful for music

Of course, that is always the main concern!

Something else: I struggle to understand the exact meaning of the factory Units > Spans settings. Just for testing purposes, I "froze" most musical parameters of generated figures: the rhythm (Step), pitch intervals (Type Runs with very short fixed Segments settings), and Pause. The only texting parameter that I wanted to make a difference then remained the Spans settings (Flow and Velocity can be ignored here).

With a standard 4/4 meter setting, I noted that with setting Span 4/4 to 100%, resulting short figures always started exactly on the beginning of a bar, and the resulting music was completely regular. With the 2/4, and 1/8 Span setting set to 100% respectively, the result was again fully repetitive, just with new figures always starting at every 2/4 or 1/8 Span.

However, results are different with the 1/4 Span setting set fully to 100%: there can longer rests happen between the resulting short figures, i.e. the onset of new figures can be delayed and the result is not fully repetitive -- in contrast to the other three Span sliders. Is this by design, and if so why only for a single slider? Or is this perhaps a bug?

I would argue that an irregular Span between the starts of consecutive figures can be very useful, musically, but I would very much wish that this regularity and/or irregularity can be controlled by users. To start with a musical example, the clave rhythm of Afro-Cuban music does not fit into the fully regular Span concept, Synfire factories seem to be based on so far (except for the above-mentioned irregularity, that I assume is just a bug). It would be nice, however, if factory users could generate such rhythms, that are irregular at a lower level, but then again regular at a higher level. Other musical styles also require a more flexible handling of Spans.

I noted that these Span settings are actually already a bit more flexible than it seems initially. When generating phrases with an odd length (e.g., phrases that last over 3 or 5 eighth or quarter notes), these Spans sliders no longer control the spans indicated by their labels, but rather different span durations. However, the resulting rhythmic phrases are still always of exactly the same overall length. What is missing is a way to create rhythms where the length of rhythmic units can change ove time.

How about it would be possible if users could optionally specify sequences of span values, where these spans could in principle be arbitrary rhythmic values? For example, these could be generated like number lists of other parameters that are generated by specifying a sequence of numbers in a first box, and some algorithm to process these numbers in a second box (e.g., keep, permutate...). It would then be possible, e.g., to create music that fits the accent structure of the clave rhythm or other irregular rhythmic formations.

>> if users could optionally specify sequences of span values

> They will certainly be useful for spans, too.

Thanks for considering it! 

> When you think about ideas for new factories, keep in mind that KIM is not a procedural language. It's rules-based

Sure, I appreciate that!   (worked on constraint/rule-based algorithmic composition before; different solvers, but I can relate.)

Two Factories (or amendments to current Factories) I would like to see added:

1. A guitar picking Factory. I like to be able to create guitar picking phrases similar to those inthe Example Phrase Library. The current factory that comes closest to this is maybe Apr One. But it's not quite the same.

2. I'd like a Factory (or an amendment to the current Piano Phrase factory) that would allow me to easily create bass & chord pedal point phrases similar to the "Rhodes G" and "Rhodes B" phrases in the Example Phrase Library. Perhaps there is a way to do that now with the current Piano Phrase Factory, but none of the experiements I've tried up to this point have been successful. If it is possible with the currnent Phrase Factory, I would love to see a tutorial as to how this can be acheived. Otherwise, it would be a great addition to the current Factory setup.

The factories are amazing and I love the detailed control available over them.  I haven’t found any big issues but wanted to mention a few things.  If they are bugs, they are minor and there are easy workarounds.  Other things may be “intended behaviour” but I find they interrupt my flow of thought a bit.  They certainly don’t spoil the utility and sheer fun of the brilliant factories. They are just things I noticed while testing. 

• If an arrangement is saved from the factory page and re-opened, the factory phrase pool appears empty.  You have to click elsewhere on the library and then back again to get the phrase pool to reappear. This can be a bit scary as it looks as if you lost all the phrases.
• Is “System.core” needed on the menu of factories as there is “no user interface” and, to the user, it just seems to generate silent phrases?
• Deleting a phrase causes playback to start on the next available phrase.  I find this unexpected.  It happens even when the last phrase in a pool is deleted (the orange playback progress bar starts moving silently).  Harmless but distracting. 
• Clicking back and forth between phrases sometimes causes an audio “glitch”/click (at least on my system).  Could this be cleaned up?
• Your changes are lost if you make changes to a factory phrase, click on another factory phrase and then click back.  Easy to get round by generating a new factory phrase but you have to remember to do so!  Is it worth putting up a warning box when clicking away from “dirty” factory phrases?  Or maybe changing it in some other way?  (By "Dirty", I mean settings changed and no new phrase generated with the new settings.)
• Some details of Segments cannot be changed when this should be possible. For instance, to change the Transformation setting (“Keep”/”Pick” etc), you have to change the number sequence first.  Similarly, if you tick the Tails box, you can’t change the number sequence.  You have to generate a new phrase and then you can change it for the new phrase.
• There seems to be a logic bug when part of a segment appears to the left of the anchor (for instance when Tails is ticked).  The signs of the Heads number sequence are reversed so the phrase goes up when it should go down.
• There seem to be a couple of bugs with “Start Heads from Anchor Backwards”.  The Start position is off by one grid unit.  Also, the number sequence is used in a strange way – not simply reversed.

[Settings were Lines.Melody/General Melody (single) set to Step 100%1/8; Spans 100% 2/4; Types 100% Runs; Variations 0%; Similarities None.]

Something I meant to raise from the beginning: in Factory results, the Preview harmonic rhythm ignores any custom meter (Metrics).

Update: I manually adjusted the Preview harmonic rhythm to fit the meter, and for later generated examples that manual edit is preserved. So, if one knows about such workarounds, it is not too bad...

Great that meanwhile all the controllers can be added to library figures. Thanks!

Accidentally deleting a figure in a library cannot be undone?? I was editing some figure in various ways, and tried to delete a bunch of selected symbols (selected with span selection, perhaps that is not the right way) with Ctrl-X, but that instead deleted the whole figure -- seemingly with no way to undo that.

Accidentally deleting a figure in a library cannot be undone??

Nope.

Where is the parameter "preview". Plz  send me  a jpg

Sorry my bad english

Thanks