Well I've been Kontakting away here. The Strings, Woodwinds and Keys are done. Brass up next and eventually percussion.
http://bigcatinstruments.blogspot.com/2 ... ntakt.html
The SFZ is available from the mothership.
http://vis.versilstudios.net/vsco-community.html
Note that you might want to try both. I'm a bit obsessive sometimes these days about tightening things up, so the Kontakt is more an instanter response, while the SFZ takes its time. There is a bit of different feel.
VSCO 2 for Kontakt
-
- KVRian
- 1367 posts since 30 Jul, 2013
-
- KVRian
- 1094 posts since 13 Mar, 2008 from Arnhem, Netherlands
Would changing the Attack parameter in the GUI get the SFZ versions feel back or did you alter the samples themselves (cut attack portion off)?bigcat1969 wrote:Note that you might want to try both. I'm a bit obsessive sometimes these days about tightening things up, so the Kontakt is more an instanter response, while the SFZ takes its time. There is a bit of different feel.
-
- KVRian
- Topic Starter
- 1367 posts since 30 Jul, 2013
You could go change the start sample times, but the gentleman who made the orchestra is going to come out with his own Kontakt version which I'm sure will have a similar feel to the sfz version. Its probably just worth waiting for his official Kontakt version.
-
- KVRist
- 286 posts since 9 Jun, 2015
If you use Zampler as a SFZ player there's no point of changing anything when you can adjust ADSR values with it.evilantal wrote: Would changing the Attack parameter in the GUI get the SFZ versions feel back or did you alter the samples themselves (cut attack portion off)?
For example, I loaded the entire Sonatina symphonic library into FL Studio, all instruments are played by Zampler, I did some tweaks here and there and then saved the project as template.
When I'm in the mood to play around with it everything is set.
-
- KVRian
- Topic Starter
- 1367 posts since 30 Jul, 2013
Its all about latency and response times. So suppose you have a good computer and lots of memory (samples in memory are far better than disk streaming btw), you might have say a 20 millisecond delay. So from the time you hit your midi key to the time the sample starts playing is 20 milliseconds. Now imagine the sample has 20 milliseconds before there is any audio. Your delay has just doubled. This is why I just edited every sample in the Salamander piano (which took many hours) to take it from something like a 20 millisecond delay on most samples to about 5 milliseconds.
Now imagine a string sample that takes 250 milliseconds to swell to any significant volume and 2 seconds to reach max volume. Do you simply allow 250 milliseconds of something close to silence? Now imagine the sample below it has only 70 milliseconds of 'silence'. So if one plays a G2 there is, sort of, 90 millisecond latency and G#2 has, sort of, 270 millisecond second latency which is long enough to blink your eyes. Now imagine that these are the soft notes in a two velocity layer instrument and the loud notes have much faster attack rates as higher velocity layers usually do. When you hit your midi keyboard at 81 (128 max) pressure you get a solid burst of noise in 40 milliseconds, when you hit it at 79 pressure you get half as much noise for the first second and almost no noise at all for a quarter of a second. You are making a Kontakt instrument, what would you do?
Attack in ADSR only cuts off the front of a sample, it doesn't make the sample start faster than the sample start point in Kontakt, so 0 attack and finely tuned powerful 5 millisecond latency computer mean little if the sample has a 200 millisecond delay.
https://www.presonus.com/community/Lear ... io-Latency
Note I'm not saying anything negative about the VSCO samples or the Salamander samples. Just explaining the thought process behind certain decisions.
Now imagine a string sample that takes 250 milliseconds to swell to any significant volume and 2 seconds to reach max volume. Do you simply allow 250 milliseconds of something close to silence? Now imagine the sample below it has only 70 milliseconds of 'silence'. So if one plays a G2 there is, sort of, 90 millisecond latency and G#2 has, sort of, 270 millisecond second latency which is long enough to blink your eyes. Now imagine that these are the soft notes in a two velocity layer instrument and the loud notes have much faster attack rates as higher velocity layers usually do. When you hit your midi keyboard at 81 (128 max) pressure you get a solid burst of noise in 40 milliseconds, when you hit it at 79 pressure you get half as much noise for the first second and almost no noise at all for a quarter of a second. You are making a Kontakt instrument, what would you do?
Attack in ADSR only cuts off the front of a sample, it doesn't make the sample start faster than the sample start point in Kontakt, so 0 attack and finely tuned powerful 5 millisecond latency computer mean little if the sample has a 200 millisecond delay.
https://www.presonus.com/community/Lear ... io-Latency
Note I'm not saying anything negative about the VSCO samples or the Salamander samples. Just explaining the thought process behind certain decisions.
-
- KVRist
- 44 posts since 7 Aug, 2013 from Connecticut, USA
(for the record, I'm not planning on making an official Kontakt version, just the sfz and MAYBE an education-centric VSTi)
Basically Bigcat and I follow two different ideologies regarding samples. Both are perfectly legitimate mindsets, and both see regular practical use in the commercial end of the industry.
My philosophy is "laissez-faire", or 'let it be'. In the real world, no instrument on earth comes in instantly- they all have a regular wind-up period as the lips open, or the bow accelerates, or the wind reaches the end of the pipe. Even pianos and harpsichords have small wind-ups as the hammers transfer the mechanical force into the strings, and come off, or as the plectrum catches on the string before reaching the point when it can no longer pull the string and releases it. I seek to include as much of that in my samples as possible, without including needless blank space ahead (in fact, the samples are cut so that an identical amount of wind-up is included per instrument). Everything about the samples should be left as natural and untouched as possible, aside from critical tuning and denoising. Wavers, bits of background noise, and other elements only add to and enhance the sense of realism (this philosophy being developed by Spitfire early on; I don't know if they still go by this, but they used to have a big long page about it). The future of this is in new approaches to dynamic crossfading without requiring phase-locking, new methods of improved destructive editing, and ambisonics, which I feel is going to revolutionize the sampling industry in the next few years.
Bigcat's philosophy (and please, correct me if I misunderstand) is that the computer can simulate many of the elements of a performance, including attack and the general response of the instrument over time (with adsr and other elements). Therefore, the sample should be cleaned up and 'purified' to the highest extent possible to allow the computer to do all the work regarding performance, allowing for higher responsiveness and greater playability, at the possible expense of realism. The logical extent of this approach is synthesis, where all factors of performance can be exactingly controlled. However, a certain medium approach between sampling and the capabilities of the CPU (as outlined in Christoph Hart's paper here- http://hartinstruments.net/technology.php) is arguably the bright future of this approach. This angle also ties into 'human playback' or 'automatic performance'-type instruments and is very present in the solo instruments end of the market. The future of this is in granular re-synthesis, synth-sample blends, and human-emulative interpreters.
As I said before, both approaches are equally valid, and manifested innumerable times and in many different blends between the two across the industry.
Now, to examine usage-
In my approach, YOU serve the SAMPLES; an intimate knowledge of the samples is required to produce the most realistic sound. Take the cello vib sus patch in the community edition sfz's for example. Play a quiet velocity- notice how the sound softly grows from nothing and sits at a gentle, sweet timbre. Every note at that dynamic level was conducted and precluded by a specific instruction to play sweetly and gently, even bowings and breathing were designated with this in mind. Now play a loud velocity- notice how it bites into the note, as if in the midst of a climactic frenzy; the vibrato is more intense, the bowing is stronger. This was instructed to be 'fiery' and 'passionate', with a strong down-bow at the start. Now, you might think, "ok, when I want a loud sound I press a loud note, when I want quiet, I press gently," but this is not entirely true. Each dynamic layer is a different emotional timbre as well as a volume. You can get by with a quiet dynamic layer in a loud section even. You will find yourself automating the velocities and note lengths a lot after performances to get 'that sound'. Think of this like being given chunks of a once-beautiful marble statue, so that you might reassemble it using glue (reverb) and fasteners (phrasing).
In Bigcat's approach, the SAMPLES serve YOU; you really don't need to know much of anything about the samples to make a good performance. Like pads, just load up and play, shaping the sound as YOU see it using the controls and modulation parameters to taste. In this angle, you have to generate the emotional factor yourself; failing to do so (very often the case with inexperienced composers) will result in an emotionless performance. A great deal of automation is required in order to create a realistic performance, including regular automation of ADSR parameters, volume, and so on. Think of this like being given a rough-hewn block of marble in the general shape you want, so that you might carve the sound you desire using a variety of tools (eq's, envelope, reverb, spatial manipulation/stereo field, etc.).
Hope this helps explain some of why these two versions exist, how they differ, and what it all means for you.
Basically Bigcat and I follow two different ideologies regarding samples. Both are perfectly legitimate mindsets, and both see regular practical use in the commercial end of the industry.
My philosophy is "laissez-faire", or 'let it be'. In the real world, no instrument on earth comes in instantly- they all have a regular wind-up period as the lips open, or the bow accelerates, or the wind reaches the end of the pipe. Even pianos and harpsichords have small wind-ups as the hammers transfer the mechanical force into the strings, and come off, or as the plectrum catches on the string before reaching the point when it can no longer pull the string and releases it. I seek to include as much of that in my samples as possible, without including needless blank space ahead (in fact, the samples are cut so that an identical amount of wind-up is included per instrument). Everything about the samples should be left as natural and untouched as possible, aside from critical tuning and denoising. Wavers, bits of background noise, and other elements only add to and enhance the sense of realism (this philosophy being developed by Spitfire early on; I don't know if they still go by this, but they used to have a big long page about it). The future of this is in new approaches to dynamic crossfading without requiring phase-locking, new methods of improved destructive editing, and ambisonics, which I feel is going to revolutionize the sampling industry in the next few years.
Bigcat's philosophy (and please, correct me if I misunderstand) is that the computer can simulate many of the elements of a performance, including attack and the general response of the instrument over time (with adsr and other elements). Therefore, the sample should be cleaned up and 'purified' to the highest extent possible to allow the computer to do all the work regarding performance, allowing for higher responsiveness and greater playability, at the possible expense of realism. The logical extent of this approach is synthesis, where all factors of performance can be exactingly controlled. However, a certain medium approach between sampling and the capabilities of the CPU (as outlined in Christoph Hart's paper here- http://hartinstruments.net/technology.php) is arguably the bright future of this approach. This angle also ties into 'human playback' or 'automatic performance'-type instruments and is very present in the solo instruments end of the market. The future of this is in granular re-synthesis, synth-sample blends, and human-emulative interpreters.
As I said before, both approaches are equally valid, and manifested innumerable times and in many different blends between the two across the industry.
Now, to examine usage-
In my approach, YOU serve the SAMPLES; an intimate knowledge of the samples is required to produce the most realistic sound. Take the cello vib sus patch in the community edition sfz's for example. Play a quiet velocity- notice how the sound softly grows from nothing and sits at a gentle, sweet timbre. Every note at that dynamic level was conducted and precluded by a specific instruction to play sweetly and gently, even bowings and breathing were designated with this in mind. Now play a loud velocity- notice how it bites into the note, as if in the midst of a climactic frenzy; the vibrato is more intense, the bowing is stronger. This was instructed to be 'fiery' and 'passionate', with a strong down-bow at the start. Now, you might think, "ok, when I want a loud sound I press a loud note, when I want quiet, I press gently," but this is not entirely true. Each dynamic layer is a different emotional timbre as well as a volume. You can get by with a quiet dynamic layer in a loud section even. You will find yourself automating the velocities and note lengths a lot after performances to get 'that sound'. Think of this like being given chunks of a once-beautiful marble statue, so that you might reassemble it using glue (reverb) and fasteners (phrasing).
In Bigcat's approach, the SAMPLES serve YOU; you really don't need to know much of anything about the samples to make a good performance. Like pads, just load up and play, shaping the sound as YOU see it using the controls and modulation parameters to taste. In this angle, you have to generate the emotional factor yourself; failing to do so (very often the case with inexperienced composers) will result in an emotionless performance. A great deal of automation is required in order to create a realistic performance, including regular automation of ADSR parameters, volume, and so on. Think of this like being given a rough-hewn block of marble in the general shape you want, so that you might carve the sound you desire using a variety of tools (eq's, envelope, reverb, spatial manipulation/stereo field, etc.).
Hope this helps explain some of why these two versions exist, how they differ, and what it all means for you.
Founder of Versilian Studios
-
- KVRian
- Topic Starter
- 1367 posts since 30 Jul, 2013
Elegantly said. Actually I didn't realize that you weren't going to do a Kontakt version. As such after I get done with this version and get back from my vacation, I'll probably go back and toss up a set of nkis with the start points set to zero. Then folks can use whichever they please. Of course if I had been clever I'd have saved the nkis before I ever changed the start points...
-
- KVRian
- 1094 posts since 13 Mar, 2008 from Arnhem, Netherlands
That would be awesome!bigcat1969 wrote:Elegantly said. Actually I didn't realize that you weren't going to do a Kontakt version. As such after I get done with this version and get back from my vacation, I'll probably go back and toss up a set of nkis with the start points set to zero. Then folks can use whichever they please. Of course if I had been clever I'd have saved the nkis before I ever changed the start points...
-
- KVRAF
- 4315 posts since 31 Oct, 2004
Very interesting! Thank you for taking the time to write that, I learned a lot as I'm more into sampling synthetic sounds (currently making my own electronic devices). Very informative post!Samulis wrote:(for the record, I'm not planning on making an official Kontakt version, just the sfz and MAYBE an education-centric VSTi)
Basically Bigcat and I follow two different ideologies regarding samples. Both are perfectly legitimate mindsets, and both see regular practical use in the commercial end of the industry.
...
-
- KVRian
- Topic Starter
- 1367 posts since 30 Jul, 2013
Alright I took a vacation from my vacation and whipped up the 0 version of the strings as in I did zero to the samples or the start points.
http://www.mediafire.com/download/o8zwz ... trings.zip
http://www.mediafire.com/download/o8zwz ... trings.zip
-
- KVRian
- 1094 posts since 13 Mar, 2008 from Arnhem, Netherlands
Thanks for that, Joe.
Gonna compare them and see which one fits best to my needs and workflow...
Gonna compare them and see which one fits best to my needs and workflow...
-
- KVRian
- Topic Starter
- 1367 posts since 30 Jul, 2013
Took a shot at the percussion...
the normalized edition
http://www.mediafire.com/download/oh918 ... ussion.zip
and the untouched edition
http://www.mediafire.com/download/xjn0l ... ussion.zip
Creative Commons Zero 1.0 Universal : Versilian Studios / Sam Gossner
the normalized edition
http://www.mediafire.com/download/oh918 ... ussion.zip
and the untouched edition
http://www.mediafire.com/download/xjn0l ... ussion.zip
Creative Commons Zero 1.0 Universal : Versilian Studios / Sam Gossner