| |
Laurent
Registered: Apr 2004
Posts: 40 |
The 6581 has fallen into oblivion
Musicians seem to have lost interest for the original 6581 model.
Is it because of how it sounds (low resonant filter, combined waveforms) or because of technical challenges (huge filter variability between revisions, quality of filter emulation) ?
There are so many 6581 tunes that we all enjoy that I cannot believe the "sound" is the issue.
The funky distorted 6581 filter is absolutely loveable, not better or worse than 8580's, just different.
There are so many awesome examples..
Googachild
Miami Vice
4-Mat's Filter
Dirty Pair
Now that reSID is doing a good job at emulating the 6581, I assume one of last issues is that musicians have no guarantee that a tune will be played back correctly, with the same filter characteristics that they used when composing their tunes. Especially when they're played on a real c64, it just cannot work perfectly.
As of today, even if we wanted to organize a 6581 music competition it would be laborious to play them back correctly even with an emulator.
Some extra info could be certainly added to a .sid header, but running the .prg in Vice would be more difficult, the SID player init code would have to write to some unused SID addresses to "configure" the SID. This would please all emulators but would have no effect on the real thing (except if it had an emulated chip)..
Do you believe the 6581 is doomed ? :( |
|
| |
Jammer
Registered: Nov 2002
Posts: 1296 |
Some musicians prefer 6581 and tailor their tunes for it strictly. I wouldn't fear too much, tbh ;) Kinda contrived topic. |
| |
Mibri
Registered: Feb 2018
Posts: 209 |
One word: Mutetus. |
| |
DeMOSic
Registered: Aug 2021
Posts: 126 |
Its a shame the 6581 has a VERY small scene, but there are still some people using it (And more people should too!) The 6581s Filter Distortion may be sometimes a pain but use it good and OOOOOH your gonna get some good stuff!
Quoting MibriOne word: Mutetus.
Yeah Mutetus is the KING of 6581. |
| |
Adam
Registered: Jul 2009
Posts: 321 |
Quoting LaurentMusicians seem to have lost interest for the original 6581 model.
I haven't.
Quote:I assume one of last issues is that musicians have no guarantee that a tune will be played back correctly
this has always been an issue.
Quote:As of today, even if we wanted to organize a 6581 music competition it would be laborious to play them back correctly even with an emulator.
musicians can clearly state what model 6581 they used, or a compo could recommend what models of 6581 they should use. not complicated.
Quote:Do you believe the 6581 is doomed ? :(
how can it survive four decades and somehow only now in 2022 it's doomed? makes no sense to me at all. |
| |
Oswald
Registered: Apr 2002
Posts: 5034 |
Evolver |
| |
F7sus4
Registered: Apr 2013
Posts: 113 |
Quoting Adammusicians can clearly state what model 6581 they used, or a compo could recommend what models of 6581 they should use. not complicated.
It still doesn't help much since the filter offset discrepancies between each 6581 chip, even inside the same revision, are huge, to say the least.
However, R4ARs seem to be an exception here. No matter what their production date/batch is, they tend to sound quite consistent with each other (at least from my personal tests with 10+ different chips).
This is why I've chosen this particular revision while working on Timewarp Spheres [2sid]
Quoting deMOSicThe 6581s Filter Distortion may be sometimes a pain but use it good and OOOOOH your gonna get some good stuff!
The output sounds different depending on the current filter offset and current volume, which is quite unique feature. However, switching the filter mask on/off leads to awfully loud clicks, so it requires some additional planning to properly handle them. In the end, it's quite impossible to use the "usual" 8580 editing logic and expect simirarily successful results when conducting 6581.
Quoting Adam how can it survive four decades and somehow only now in 2022 it's doomed? makes no sense to me at all.
Exactly. As long as the very last ninja still uses 6581, it can't be claimed dead. |
| |
chatGPZ
Registered: Dec 2001
Posts: 11165 |
Quote:R4ARs seem to be an exception here.
its just yet another myth. the R4AR chips are not special in any way. |
| |
F7sus4
Registered: Apr 2013
Posts: 113 |
Quoting Groepazits just yet another myth. the R4AR chips are not special in any way.
I have considered that somehow all the owned R4AR used for tests might have provided similar filter sound by pure luck, but unless appropriate recording samples show up to prove otherwise, it's merely just a claim. |
| |
chatGPZ
Registered: Dec 2001
Posts: 11165 |
I'm more interested in what the AR marking actually means ("advanced resonance" is... just another myth =P) |
| |
hedning
Registered: Mar 2009
Posts: 4628 |
Spread Eagle Idaho. <3 |
| |
Hate Bush
Registered: Jul 2002
Posts: 459 |
from my perspective: one of the biggest challenges for a SID composer is drum programming. the 8580 filter is tailored to overcome that challenge (no clicks, lower values boosting freqs) and once you try it... 6581 is seriously crippled here and we're not talking tastes that differ, we're talking things technically impossible.
but if contemporary beats are not a priority, 6581 is still highly lovable. for that matter, many of us new SID embracers have done some tunes for it in the past. |
| |
MCM
Registered: Apr 2018
Posts: 21 |
There are still new demos with 6581 sid´s , just check out my demo for the last transmission party.
This one has a killer soundtrack !
RAGTAG |
| |
TheRyk
Registered: Mar 2009
Posts: 2111 |
Nothing to worry about, really.
One argument to use 6581 - against the trend - would be, if you do something for 6581 (apart from $D418 sampling which hardly anyone does anymore since late 1990s because there are better ways of doing samples/digis), it normally does NOT sound shitty automatically but mostly as intended on 8580. The other way round, if you used filters, chances are, it sounds borked on 6581.
So new SID is far more popular with composers for at least a decade or two, agreed. But there are still enough people who care enough to include a 6581/8580 detector and then adjust filter settings in Init, so it doesn't sound broken on 6581. And the more experienced composers are, the more they tend to do it if they think it's worth it. Also, there's a fistful of composers who stuck with 6581, either for traditional reasons or because of their own hardware. |
| |
spider-j
Registered: Oct 2004
Posts: 454 |
6581 has broken mixed waveforms and (kind of) broken filter – noone sounding like the other as already mentioned, and also that horrible crackling noise when you switch filters.
For me it wouldn't be fun to put work and effort into a tune knowing it may sound completely different as I intended at a party.
In 6581 land every SID seems to have different filters. In 8580 land only *one* SID has different filters: the one they use downunder at Syntax XD |
| |
chatGPZ
Registered: Dec 2001
Posts: 11165 |
heretic! |
| |
Oswald
Registered: Apr 2002
Posts: 5034 |
around 95-96 I sensed a revolution in sid music, before that I mostly knew late 80s SID music, apart from improved players this revolution in sound I experienced might be partly or greatly attributed to musicians moving to new sid. |
| |
anonym
Registered: Jan 2002
Posts: 250 |
Waz makes music exclusively for the 6581, too. |
| |
Laurent
Registered: Apr 2004
Posts: 40 |
Quoting F7sus4It still doesn't help much since the filter offset discrepancies between each 6581 chip, even inside the same revision, are huge, to say the least. Indeed. Like I was saying above that kind of info could be embedded in the .sid if there was enough motivation to extend the header.
But that still doesn't solve the issue where the tune is played within a .prg or others by a c64 emulator.
The few bytes describing the SID model for that tune would have to be stored in the tune data and somehow passed to the emulated SID during player init.
That's feasible, but requires the SID emulators communities to discuss and specify this, together with other things like addressing multiple SIDs.
On emulators we could then play 6581 and 8580 tunes made without compromise within the same multi-part/multi-tunes demo.
Sadly this would not work on a real c64 with a single real SID. |
| |
Krill
Registered: Apr 2002
Posts: 2871 |
Quoting LaurentSadly this would not work on a real c64 with a single real SID. Which sort of is the only platform anybody should care for, no? =) |
| |
chatGPZ
Registered: Dec 2001
Posts: 11165 |
Krill has leading |
| |
Laurent
Registered: Apr 2004
Posts: 40 |
Of course :-P but there is still value to store info that describes the SID model that was used to compose the tune, and I believe the best place is in the player, which would prevent fiddling around with the emulator configuration. |
| |
Laurent
Registered: Apr 2004
Posts: 40 |
Really, this is about improving things for the 6581. |
| |
chatGPZ
Registered: Dec 2001
Posts: 11165 |
You'd have to put much more info there than a byte or two - and then you couldnt reproduce it on a different SID anyway. So what is the point?
Different Guitar sounds different. Deal with it :) |
| |
Laurent
Registered: Apr 2004
Posts: 40 |
Maybe not two, but not that many, and you could also refer to a bunch of common 6581 presets with a single byte.
Isn't it valuable to know how a 6581 tune sounded like when the composer made it ?
I agree with you this cannot be known with a real c64/real 6581 but at least this can work with a c64 emulator or when using deepsid. Better than nothing.
Jakob's Lullaby #1
Jakob's Lullaby #2
Jakob's Lullaby #3
Jakob's Lullaby @ X2010
I assume the "good" one is the last one because it was played at X ? But maybe Stein composed it with another SID that sounded like #3 ? |
| |
chatGPZ
Registered: Dec 2001
Posts: 11165 |
Still not sure what the point is though. I listen to C64 music on my C64 - and then it sounds like it sounds. Couldnt care less about emu in that regard :) |
| |
Viralbox
Registered: Nov 2021
Posts: 15 |
Quoting Laurent
Do you believe the 6581 is doomed ? :(
No. |
| |
TheRyk
Registered: Mar 2009
Posts: 2111 |
Quoting LaurentReally, this is about improving things for the 6581.
All you can do to workaround against the well-known shortcomings is
a) DON'T use filters at all (you'd be surprised how many cool tunes from the 1980s don't)
b) use filters and then fiddle around with the registers as long as it sounds adequate on a wide range of ICs (or reSID emulation if you like, although some of the reSID 6581 examples are really extreme, did not come across real ICs who behave like that yet) |
| |
Conrad
Registered: Nov 2006
Posts: 839 |
Quote:b) use filters and then fiddle around with the registers as long as it sounds adequate on a wide range of ICs
That would be pretty straight forward for SIDs that use filters for leads or simple bass-line. For filtered drum-sets, not as much.
For the latter I would have code to overwrite parts of the wave-table/filter-table to make 6581-beautiful drum-sets, if one was to really make one's 8580 masterpiece 6581-friendly.
BTW, come back Soren/Jeff! ;) |
| |
spider-j
Registered: Oct 2004
Posts: 454 |
Quoting ConradFor the latter I would have code to overwrite parts of the wave-table/filter-table to make 6581-beautiful drum-sets, if one was to really make one's 8580 masterpiece 6581-friendly.
I doubt it is even possible to convert any 8580 drum sound with fast filter switchting to 6581 without loosing the characteristics that defined the sound in the first place.
You can adjust the values, but you can't reliably get rid of that horrible "pop" / "crack" sound when switching filter bands. It works for some combinations (I guess waveform and frequency does also matter here) but everytime I thought I had a strategy it turned out to be just lucky guesswork in a particular case.
It's maybe a compo idea:
"6581 heavy filterband switch without crackling compo" ;-)
The entry with the most switches and least crackling sounds wins.
Of course all participants would also have to send a 6581 around via snailmail so the results can then be recorded on that SID and be compared :-) |
| |
chatGPZ
Registered: Dec 2001
Posts: 11165 |
Reminds me of Cybertracker and how horrible "filter interlace" sounded on my C64 (with 6581) :) |
| |
Soren
Registered: Dec 2001
Posts: 547 |
Hi Conrad! ;)
Filtertype switching on 6581, as far as I remember.... There are ways of avoiding the clicking somewhat... Having 2 types active and only switching one of those off.... etc. I wonder if it also works with filtervoice control?
:) |
| |
Viralbox
Registered: Nov 2021
Posts: 15 |
Maybe you can just set the filter to the highest cutoff and the resonance to the lowest value when you switch instruments? |
| |
Viralbox
Registered: Nov 2021
Posts: 15 |
Quoting TheRyka) DON'T use filters at all (you'd be surprised how many cool tunes from the 1980s don't)
A favourite of mine is Orion by Jason Page. :) |
| |
Viralbox
Registered: Nov 2021
Posts: 15 |
Also Morpheus by Steve Turner. |
| |
acrouzet
Registered: May 2020
Posts: 80 |
Quote: Hi Conrad! ;)
Filtertype switching on 6581, as far as I remember.... There are ways of avoiding the clicking somewhat... Having 2 types active and only switching one of those off.... etc. I wonder if it also works with filtervoice control?
:)
I believe what causes clicking when changing the filtered channels is changing the actual number of channels the filter is active on, so as long as you keep the amount of filtered channels the same at all times you shouldn't get any clicking. |
| |
Jammer
Registered: Nov 2002
Posts: 1296 |
In general, most clicking seems to be caused by enabling/disabling bandpass - also on 8580 ;) |
| |
F7sus4
Registered: Apr 2013
Posts: 113 |
Quote: In general, most clicking seems to be caused by enabling/disabling bandpass - also on 8580 ;)
This is true for 8580, but on 6581 turning the filter mask on/off is just enough to trigger the clicking noise, no matter what the filter type is.
I didn't test acrouzet's approach yet, though I have a bit of doubt - even if it consistently worked that way, it would require an extraordinary amount of resources to become wasted only to keep proper filter ratios at all times. In the end, I think it's just easier/better to "hide" the artifacts among other sounds.
Quoting spider-jI doubt it is even possible to convert any 8580 drum sound with fast filter switchting to 6581 without loosing the characteristics that defined the sound in the first place. You can adjust the values, but you can't reliably get rid of that horrible "pop" / "crack" sound when switching filter bands
Ed did a lot of great work when it comes to progressive 6581 sound (and especially drums). |
| |
Stone
Registered: Oct 2006
Posts: 170 |
@Laurent: Jakob's Lullaby was made for 8580. The SID version in the file is wrong. I think the same goes for all the tunes in Another Beginning
Our latest demo Lifecycle was made for 6581, the grittier, the better ;-) |
| |
1BM
Registered: Nov 2002
Posts: 47 |
following this post i thought i would make some experiments with goattracker. what i noticed first is that the click has more amplitude the more channels (+ext as fourth channel) you feed thru the filter.
as most of my songs are more sound experinments than composition i decided to dig more into "6581 filter pop drones".
looping a simple lowpass on/off table you can create a boring buzz, but if you switch to multispeed (4x) it is possible to create proper audible frequencies.
i thought then i would use these buzz drones to create a song.
i used the G#2 note and 432Hz tuning, because i fancy it and it created less vibration than using 440Hz.
I played some pwm sounds on channel 1, also with a looped pwm table, to have some interference sound counterpoint to the filterclick drone buzz.
further experimenting with the duty cycle of the filter on/off relation it was possible to create 4 different nicely audible octaves of this buzz.
the highest possible frequency would be generated by:
00 90FF
01 80FF
02 FF00
then i started to play with duty cycles, this generates the lowest possible audible frequency:
00 90FF
01 0200
02 80FF
03 0200
04 FF00
from there you can change lines 01 and 03 to 01/01, 00/01, 00/01 etc etc. all of them create either raspy sounding drones or a very phased sound.
but also inserting the small break after the switch will sound different than a 50/50 duty cycle.
and not to mention, one could even set different cutoff values during the on phase and further mangle the sound.
this all was very exciting and i started to evaluate which musical note would result in a neat sound in combination with the filter pop drone. sure this is very minimal, one frequency only (unless someone writes a routine to modulate framerate?) and the resulting sound is more atari than commodore, but that does not hurt me. the great thing is that it is very loud, and loud also does not hurt me much.
but then i encountered a problem, else i would have already written a full tune with this technique and called out for a compo:
my filter duty cycle modulation tables do not work! they only work when i use the table starting from 00. if i copy the values to 05 and call it from the filtertable pos the drone does not work. keeping the values in table pos 00-04 and editing them on the fly produces audible results always.
and on the other hand, probably it will sound totally off on the real thing (hello filter variations... what will it sound like at all). sure it will produce audible results, but the timbre would be very varied.
if anyone could help me finding out why it behaves like that and how these modulations would work from any position in the filtertable? this really could be something for a competition and yet another way to exploit the 6581, and to pull it out of "oblivion". |
| |
acrouzet
Registered: May 2020
Posts: 80 |
4-Mat was able to create a pseudo 4th channel by modulating the filter-type and filtered channels settings very quickly, seen here in oscilloscope view: https://youtu.be/WKvJ3XkIz2Y
Keep in mind this is on 8580, so the clicking is much quieter here and is able to be mixed with the 3 SID voices better. |
| |
F7sus4
Registered: Apr 2013
Posts: 113 |
Quoting acrouzer4-Mat was able to create a pseudo 4th channel by modulating the filter-type and filtered channels settings very quickly, seen here in oscilloscope view
This particular trick was earlier used in the concluding part of Triangle My Amelie in the very same compo, but what is important is that it does not require the SID model to specifically be 6581. |
| |
Laurent
Registered: Apr 2004
Posts: 40 |
@Stone : Thanks for the info, who would have guessed ! |
