It's good to have if you're a real, hardcore musician. I mean; if you're able to create something with multiple channels, why not do so?
Because there are other venues much more suited for big productions. I'm not dissing FamiTracker here, I love the program. I'm just saying that NSFs are a pretty unrealistic option for this kind of thing.
Besides, handling lots of channels is a tough task and a mixing nightmare. It's impractical, it's clumsy, and it's over-complicated.
Not to mention that using more channels for a song doesn't mean you'll get a better song out of it. As a matter of fact, there are more chances of making the song sound more difficult to listen to.
EDIT: I forgot to say: You don't need lots of audio channels to make the song sound big. Ask Virt, ask Danimal Cannon, ask Heos, ask... ask around. Most people who know their business will tell you.
Because there are other venues much more suited for big productions. I'm not dissing FamiTracker here, I love the program. I'm just saying that NSFs are a pretty unrealistic option for this kind of thing.
Besides, handling lots of channels is a tough task and a mixing nightmare. It's impractical, it's clumsy, and it's over-complicated.
Not to mention that using more channels for a song doesn't mean you'll get a better song out of it. As a matter of fact, there are more chances of making the song sound more difficult to listen to.
EDIT: I forgot to say: You don't need lots of audio channels to make the song sound big. Ask Virt, ask Danimal Cannon, ask Heos, ask... ask around. Most people who know their business will tell you.
Well put. This stuff is great for arpless polyphony, put any ppmck borne nsf I've ever heard with tons of channels ends up sounding really thin and tinny, even if mixed well. Not mixed well, it sounds almost like assy beeper music. We've got basic waveforms with comparitively limited volume control generating a ton of energy(RMS) unless having been completely anal with creating transient rich instruments. Mono certainly doesn't help, tho Chibitech found a neat solution for this by simulating the effects of sidechain compression across channels, which in turn simulates slapping a limiter on a master bus. Sidechaining aside, there's a lot mixing possibilities within simulating modest compression.
I think there's a lot of potential with this since FT makes it really easy to play around with clock speed, something that wasn't very practical with ppmck. Transient rich waveforms via high clock envelopes paired with simulated compression would really take the nsf format into some interesting new places. FT becomes an additive synth playground.
You guys do have a good point, but provided that a composer might not simply stick with the pure NSF format, he could, technically, export each individual channel into a WAV file & import it into a sound editor for remastering (which can include other effects not available in FamiTracker such as dynamic wave distortion).
And given the fact that FamiTracker now has more channels to play around with extra sine/triangle/sawtooth waves, post-FamiTracker editing can create endless possibilities when merged with more common forms of audio.
And yes; I know you don't really need any extra channels to make a song sound rich or full (it's all in how you manipulate effects & notes with a proper volume balance).
_______________________
Technology: the one thing that's hated & cursed at by all engineers, technologists, scientists & technicians!
First, the Japanese did not experience the arpeggio-intensive demoscene as much as the western world, and it still makes sense whatsoever to prefer 28-channel polyphony over 5-channel arpfest; it is like comparing streams to sequences, or samples to frequency modulation;
Second, the simultaneous use of multiple expansion chips <i>is</i> part of the NSF1 specifications. No post-editing should be necessitated for any combination of expansion chips alone;
Third, a lot of trackers easily have support for dozens of channels. I don't see how it is impractical since there are always ways to clean up the GUI, hide channels, implement frame editor in separate window, swap between channel positions, they all work well.
Just saying. Though I do not assume it will be an official solution, the NSF engine probably needs an overhaul.
Third, a lot of trackers easily have support for dozens of channels. I don't see how it is impractical since there are always ways to clean up the GUI, hide channels, implement frame editor in separate window, swap between channel positions, they all work well.
However, with the limited volume and (zero)panning control, you certainly reach a point of diminishing returns as far as channel count is concerned. These limitations leave you with way too much overtone activity and you have a hot, fuzzy mess with all those channels, no matter what you do.
Give robokabuto 64 potential volume levels and some panning to play with and just witness the tracks really punch you in the face. It'd lose the nostalgic nuances at the expense of clarity, but would still retain the aesthetic quality of the chips.
Quote:
Just saying. Though I do not assume it will be an official solution, the NSF engine probably needs an overhaul.
I really like this notion that nsf needs an update, as controversial as it may be. A lot of people have been expressing a thirst for progress lately. Nostalgia's starting to get boring
I highly recommend listening to Robokabuto's tracks in my NSFPlay beta with the per-channel panning settings going.
Also, gyms, there are two updates to the NSF format, but neither are about facilitating multi-expansion mixing. NSFE basically has per-track times and titles, NSF2 supports a couple of hardware features missing from NSF (e.g. it would be possible to make a proper rip of Battletoads with the PCM samples, or use Blargg's DPCM + IRQ saw wave) plus the possibility of other data like track times. The reason neither of these has taken hold I guess is that NSF can do like 99% of what it needs to do already. (I plan to support them both with NSFPlay in the future, though.)
Things like extra volume graduation will never be part of NSF. That just defeats the purpose of NSF. If you want a less limited format they already exist.
Things like extra volume graduation will never be part of NSF. That just defeats the purpose of NSF.
Right. And that's kind of what I was implying with my whole spiel up there. The more I learn about and come to truly appreciate strict 2a03 music for what it is, the more it seems like any expansion starts to ride the fence on serving that purpose. Expansions are acceptable, of course, because they were used in real games which demonstrate their practical real world application. As far as making nes music is concerned, they allow you to tinker around within a proven framework, which is what really makes it fun I think.
But multi-expansion is just a facet of emulation and really seems unnatural and almost pointless in that regard; I'd argue that robokabuto's multi expansion work sounds strange and unnatural(although I do appreciate and respect the craftsmanship). By unnatural, I mean that the music itself is almost saying out loud, "My complex sound is musically unsupported by the limited framework of my format."
Then how would you support it? That's what my earlier words were kind of addressing.
If we're going to progress into adding more channels, almost arbitrarily, then why not progress into supporting them with more volume control, panning and other features found in the more advanced sound chips that came later? That's the idea.
And this is the main thing I was getting at before; multi-expansion use itself almost defeats the purpose of NSF. But since it is a desirable thing, it seems logical that the next step would be to actually support it by addressing the fundamental sonic issues it presents.
Quote:
If you want a less limited format they already exist.
Yes, but the amount of control you have with tracking an emulated chip, nothing will ever give you the same sound. Even if you're not sticking to the original hardware specification, the nuances of the chip emulation still carry with it the target aesthetics of the sound.
The point I'm trying to make is that the interest in multi-expansion work itself, something that goes against the true-chip mentality, demonstrates a desire to do more with the format. Why not address the issues and support it? It certainly doesn't defeat the purpose in this particular progression of interest.
If we're going to progress into adding more channels, almost arbitrarily, then why not progress into supporting them with more volume control, panning and other features found in the more advanced sound chips that came later? That's the idea.
The only thing I agree. Actually before even venturing into out-of-NSF features, we haven't got the 2A03 hardware decay and a full 64-step VRC6 sawtooth and perhaps some FDS volume thing.
And no. This is in fact weaker than even systems with two AY-3-8910 chips.
Not only is FamiTracker an NES tracker, it is also an NSF tracker. The NSF specifications are there, and we exploit its transfinite possibilities. I suggest any of you to contact Kevin Horton himself for information on why the NSF is so unreal.
If we're going to progress into adding more channels, almost arbitrarily, then why not progress into supporting them with more volume control, panning and other features found in the more advanced sound chips that came later? That's the idea.
I understand the idea, but I will say that very little about the NSF format is arbitrary. Every bit of control that you have is there because a piece of hardware in the NES or a cartridge had that feature. The 2A03 doesn't have 16 volume levels because Kevin Horton thought it was a good limit, it has this because it's exactly what you can do on an NES.
The reason multi-chip expansion exists is because by some stroke of luck, the memory interface for all 6 expansions does not overlap, meaning that (at least in theory) you could map them all from the same monster cartridge. It's there because if you happen to implement them all in an NSF player, there is no natural impediment in any of their specifications to running them all at once.
The other thing is, well, the NSF spec does not actually describe the behaviour of any of the chips, and this goes back to almost nothing being arbitrary. The NSF spec is merely a container for working NES music code. ALL of the specifications about what the chips do comes from analysis of the chips themselves.
There are still some nebulous areas about what some of the chips do, and I can't begin to explain what a problem it is to try and deal with compatibility issues between players/emulators that presume different behaviour. The goal for all of them is to do what the hardware does, and while there are missteps along the way, eventually we work them out as the hardware becomes better known. To compound this problem with (extremely) arbitrary decisions about what else the hardware might to is kind of an absurd proposal. Do you want to argue with the commonity about where to put arbitrary extra bits for extra volume control? There's no way to arbitrate this because no-one has authority (not even Kevin), and the only natural thing to agree on is what the chips already do. A fantasy feature like this would simply be unsupported by all but one rogue player. There is no point.
Now, there are certain adaptations that can be made on the player side that are more or less external to NES/chip emulation, and those I think are fine to implement. This is why I put stereo controls in NSFPlay. It also has features to turn of Zxx pops, phase reset clicks, etc. which customize the sound output. These are all just player settings; an attempt to add them to the NSF format would be met with the problems of authority and poor support I mentioned above.
If we're going to progress into adding more channels, almost arbitrarily, then why not progress into supporting them with more volume control, panning and other features found in the more advanced sound chips that came later?
rainwarrior wrote:
The reason multi-chip expansion exists is because by some stroke of luck, the memory interface for all 6 expansions does not overlap
Ah, that does clarify some things for me.
This does go to show my ignorance for these kinds of details. I assumed that multi-expansion nsf support was some kind of glitch or fault in the emulation and people decided to just roll with it. The rest of my thoughts stemmed from that.
Has there ever been a multi-chip nsf, of any combination, played on actual hardware?
) 
