Toy Story's Hardware Defying Music - How We Did It 

Having programmed through all these years the changes have been incredible. Almost everything I learned is now wrong. I remember when memory was precious and fast, running synchronously, or almost, with the CPU. Lookup tables could speed up complicated operations (like here the pitch and volume LUTs saved a number of divisions and multiplications), but each LUT took up a lot of your precious memory (only handful of megabytes!). Now you have all the memory you could want to make look up tables, but memory is glacially slow and lookup tables blow your CPU caches, so it's faster to do complex math than look up a number in a big array! I remember when software implemented fixed point math was faster than even the built-in FPU math on a 486 (let alone simulated FPU math on a 386). Now floating point math is much faster thanks to better FPUs, pipelining and SIMD instructions (SSE). I remember when colour change effects were practically free (because you could do them by changing the on-screen palette), then they became really expensive when lost the palette in exchange for true-colour images, and then we gained GPU shaders and now we can do more calculations on every single pixel on screen than a game used to spend to draw the entire frame... A prototype I made back in the DOS days (that unfortunately never turned into a full game): ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-d6gLhPaLnAs.html I pulled so many rabbits out of my hat to make that happen. Probably half of my available memory was used up for various look up tables and caching the lighting to make it all happen...

@@tylisirnthe lighting effects really are impressive for a DOS game.

@@tylisirn These guys are just doing reactionary comments because they think modern workers are lazy you are bringing actual experience lol

@@tylisirn that looks awfully a lot like a simplified version of Alien Breed?

@@BradleyQuerruel The graphics were very much inspired by Alien Breed, you are correct. That's one reason why I never moved further than prototype stage (the graphics weren't stolen, they were custom drawn, but heavily inspired by.) The lighting system is much more complicated than AB ever had, but the gameplay and of course levels are mostly not implemented.

This is back when alot of licensed games were actually GOOD games.

@@KGR7Gamer Yeah, that's gonna be a no from me. They were decent at best.

KGR.7 haha noooo there were tons of bad licensed games even back then. You never heard of LJN?

@@retro_boy_advance oh yeah LJN no thank you. Yeah looking back at my comment a lot of licensed games were horrible but there are some pretty great licensed games from that time.

it's why the lego games took off

I'm pretty sure it would have been Jon and/or his cohorts who figured this out. If tricks like this were in the documentation from Sega, the Genesis would likely have had a lot fewer garbage-sounding titles.

*the engineers at Traveller's Tales

I'm watching this, and I'm thinking why isn't this guy (and the hardware engineers) curing cancer?

@@Electrify85 Completely different field of expertise?

@B3ro1080 Sonic 3D Blast was coded by the same very guy who's speaking on this video....... it was also made by Traveller Tales, as Toy Story was.

its been nice going through them as he reuploads them!

a lot of it gets caught in your memory buffer and by the time you've come back your brain has fully uploaded the data into a data bank that is easily addressed. I think if we upgraded to USB3.0 we would allow a higher bit rate transmission. Sadly I think our brains would then need upgraded to handle USB3.0.

dude TL;DR: BLAST PROCESSIN

Olosnah Crafts and More Personally I would love to one day upgrade to an NVMe brain interface but for the foreseeable future I will have to be content operating on a 56k modem instead.

Nvidia brain

you are a legend sir!

Is that a reference to a Sega Genesis-esque remixing channel?

You didnt have to make that reference.

@@lev7509 No it's a gimmic Sega used to sell to kids. "Blast Processing" Which isn't a real thing, was just saying that the sega games moved faster then nintendo. (Sonic)

@@John231984 DF Retro Extra: Sega's Blast Processing Was Real - But What Did It Actually Do? ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-rvvL6S5Buiw.html

Yeah now you just throw some shitty css into electron and tell people to install more ram :/

now it's just GrApHiCs which always end up looking outdated a few years down the line whereas a more stylised game will always look good

The current consoles are PC architecture because devs blamed the exotic hardware for not producing a quality game. But today we still have bad performing games(crashes, 20FPS, missing textures, etc.) - overall unenjoyable experience. Still even today there're some really good devs out there who are able to create amazing games on this limited hardware. As a gamer I wish more of them to show up and raise the bar.

If games today often are poorly optimized it's because the effort it takes to optimize them will only reduce loading times by a few seconds. More effort can instead be spent on the game which is why a lot of the best games made today are far superior to these old shitty games that were made in the 90s.

The one issue I see are games that take up half of your computer's disk space. Like wtf is all that shit, uncompressed textures??

@@DanIel-fl1vc is this even bait anymore

Yay. Nice to see another Jazz Jackrabbit fan in the wild. It was the first game I ever played as a kid. The music is definitely the best thing about it.

In this scenario having a much larger sound buffer that was dual ported would make all the difference. But that would have cost a lot of money in 1989

@7MGTESupraTurboA I can certainly agree that they did the right choice. It was more a "what if" these days since it's a fairly simple processor to transplant and hack into systems. One can even remap the address base for internal devices in them. Quite fun little processors

@MageLocc Question: How much would that have increased the price? Not sure if that can be verified or not.

You should ul some of ur music if you've not already

AtGames hates that stage because its too complex LOL

sounds Great!...Wish they'd spent a little more time on color choice. However, level design, gameplay, and animation look superb!

Good question

It’s only on menus - the actual music during gameplay is awful in comparison.

@@Jimmy42222 Or nonexistent. This game really pushed the Genesis to the max and it shows.

I think it was using most (if not all) of CPU power back then, so answer is most likely no = )

they sorta did the "debunking" was a nintendo ad

The nes hasn't got an independent cpu to drive the sound chip. Nor any independent sound buffer as far as I know. It would be driven directly by the main cpu and essentially streamed right from the cart or main memory block.

There is indeed a homebrew soundtrack cartridge from a couple of years back, that does exactly that! I can't remember what it was called, though.

It's kinda a mess to do this on NES because you only have one CPU and the playback mode which automatically streams data doesn't use a format that's conducive to adding samples together. You'd end up having to manually write all the bytes, and it could probably be done for the case of just playing music with carefully written code, but running any other game code would probably mess up the timing.

Actually, there are a few games that do 4 channels of pcm during gameplay. Skitchin is one example from back in the day, and nowadays you have games like Xeno Crisis (and probably some others that use SGDK). They aren't playing amiga mods though.

@@jc_dogen Pretty sure Skitchin uses 1 or at most 2 channels of PCM. The rest of the instruments are standard Genesis FM synthesis I think

@@CodingSecrets You could be right, I'm only passing on what I've read. But I do know SGDK does have a 4 channel PCM driver, with some limitations.

@@jc_dogen Just had a quick look at SGDK and it can do 4 channels of PCM but you are right about limitations as no pitch changes, so useless as a tracker. I'd imagine all the samples have to fit in the 8K Z80 memory too, but I might be wrong

Savaged Regime himself made a video on the subject of Mega Drive PCM quality: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-2W4yTL-9gZE.html You could also go the route of Titan Overdrive 2 by playing separate high-quality samples in sequence without any sort of channel trickery, but not everybody has eight megabytes of storage space to burn (with the majority just for sound). The YM2612 was very much a stripped-down version of the YM2608, and lost its ADPCM and most of its PCM capabilities in the transition. As to the Amiga, it was a great sounding system, but having two left and two right channels is a bit of a bind - it reminds me of 60s stereo recordings where singing would come from the left, and drums from the right.

like wise I loved that back in the day

what company were you in?

@@xddstudiosbywr3cked845 Apache Software, making games for Sega Europe

in HolyC xD

I agree. The videos are brilliant, but the background music is far too loud.

@@fabrice143 ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-BAVqOBCHxkM.html Star Glider 1986. Jez San created the Super FX as well for nintendo

Atari ST for sure. We were even doing these tricks on PC (one bit channel used with timer PWM and software mixed to 3 or 4 channels) well before the Genesis / Megadrive came out. The Book Musical Application of microprocessors by Hal Chamberlin (1980) showed how to do such tricks on early microprocessors such as the 6502.

Sound is a waveform. That waveform moves a speaker back and forth quickly to produce a note. If it moves twice as quickly the note is higher. That’s why you can see bass sounds rumbling your speaker - because the movement is slow because the note is low.and so you can perceive the movement. High notes move the speaker much quicker so you can’t perceive the movement. But you’re right about the tempo changing - in as much as you’d use up the data describing the sound more quickly. If it was a sample of speech the pitch would be higher AND the speech would be sped up.

@@CodingSecrets Aaah, I see now! I was definitely aware that shorter waves = higher pitch, but I hadn't recognized the data as being directly representative of the waveform the speaker would be producing. I thought each point was it's own note, and the song generated just so happened to have a wavy look to it! Thanks for the explanation!

@@CodingSecrets Hey Jon, could we get some Coding Secrets on the second Toy Story game? The draw distance is quite good for a fifth gen title and on the N64, it runs very well, something quite unusual for that system.

The processor runs much faster than 60 interrupts per second = )

Is that a reference to a Sega Genesis-esque remixing channel?

Using a separate wave table for every pitch will make playback code simpler, but will either limit the variety of pitches one can use or massively increase memory requirements. An approach I used on the 6502-based Stella's Stocking was to use twelve wave tables with lengths from 32 to 60, but unroll the playback loop 4x, allowing the pointer to be advanced by 1, 2, 4, or 8 after each group of four samples, and allowing the four samples within each group to be played from the same sample, or two from one sample at two from the next, or from four consecutive samples, or from samples skipping ahead by 2 or 4. This made it possible to generate 61 pitches using 12 wave tables. Trying to include fine pitch shifting on the 6502 would have made four-voice music at a 15.75khz sample rate impossible, but on the 68000 there are some tricks to greatly facilitate it.

Those are stock wires from the factory. Most revision boards have them. No real idea what they do.

Nah, you know what the console is running at by the region you set in the ROM. Just need to adjust the scan line position of the interrupts.

This is how I feel about "streamed" (or "pre-rendered" as I call it) music. I kind of hate how those boring, basic sound files have taken over in most modern games. Even if it sounds "cheap" at times, I find it way more impressive when I listen to sequenced music in older games (or in a lot of cases, DS games) and think about how it's playing every single note in real time instead of lazily going through a single sound file... I don't like how CDs and stuff made it easier for developers to rely on their DAWs and recordings, playing them during gameplay instead of programming the game to do it manually. Some CD-based games still used sequenced music, but it was usually due to technical limitations, such as having to stream other data like voice samples or level data in the background; To think streamed audio would've taken over sooner if CDs weren't so slow to access back then... Surprisingly, there are games on the PS2 and Gamecube that used sequenced music for seemingly no technical reason, and at times, it sounds amazing, almost pre-rendered sound quality... It annoys me to think that sequenced audio was starting to sound really good, then that *stupid streamed audio* had to ruin it. On PS2, _Graffiti Kingdom_ (level music, not bosses for some reason) and _Harvest Moon: Save the Homeland_ have sequenced music, and they sound really nice! On Gamecube, they're _Harvest Moon: A Wonderful Life_ (most songs sound cheap and "MIDI", but Breeze is good), _Pokemon Colosseum_, and _Tales of Symphonia_ use sequenced MusyX music. Personally, I want to bring sequenced music to a platform that seemingly doesn't use it, Roblox, but I haven't worked on my library enough...

Maybe

@@user-rl5zr1cg1o Indeed. On a side note, I recently saw LexSNES' posts about how he's using the SNES' HDMA to stream in 32 KHz PCM samples on the fly. And he even thinks there's a potential he might be able to push the system to be able to stream multiple 32 KHz PCM samples like this, even possibly up to eight of them, or something like that. I dunno the exact details, so just trying to give an overview. Look up LexSNES RU-vid channel and his latest video if you'd like to find out more about that.

That's what they do, now isn't it? The first thing that's done is combine the 4 sound channels from the mod-file to 1 channel.

new channel for cs only