Prototype Through Production: Pro Guitar in Rock Band 3

Hello. Okay. All right, so we're going to get started. As a reminder, um please uh mute your cell phones, and I think that's it. Fill out the evaluations. Fill out the evaluations at the end, please. Thank you. So, yeah. Um this talk is about uh Pro Guitar uh feature in Rock Band 3. Uh my name is Jason Booth. I'm the senior technical designer at Harmonix, and uh I was responsible for designing and prototyping uh the feature. I'm Sylvain Dubrofsky, um senior designer at Harmonix. On Rock Band 3, I was in charge of gameplay systems. Uh one of the things I did was uh work on Pro Guitar and strike team lead for most of the project. Okay, for those of you who don't know what Pro Guitar is, first let's check out this video.

All right, so thanks to Fender for that video. Um this talk is about Pro Guitar in Rock Band 3 and our experiences making it. It was a big ambitious feature, and we learned a lot of things in its production. So, first I want to talk about why we made this feature. Um as a little bit of context, in around in 2008, music game sales were around 16% of North American video game sales.

Um we knew we had to innovate for our hardcore players looking for a new challenge. Um and we knew that uh being able to play real guitar in Rock Band 3 is a good extension of Harmonix's mission, which is to teach non-musicians the joy of um making and performing music. So, you might have heard about this in previous Harmonix talks, so I'll go quick. Um this is the one question. The one question is a question you can ask about any of your features in your game to see if they make sense. Um you can use it to align your team. Uh Guitar Hero's one question is is it rock? So, you can see like uh iconic characters with their own over-the-top special moves is a good answer to that question. When it came time to work on Rock Band, Harmonix added features like character creator with user expression, um being able to have more realistic graphics, putting tattoos on your guys. Those are all good answers to the question of Rock Band, which is is it an authentic band experience? When it came time to work on Rock Band 3, you can see features like real instruments, we added pro keys and symbols to the drums, pro guitar and harmonies are all good answers to the question is it an authentic band

experience? So, the first period we're going to talk about is the first 3 months. Um this is where we had a group of senior developers and management to talk about the feature. Yeah, so I started thinking about um this problem space, right? Of uh well, what is it we're really trying to do? Are we trying to, you know, teach somebody how to play guitar who's never played one before or is it this really about uh playing Rock Band with a guitar? Um and when I started thinking about this, I realized that they they don't actually intersect that much. Um and to compound things, uh the sort of feedback that we're going to need to give, um the information we're going to need to display actually radically changes as somebody becomes more comfortable with the guitar. Um So, right off the bat, we uh established our target audience as being sort of the hard to expert Rock Band players, the people who've been, you know, playing a a of our game, but they didn't necessarily have guitar experience, but were interested in learning. And we set a sort of modest goal of like, "Hey, you know, if these guys are able to sort of play a campfire version

of songs at the end of this, then we we've kind of won, right?" If we can get them to play, you know, a couple song version of of you know, a couple chord version of some song, then then they're going to feel really good and maybe they'll go on and learn more about guitar and music. And sort of our approach was to use this sort of ramp we have in our games where you start out on the easy and you're playing a very simplified version and you work your way up to a much more complex version over time. Another sort of initial decision was we have this great illusion in our games. The first time you play, you know, the first time you play Guitar Hero or Rock Band, you hit some silly little gem, you know, representation on this guitar and then this huge sound comes out of the speaker. That's really rewarding to players. So, rather than, you know, the traditional instrument learning experience where the first sound you make actually sounds really horrible, we were going to use that illusion. So, the first time you get something right, you got that big sort of feedback. Okay, so there's a lot of ideas and there's a a large possibility space at this time, but the problem is, you know, Harmonix

is a big company and we have a lot of projects and at this point we didn't actually have coding resources to get things in the game. So, what we did was we explored the space as much as we could, you know, with designers and artists, design docs, pictures and video. I'm going to show you some of that stuff right now. So, as a quick primer um for those of you who don't play guitar, if you're looking at a guitar neck, it has these raised metal objects called frets. Um you push the string down behind the fret and then you pluck the string with your other hand and that actually will make what sounds. And as you move your finger along different frets, you'll actually make different types of sounds. On a Rock Band guitar, this is actually replicated by the five buttons. They represent five different frets. So, when you're holding the second red button, you can see that that is replicating holding down behind the second fret and and picking a string. So, we call this fret relative. As you move from left to right, you're actually moving up from the first to the fifth fret. And these things coming down the track, which we call gems, are colored

to represent that. When you have to hit that as it crosses the bar at the bottom, you're hitting a fret. So, this red one coming down, that's the second fret you'd have to hit. Um and time is the next axis. Um one of the things we uh wanted to experiment with first is the idea of fret relative because like millions of our players have been playing uh this game and we kind of baked this knowledge into them. We thought this might be the first thing we should try. So, here's an example from an early design document. You can see in the top um if the player was to be playing a real guitar um in this example, the third position represents the third fret. So, the player would be holding the third fret with his left hand. And each of these gems have a color on it and the color represents which string you'd have to hit. So, in this example, um I'm holding the third fret and on the green one I'd have to hit the lowest string. On the red one I have to hit the next string up from that. Uh here's another example of that where color and position represent which fret you'd have to hit. Uh the string position that you can see on the gems is represented by stars in this example. Oh

or actually there's numbers that are not represented there that we could have tried. So, we quickly moved into videos. Um here's a 3D Studio Max mock-up mock-up that our artist made. Um it's always good to do a Max mock-up with real data. In our case, this is songs. Can you turn it up a tiny bit? Can you guys hear that? So, this Max mock-up you can see um kind of was hard to understand and kind of hard to read, right? Um it's a pretty standard song and there's a lot of stuff going on and and a lot of this convinced a lot of us that maybe fret relative was not the best approach. So, then we started thinking about the idea of string relative. If you turn the guitar neck 90°, instead of looking at the frets, you're now looking down the fretboard. Um, you'd have your lowest strings on one side and your highest strings on the other side. Gems coming down, instead of telling you coming down what fret they would on, they would come down and tell you what string you have

to play on. And something about the gem would have to represent what fret to play instead of what string. So, let's look at one example here. This is an early art mock-up. It's It is actually pretty confusing, so let's talk through it a little bit. On the right hand side, you can see each of these gems are coming down on a unique string, meaning you'd have to strum that string when it reaches the bar at the bottom. And there is a light on it that have tells you relative position of actually which fret you'd have to hit. Kind of confusing. On the left hand side, there's the idea of what a chord could look like in this. Um, we still hadn't quite nailed that. Then we put this into video. Um, this is an actual song in motion. There's no sound on this one, but you can see that there's a lot to process this here in in real time. And this is kind of scary for us. Uh, we were happy to try string relative, but that wasn't maybe super successful. But then we got to this one right here. This came towards the end of our period here. And you can see like with that same real data we had before, this is looking a lot more promising. This is actually closer to what we end up with. Chords aren't quite figured out, but

notes and and their relative positions are a little bit easier to understand in this way. So, this video gave us a little bit more confidence um coming towards the next period. In fact, uh, with this, we were able to work with management and senior developers in the company to kind of come up with our minimum goals and constraints for the system. Pro Guitar had to display notes and chords on the track. It had to be pitch accurate, means you never play the wrong note. We want to teach people the right thing. Um, it had to target non-musician Rock Band players, as Jason said before, people that were looking for a new challenge. It didn't necessarily have to work with bass. It didn't necessarily have to work with all the Rock Band 3 songs. We were targeting maybe say 20 of the 80. Um and it didn't have to work in all game modes. It could be its own menu option off the main screen if we needed it to be. So, our takeaway here is um when we we had uh low resources, uh we used whatever we could to actually learn about the problem space. You know, getting things in the game is always best, but if we can't, we like video and pictures and

text are also useful to find more information. So, the next period we're going to talk about is the prototyping period. This lasted for about 7 months of the almost 2-year period. So, the problem space here is we need to iterate really fast. We want to get things in the game to try it. And we had a large group, and when you have a large group, there's a lot of disagreements about what you want to try and get in the game first. So, one of the first things um Dan did as strike leader at this point in time was reduce this group to a a smaller size. Um we added Jason Booth and Mike Monahan as Jason was uh our coder on the team. Uh Jason is a multi-talented coder that can do anything we needed him to do. Jason could do graphics, gameplay, menu work, and Mike is a similar as an artist. He can do 2D, 3D. These guys could do tons of iteration really fast. So, I went through this a little bit already. Like the best part of uh small strike team is like the implementers are the people doing the work. Jason and Mike are able to do all that.

I was able to contribute wherever I can, uh whether it's meeting notes or lesson designs or song ideas or um art mock-ups. Another thing is we were able to have more focused meetings. We get things in the game, meet about it right away, and decide a course of action after that. You know, if we don't if we didn't decide on the course of action, a lot of times our philosophy is like just try the fastest solution, and we'll learn something about it. We also were able to sit closer to each other with a smaller strike team. Um Jason sat really close to Mike uh like three desks away, and later in production, I moved into the room with the coders in the team. Um one thing I forgot to mention is in case you're not familiar with the term strike team, it's kind of like Scrum, although our version was a little bit less less strict than Scrum, but you can you can imagine it's a group of people focused on the feature like music library, or in our case Pro Guitar, Pro Keys, stuff like that. All right, so now I'm going to talk a bit about how we actually went about prototyping this and coming up with the various design solutions we did.

Um so the first thing that we noticed you know I can't we rolled onto the project and there were all these mock-ups and stuff, but we didn't really have any hardware and I kind of looked out into the landscape of you know possible solutions for getting data out of a guitar and getting that as MIDI or something like that. And all of these solutions have various problems. For one they're all pretty expensive, they have latency, they have accuracy problems, there's all these different types of issues. Um and rather than try to tackle any of these things right away, we basically just decided you know what, we're going to use whatever works best, not worry about it right now, and we'll use the information we gather to inform our hardware department of you know what we need out of a guitar. Now I have to point out that building hardware is not exactly an easy process. It's very complex, very risky. So you know if we could have done this by just plugging in a guitar, it would have been our priority. But for now what we did was use some off-the-shelf stuff to figure out what we really needed. And so the first question that I set out

to to answer is okay, how are we going to communicate what to play to the player? You know they they may not know anything about music notation and that's really kind of the the crux of you know the first half of the game is communicating to the player. Um so for this we decided to take sort of a Montessori approach which is you know you in Montessori you know kids are basically playing with models and various things and they're learning a lot from this and they're learning actual you know sort of factual things but not in a in a very direct way. And so, our our thinking here was like, if we can get your hands into the right position, if we can get you to play a G chord, then it doesn't really matter whether you understand, you know, that that's a G chord or know any of the music theory about it. It's just about getting your hands to do the right thing. And so, that was really our focus is like, let's get their their muscle memory built up, let's get them doing the right thing, and not worry about anything else. Um now, we wouldn't present anything wrong, but, you know, we would uh just focus on that. Um so, I started thinking about this idea of like, okay, we're going to need to notate this to the player, and I started looking at traditional notation. And um traditional notation is kind of

interesting because it was designed for uh conductors who had to fit an entire orchestra onto a page. Um and to do this, they came up with various forms of compression to fit everything onto a page. And one of these is the idea of key signature, which is that um the player has some knowledge of which notes are going to be played within a um you know, a possibility space. Uh and then you don't have to notate that stuff. And I was like, oh, that's, you know, that's kind of interesting. Well, kind of what we're facing is a compression problem, right? Like, how do we take all this information and get somebody to recognize it um right away. Now, most guitar players don't actually even read traditional notation, or at least not very well. Um they tend to read tablature a lot. Uh the reason is is tablature is physically centric. It tells you exactly where to put your hands. On a guitar, you can actually play the same note in multiple places. Um so, tablature is a way to to say, "Look, put your hand exactly on the 12th fret." Um Uh but, it has other problems. It doesn't work with chords very well. It's kind of hard to read. When you get a lot of stuff, it doesn't represent rhythm well. Um and so, there's actually a third type of notation that guitar

players use, which is chord charts. Uh these are great because they show you kind of a layout of what your hands are supposed to do, uh but, they only really work for chords. Um So, often you find guitar notation is written in these three different forms, sometimes all at once. Uh and there's no way we can really ask people to to learn all these three notation systems, uh or, you know, uh uh, understand them um in real time. And so we knew right off the bat we're going to have to invent some new form of music notation. And what we wanted was something which was physically centric like, uh, tablature is, um, but it would need to have some form of compression for dealing with chords and and sort of complex, uh, musical passages. Um, and we knew from our experience with Rock Band and Guitar Hero that, you know, we had a very good concept of time, um, to give to give users. So, I started thinking a lot about the chord problem and sort of what I was doing when I was playing a chord. And, uh, you know, I started thinking about a D chord and I'm like, wait a minute, a D chord, you know, it kind of makes a shape, right? Uh, with my hands I'm making the shape. So, maybe if we we took that and we

actually, uh, you know, map shapes, uh, into gems that would come down the track, then the player who had already seen these shapes would be able to recognize them. And so when you see different, uh, chords mapped out like this, Mike would sort of, you know, sketch these things out and it was like, oh, okay, I can see a unique signature for each one of these. I can see how all these chords look different. Um, additionally when we started messing around with like variations, like these are all of different variations of major minor A chords, um, they actually all produce unique shape signatures. So, when we started messing with this, we thought, you know, this is this might be kind of the key for compressing all this information down for the user. Um, so, uh, we also started thinking, okay, well, I can play the same chord in multiple places when you think about it as a shape. Uh, and so we're going to have to number that, uh, somehow to let the player know where to anchor that chord. Um, and, you know, we were also looking at music in general. There's all these other opportunities for compression, right? Uh, music tends to repeat a lot, so you're going to have that sort of, uh, compression of repeating riffs. Um,

there's various playing styles of, you know, strumming the guitar loosely like Bob Dylan. Um, you know, you don't necessarily need to notate every note in that. You just need to notate, okay, he's playing this chord and here's the rhythm he's playing. And then there are things like arpeggios. And all an arpeggio is is it's a chord that's played one note at a time instead of all together. So, we're looking at these things and you know, my basic thought was there's a lot of opportunities for uh sort of compression within our notation here. Um and Mike would sketch some of these out. Like here's an example of, you know, how we might represent arpeggios, um which is pretty close to what we finally uh shipped. And so, here we have uh essentially a D chord and then we're just sort of lightly notating which strings to pluck when. Um we even played with things like changing the plane that these exist on to sort of create mode switches in the player's head. Uh So, um you know, while while we were thinking about all this stuff, uh we still had some some sort of lingering questions that have been uh coming up. Uh one of the things for me was, you know, we'd dismissed string uh

sorry, fret relative and sort of decided that, you know, we had this chord language and this string relative system to work with, um but we still have this question of like horizontal versus our perspective track like we have in our games. And my initial thinking on this was, "Hey, you know, all music notation is horizontal. We read horizontal. There's got to be something about that that's actually better um for people." And uh when I made the first version of the prototype, I made it visually agnostic so we could sort of plug in new models and then just see how they worked. And uh this is an example of the horizontal one and it broke down uh right away for a couple major reasons. Uh one is that when things are scrolling horizontally um or on any flat plane, they're going to move uh many more pixels per frame than when they're coming at you. Uh they tend to move in very small deltas when you consider the pixel motion. Um so, it's actually easier to read details on them when they're coming at you. And additionally, this idea of this chord shape language just broke down in a horizontal mode. Um you can see Mike was working various sketches trying to figure this out, um but it there just

wasn't a really great way to do it. So, as we got the prototype up, I needed some songs. And um you know, when when people talk about prototyping, a lot of times uh you'll hear them say you have to tackle, you know, all your hard problems first because otherwise, you know, you'll never get to them or whatever. And I I think that there's some truth in it, but um but I think you have to make sure you don't uh just do those things. I think sometimes you need a prototype to build momentum for your project, right? You need to build some success to get other people bought in, to get them psyched up about what you're doing. So, when it came to song selection, which is really um you know, a core component of what we do, I chose songs that were were definitely going to work. They were going to be easy enough to play for a new player. Um they would be simple enough that when you were playing a simplified version of them, you still felt like you were playing the the full version. Um so, things like I Love Rock 'n' Roll and Twist of Cain. Twist of Cain doesn't have any solos in it. It's really powerful-sounding song, but it's actually very, very easy to play. Um so, rather than, you know, jumping right into Crazy Train or something, I I sort

of focused on these. Um additionally, Mike and I uh just came up with a system, rather than coming uh or sort of a convention to solve the the Chord Gem problem. Rather than uh actually building a system which would, you know, graphically build these elements, he would just model the ones that we needed for a song. So, you know, I Love Rock 'n' Roll has three chords in it, so he would go model those uh chords, and that saved us from having to solve any of these problems right away. Um so, uh at this point, we had a a pretty good uh prototype built. Um and uh you know, had some some uh so most of our answers kind of answered here, um but we still had a had a bunch uh more after this stage, which is uh things like, you know, okay, great. We have this chord shape language, and you know, I can recognize that that's a D chord, but if I've never seen it before, I've no idea what it is. Uh so, we're going to need to teach that to players. And to translate that shape into something they do with their hand, there's going to be some information we need to give them to walk them through that process so they can really understand what they're supposed to do with their hand uh from this notation.

And then there's going to be a bunch of other things that you need to really learn a song, right? Uh it's not just about learning a single chord, it's about learning a song in the end. That's what people want to do. They want to be able to play a song. Um So, at first, you know, our attitude was wow, we don't want to make a big training edutainment program, right? Like we want to make something that's fun and exciting. Maybe we can uh sneak this education in along with our our game somehow and inline the training right into the songs. Uh but this proved really jarring for people. Um we tried things like stopping, you know, and sort of bringing up uh menu displays and stuff. Um So, we switched this model of upfront learning, which is okay. So, I Love Rock and Roll's got three chords in it. So, let's make them actually learn those chords one step at a time and then they'll uh proceed and actually try to play the song using what they've learned. And uh as we were mocking this up and and uh implementing it, you know, we kept thinking that we needed this this virtual hand, right? That would show the players what the what their left hand was supposed to be doing. Uh but what we

found out through playtesting was that nobody actually looked at the hand. What they really needed uh to know was just what is my you know, what is each finger supposed to do uh in simple text and then some feedback about what their hand was doing. So, in this video, you can see me sliding around my hand and sort of lighting up when I hit the right notes, um you know, and telling me whether I'm right or wrong and showing me where my fingers were. And this sort of proved to be the the most valuable port uh part um for players actually learning this for the first time. Um so, we also uh started prototyping a trainer. And now, when I think back to learning music, uh you know, it's not like I the first song I learned I I learned it in real time playing along to the record, right? I sat and I like, you know, learned how to play that one chord and then I figured out what the next three notes were and kind of kind of went in whatever time it took for me to uh figure that out. And so, we tried to imitate this by having a what we called wait mode, which is where the game would just sort of wait for you to you know, get the right answer and then it would proceed on and it would bring up various

help for that. Uh and additionally, it would let you work on certain riffs of the song and sort of practice them at slow speeds and work them up to speed, which is how you actually learn when you go uh learn a part. So, um at this time uh we started doing a lot of user testing and we had a pretty good idea of how we were going to answer all these questions. Um And this is the actual video of the uh the prototype running. Um so, you can see me going through the chord book playing the the various chords of the song. And one of the problems that becomes apparent is if you have a lot of chords in the song, it gets really tedious. Um And then here's a little clip of the prototype. And there it's waiting for me cuz I missed that note. So, yeah, that was that was kind of uh where we got to on the prototype. Um And we were doing internally play test this with uh various people in the company who had little to no uh uh

experience. Um and we would always ask them, you know, okay, so you know, at the end of this we'd be like, well, how fun is this? And and they'd sort of look at us weird and be like, I don't know if that's really the right right question. It's super compelling or encouraging or something like that. Um they were really excited about the feature. Um but at first they weren't even sure whether fun was really a word that could be applied. Um but we sort of believe that, you know, the secret to our success of of actually teaching people to play guitar was was rooted in that idea that you know, this could be a fun experience instead of a grueling experience. And so, we kept working on little details. Um so, like one of these details, uh we kept having this problem of people not knowing what their left hand was doing. When they were um you know, when they had the the help up and the the guitar was there and it was showing them exactly what they were doing, the feedback was really good, but we didn't want that up all the time because it was too much information and too distracting. Uh so, we kept trying different things, you know, to sort of show them what their left hand was doing it ways and sort of failing and then one day I was watching uh crazy Japanese game

shows and it was the human Tetris wall, right? And the the wall is coming towards somebody has a hole cut in them and the person has to get into the shape of the hole and get through. And I was like, you know, that's kind of what we're doing in inverse, right? We we have the shape coming towards you and we're asking you to get your hand into that position. So maybe what we can do is is sort of generate a wall based on what your hand is doing and then the shape can slot through that, right? And it'll give you some feedback. And that's where this idea of the position wave sort of started. Um So when the prototyping was coming to a close, um it was pretty obvious there were some inherent limitations to what we're doing. Um you know, our notation system had a lot of upfront learning. Uh this chord shape solution was certainly interesting, um but it didn't it didn't work well when you had, you know, uh songs with lots and lots of complex jazz chords and things like that. It just sort of became hard to read. Um we had tons of screen space issues. You could never make the numbers uh large enough and things like that. And as a result of all this, some songs

would work really well and some songs wouldn't. And there were still a bunch of problems uh both from the guitar technique standpoint and just sort of general problems that we wanted to solve. Um and we sort of realized that, you know, a lot of these problems aren't solvable, right? They're not totally solvable. What we can do is chip away at them. We can make them better. We can try to make the learning curve smoother, but we can't necessarily solve them. Um So meanwhile, uh from all this, we had a pretty good idea of what we needed out of hardware. Um And it was pretty apparent at this point that this idea of knowing what the left hand was doing was really going to be key and we're going to have to manufacture some type of hardware. And so really what we wanted was we wanted to know exactly what the left hand was doing. We needed that to be low latency. We needed to be very high in accuracy so the players wouldn't get frustrated when they hit a note and it said they didn't or things like that. And finally, it needed to be low cost, right? And it's a really hard challenge. Um so what our hardware team decided to do is sort of uh go on two directions, right? So the the the the real thing we wanted to do is

was have a real guitar, right? Like a real Fender guitar you could play and you plug it into an amp and it could be um real, but we also realized that there are problems with this. You have additional problems that come into play uh like tuning and string breaking and your hands hurting. Uh you know, there's a lot of extra complexity there. And so for some players, you know, some sort of plastic controller would actually be a lot better um as long as it was still representational of a real guitar. Um and on top of that, you know, there was no real known technology to do this with a real guitar and we had no idea what it was going to cost if we did figure it out. Um whereas, you know, uh sort of our button approach, we we had a pretty good idea of, you know, how that would be built and what it would cost. Um so you know, we decided to pursue both of these directions. Um and started working with Fender um and eventually went down there and built a first prototype of this uh string controller with them. And it was really exciting because even though the thing only worked up to the fifth fret, to actually like play the

game we were building with a real guitar was like a real uh amazing moment for all of us. And so we settled on this two-pronged approach. We're going to have this 22-fret uh real guitar which actually shipped uh this week. And uh you know, you can plug this thing to an amp and play it. It's it's it's a 100% a real guitar, but you can also play the game with it. Um and it doubles as a really amazing MIDI guitar. And then we would have a lower cost approach which was this um sort of button guitar uh which we would probably be able to manufacture by the time we ship the game and uh was a much safer and more cost-effective route uh for many people. Um so from this I just wanted to bring up a few sort of prototyping suggestions. Uh you know, like Sylvan uh said, I think really reducing the team size down is is crucial. The less people, the less uh arguments and random random sort of discussions you need to have to keep everyone in sync. So, it allows you to move really fast if you can get a very small number of people to prototype a feature. And really the cycle that I like to think about is, you know, you get together, you come up with these core

goals, like what is it you're trying to build and you know, what is really the the the target audience, things like that. And you sort of establish these constraints. And then you just go, you know, hack it in any way you can, try all the different variations of things that might work. And then when you sort of think you have your answer, you put that in front of actual users and you see, you know, I would have users come in and they would play and they would learn this thing and then I would give them a real guitar and I'd be like, can you play that chord on a real guitar? And they would sort of like sheepishly take it and then hit that chord like they had this amazing expression on their face because they just learned how to play something on a guitar, something that seemed impossible to them, you know, from playing this game. So, that was really exciting. Um and the other thing is like, you know, you're going to have these constant arguments about various features or different ways you can go with people coming up is a sign that you need to address them. You need to try something. So, we kept having these discussions of fret relative versus string relative and horizontal versus perspective and until we actually like put those into videos or, you know, actually put them in the game in some case, we couldn't really

solve those arguments. But as soon as you do, all of those discussions go away. So, if you if it keeps coming up, don't even bother having a discussion, just go start implementing it and figure out what's going on. And then finally, when you're prototyping, you know, a lot of times people are so focused on the problem that they're trying to solve that they forget all the little things that make make a prototype or a game polished and make it feel good. And often what happens is when you present this to somebody, all they notice is those things you didn't bother to do, right? Those little things that you're like, oh, that's easy, we'll get to that later. Sometimes it's worth spending the time to to actually do those things early because it lets people really get to the meat of your your solution and really see what's going on. So, all right. So, let's talk about the last period. This is the 13-months of production. Um at this point, the prototype branch was pretty hacky. Um we couldn't keep adding features to it, and we had to bring it over into the real Rock Band 3 branch. But, at some point during the prototype, we got really confident. Um You You saw the play test results. We were really excited about this feature,

and we decided to aim higher than our original goals. Um we We decided we'd make the the system work for guitar and bass for every song on all four difficulty levels. Um Easy difficulty level would be playable without training. It would just be notes. And expert difficulty would be every chord and note in the song. As crazy as that sounds, even if that's really, really hard for people, even if the jump from hard to expert is um a little bit higher than it is normally, we thought that was a a good goal. Um this would work in all modes in the game. So, you can jump right into quick play next to two other tracks or three other tracks for seven players. Um this feature is the same as every other instrument in the game. We would do trainers for every song on guitar and bass for four different difficulty levels. And we'd also do general guitar lessons, what which would teach people actually theory and things that they would have to learn to play the game at all four difficulty levels. Which means that although we solved a lot of issues in prototyping, that we still had a ton of stuff to figure out. Um Some of the stuff we had to figure out was just Rock Band standard stuff. We

had to figure out how to make work in this world like solos and peak states and um uh stuff that you're used to that we'd have to make look right. And then other things that involve guitar techniques. So, if we were going to make this work for DLC in the future and all the songs on the disc, we'd have to solve a lot of those issues so these songs are playable on expert. So, things like arpeggios and loose strumming like Jason mentioned and hammer-ons and slides and um there's There's dozens of these things that we had to address. Um finally, there's the issue of uh song authoring. We had to finally figure out what what's going to be working on all difficulties. We had a 22-fret guitar and a 17-fret guitar. How are we going to author differently for those and what kind of chord names are we going to put on it? So, uh as you some of you may know, developing software for emerging hardware can be complicated. Um this is uh our case as well. On the PC, we were using standard MIDI drivers. Um on the consoles, something like that doesn't exist. It's not that flexible. So, here's our our timeline for production. Um we

had a plan, an ambitious plan, for our hardware team to go off and make these two hardware versions, which are really exciting to us, but it would take a while, and we needed something to develop with so we can answer these questions. Um on the left-hand side, you can see we had seven of these things we call the bridge modules. These are made by our hardware team. They take MIDI and convert it to a signal that the console can understand. We had seven of these for 180 people. It was just enough for our team to develop it develop with, but it wasn't a lot. And they were a little idiosyncratic. Sometimes you'd walk across the floor and static electricity would make you have to reset your console. But, it was just enough for us to get done what we needed. Um a few months before we were done finishing the game, we started getting our Mustangs. You could see we had uh seven of those on the right. But, the Squier came in even hotter. In fact, 40 days before E3, we got our first version of the working Squier guitar. Um we had emergency code in there to cheat success if we needed to use it. Um I'm happy to say we didn't need to use it. So, you guys saw real playing on stage, and people were kind of blown away by that it actually worked. So were we.

Um second problem uh is uh just a simple fact of life when you work on these types of games. Uh our content uh is songs, and our songs are our levels. Um and we get those late sometimes, a lot of them. Uh we had to learn about this problem space before we got our final songs. This is like working on a shooter and not having your levels and your final controller to play them on until the last 3 months. You want to learn things before that. So, our solution was our audio designers would make uh test versions of songs or snippets. They would record them and cut up other songs so that we could learn about as much as we need to learn about before these final songs came in. So, uh for our production team, we had a lot of people that actually had played guitar for a long time. And our two audio guys who were making these uh sample songs actually taught guitar for many years. They would crank out the uh the type of content we needed to answer questions. Uh the first thing that they would give us is like famous guitar songs, you know, songs everyone wants to learn like

Smoke on the Water. Coincidentally like some of those songs that everybody learns, they only learn the easy part. They learn da da da da. They learn that, but they don't learn the arpeggiated verse that we'd have to figure out or the crazy solo that goes on forever. Uh we also got like Jason mentioned songs that are should work well in our system like I Love Rock 'n' Roll which we had in the prototype and we had in the final game. Uh finally, the thing we worked on the most was stuff like um songs that would stress the system that normal guitar players would need for all our DLC. Stuff like arpeggios and alternate tunings, really weird chords, uh tons of that stuff. Um we would have these guys go in and make it, you know. Early on we were thinking, "Well, core language works pretty well as long as we don't get Crazy Train." Then we got Crazy Train. So, on a quick tangent, there's a video I found on YouTube of someone like nailing uh you can turn it up a little more uh nailing the tapping solo uh or nailing something 97%. I'm sure he got better later. Um so You have to kind of say that, you know, often what we find on on YouTube is that people are absolutely insane.

Yeah. Lot better than we thought. It's amazing. Um so having them work ahead allowed us to uh solidify our authoring this time. We figured out what we were going to do for easy, medium, hard, and expert on uh both the instruments of bass and guitar. Um we were always going to keep the same hand position for all difficulties. So, if you're playing on the third fret on easy, you're playing on the third fret on expert. Medium would have a a simple variety of chords that we picked out that we thought people could learn. And hard would we try to make just parsable in real time so you didn't necessarily have to go to our trainers. Uh we also figured out in this period how we're going to author for the 17 versus 22 fret guitar. We also brought in Rick Peckham as a weekly consultant. Um Rick is chair of guitar at Berklee School of Music in Boston. Um he brought us a unique perspective and helped keep us honest during this period. Um among other things, Rick uh came in weekly and helped us uh with something like chord names. On the side of the track we wanted to display uh the chord name for people who can use that. And we don't have a lot of space so we kind of had to come up with our own new uh terminology there. We wanted to make sure it made as

most sense. Um and it's also done in code so our audio guys don't have to uh author that for every single chord. He also helped us with trainer lessons. Uh I mentioned this before. We um didn't have our song tape but we made educated guesses about what people would have to learn on easy, medium, hard, and expert and started authoring those. And this is great for a couple of reasons. One is uh our audio guys spend all of their day analyzing other people's music. And this was their chance to make music that actually uh people would play as lessons and made sense in the in this context. And it's also great cuz people would learn a little bit of theory and they could actually take this to their own guitar lessons later. So last one I'm going to talk about is um our problem of going from prototype to production took us a lot longer than we expected. You know, it was estimated that it would take a couple months to get all the features from the prototype in the game. And uh over 5 months later we didn't even have all that stuff in to parity. So here's a video uh can you turn off the audio, please? Um this is the first version working in

the game in quick play. Um you can see a lot of prototype graphics in the back. Um you you can see that the notes look okay. They're not quite great. There's no chord to exist there. The strings don't have any feedback. There's no position wave. Um this doesn't even match the prototype at this stage. So from prototyping we had this core group that were evaluated by our larger group. Um when it came time to production, uh our strike team membership grew. Um Jason rolled off from coding duties, and we added a project lead, um QA guy, um two audio guys, a producer, and three coders to do the work that Jason was doing. This definitely changed the dynamic of the team. So, then these new team members weren't around to see the prototype grow and learn the lessons for themselves. You know, you a lot of misunderstandings happen when three coders are trying to get previously one coder's work in the game. We also had um more reliance on external resources, even though we had more people. So, if we wanted to get a new menu screen in the game, that might get tasked if it was really big, and you

know, held off for a couple weeks. So, this exposed some process problems with this new big group. So, we need to get back on track. So, one of the first things we did to get back on track was uh Daniel Sussman reiterated to me our core focus, which was non-guitar players and their ability to enjoy and play and learn how to play guitar in this game. There he is over there if you want to kick him in the nuts. Um he was so vehement about this that uh he was willing to even cut hard and expert. And to us, this was a wake-up call because we really wanted this to work on all songs, all the way up, and be able to teach you everything that the the guitar player did. So, this helped get us back on track. Um it helped us prioritize our tasks. Um you know, if we had to do work for the trainer, check, we'll do that right away. But, arpeggio work and stuff like that, that can wait till after we solve the core problems. Um we put all these tasks in the schedule, prioritize them, and then our our guys would get excited because we would see all these stuff we

wanted to get to. We knew we had to blow away the other stuff first, so that helped speed us up. We also switched to more short-term deadlines, you know, we had an alpha date um that was many months away. Instead of just relying on that, we actually set ourselves a new short-term date that uh would get all the features in in a month and a half. And then we pushed up our external play test so we can actually find out this information sooner. The result of these changes was the next four months were the most productive for our period in production. We were now actually back on track and ready to play test. We wanted to know did we nail this stuff? We're lucky enough at Harmonix to have a really great play test department. Um, it could be a subject for a whole talk about how to run a good one of these and we don't have time for that, but we think it's important enough that if you don't have a play testing department, at least find someone that you can to own the process. Uh, it's really part big part of what we do. Our play test questions at this point are are our trainers are as effective as they need to be? Um, what info do we need in our video tutorials if any? Uh, previous Rock Band games had relied a lot on videos. Um, what happens when players don't use

the tutorials and trainers and jump jump right into quick play and just start playing? You know, this is a bad case scenario for us, but we wanted to test it because we knew we'd learn a lot of things. And you know, I suggest doing that for your games, you know, have people skip cut scenes and skip all the trainers and just see what happens. Uh, finally, can players learn how to play guitar through Rock Band 3? So, here we are a few months before. You can turn this up a little bit more. Um, you can see the track here is actually somewhat close to what we shipped. Um, the the numbers on the single note gems are smaller than ideal and we we pumped that up after this. The position may have changed a little, but that was pretty close. And I here you can see the chord book that Jason was talking about actually in action in the real game. We uh beefed it up a little bit so players would have to play chords three times because they weren't getting good feedback without that. But what we found out is with real lessons and real songs when we divided up the songs, uh, players would take a long time in learning these chords. And by the time they actually learned them all and got into the trainers, they had forgotten what they'd done. Um, that was the first major problem. Then

that the other uh issues were were smaller, but consistent. On the right hand side, you see this thing we call the gem highway, which is a preview of all the stuff you have to play. Players would just stare at that. They had no idea what it did and it was just distracting them. Uh additionally, we made a a classic mistake of separating our elements for how you switch lessons in the upper right-hand side, and in the lower middle side, uh how you speed up and down the track. Finally, as you saw there, our success feedback was pretty underwhelming. So, uh we we got a lot of little issues that we needed to solve out of this. Um and our trainer needed to be rethought. Uh teaching the chords beforehand wasn't working as well as we wanted it to. But, people were uh really learning how to play guitar. We Within an hour, we had people that had no guitar experience, and they were like moving power chords around and playing really well. And this was great. Like, all of our uh people in our strike team went to the playtest and got to see this, and we got really pumped up to actually finish off the game really well by seeing how well we were doing. And we knew we just had to nail those last issues with those

core audience people uh before we can get to our wish list items. So, some of that response was first thing we did, maybe a day or two after the the first external playtest, was add a number to the position wave. This tells you what your left hand is doing at any time. Uh I had a sneaky suspicion we should try this for a while, but it wasn't until we saw like newbie players actually wanting more information about where their hand was that we actually added this to the game. Um finally, we eliminated the chord book and actually brought it it directly in to the training. So, players could, at their own volition, decide when they wanted to turn it on and learn the chord right then at any time, as opposed to doing it all front-loaded. Um if you go to see the Dance Central talk, which I highly recommend, later, you'll see that they kind of wound up with similar uh spot in their actual production of their game. Um also, we did a better layout here. You can see on the left-hand side all the elements for speeding up and slowing down and changing uh which lesson you're on and the progress are all in the same spot. So, after solving those issues by

iterating and re-verifying via further external playtest, we were able to get to a ton of polish and wish list items, you know. This is just some of them. Um I'm going to talk about a little bit here. Like this first uh slide shows you some stuff. What What I want to really mention here is when I first came to Harmonix, I was really surprised by how much polish went into the tiny elements of the game. I'm just going to talk about one tiny little example here. So, for a series of repeated chords, um you can see it from the old prototype video, uh we would have names for every single chord, and it would and it would be kind of intimidating and hard to tell when you're the chord changed. So, one of the first things we did is um do a little bright highlight on the first chord in a series of of chords. So, every time the chord changed, you would see a little bit of a pop. And we eliminated the chord name uh next to the chord of anything but the very first chord. Um doing that, of course, meant that after the very first chord, it would scroll off screen. So, what we did was we locked it into place until that series of chords is gone. So, if you do

use the chord name, then it doesn't go away and you and you know like a series of 50 chords is still this chord. At this time, we were also getting able to get in a ton of our wish list items. Here's just a small sample. We finally nailed the arpeggio look and got it working in the game. We got uh guitar techniques such as slides up and down the neck and muted notes, which are just playing without actually fretting. Um and hammer-ons and pull-offs and tons more. One last example was this idea of being able to turn on the chord numbering option in the modifiers. Um we'd wanted to do this for months, and we had it kind of working, but there was actually a rendering bug that didn't let us get to this until the very last day. And with a lot of begging on my part for the producers, we actually got this in. So, our summary here is um when we go from prototype to production, it always takes us a lot longer. We haven't quite nailed it, but for this project, um we did a couple techniques. Uh for one, you know, we use placeholder content um for being able to rapidly develop. Um we

learned a lot from these MP3s, and uh we learned a lot about the space just like we did earlier on with the pictures and the video. Um We revisited our target audience and that helped our team realign on the core goals that we should really be working on before we think of the bigger issues or the smaller issues. Uh and finally short-term deadlines helped us get our work done faster as people wanted to see the results of that work. Um especially stuff like external play tests. So we're really happy with the response from Pro Guitar and Rock Band 3. We see a lot of people enjoying it and learning how to play guitar and going out and taking lessons. I'd like to thank everybody at Harmonix for this feature. There's a lot of people that worked on on this really hard and I want to thank all of you for coming to our talk. And please fill out your forms and we'll take questions now. Thanks. You want to roll the video? Yeah. So while we take questions we're going to run this video. This is like from prototype to the end, you know,

I love rock and roll in this series of uh time-lapse photography here. All right. So hey. Um were you working with the Pro Drums and Pro Keys teams at all? And was there any temptation to try and create a like unified visual style for the three of them? Uh yeah, I worked on the Pro Drums and Pro Pro Keys team. Um the Pro Drums is probably the easiest problem. We we kind of actually nailed that right away. We wanted kind of the the circle symbol gems and we just got a look for that and it kind of just worked right off the bat. Pro Keys is a very different problem and it required very different solutions for a ton of reasons. There was no real desire to try to solve that problem at the same time or in the same way just because we knew that they were they were just

going to come up with different issues. So, I was on both teams but most of the team were separated and we we developed them separately. Yeah, I picked up a real drumming after playing a whole bunch of Rock Band and did my first show in December. So, that's pretty awesome. Cool. Thanks a lot, man. Howdy? Hey, I'm just curious about the decision to ship the game on a separate date from the Squier guitar which just came out this week and how that impacts kind of how you see it impacting your your sales and reach of of this great platform. Uh Could you repeat the question a little louder, please? Sorry, um I was curious to to understand kind of the rationale and decision making behind the fact that the game launched, you know, several months ago and the the Squier guitar controller just launched this week and how that distance in time kind of affects the sales and reach of of this great platform. How it affects the what? How it affects the reach and kind of just the fact that, you know, are people still going to pick it up now that it's an older game versus, you know, something that came out today. hear you at all. Sorry. I'm sorry. Um I think you're you're asking about

how the since the Squier came out later, how did that affect the reach of the game? Yes. Uh I think there's a lot of people actually excited about the Squier guitar. A lot of people I think were surprised by how good the button guitar actually felt. Like I was. I was a big proponent of I didn't even really care so much about that one but then now that I got it at home, it has a lot of advantages. It's really easy to just take apart and throw away and bring to your friends' houses and it's a lot cheaper. So, I think I don't know, for us I think we would ideally like to have the the string guitar sooner but I think it's worked out well. I think people are going to enjoy both of them actually. Thank you. So, uh I haven't played the pro guitar but one thing I noticed that was disconcerting about the pro keys was the lack of auditory feedback about my mistakes. And I see here, you know, there's a lot of visual feedback about what you're doing, but did you have the same issue about audit audible feedback for mistakes? Yeah, we we actually had a lot of discussions about that earlier on and um

we sort of made a a decision to sort of uh veer towards that illusion uh at the the cost of that, right? A lot of uh playing an instrument is refining through hearing yourself. Um now the one thing that's really nice about um well, you can actually do it on both is you can actually, you know, get the output from the the keyboard to go somewhere and and have it amplified and you can do the same thing with the the uh Fender guitar so that you can actually have both, right? Um but our our sort of fear going into was that, you know, the first time somebody actually plays something on an instrument, it sounds really bad, right? And that that is so detrimental to so many people that that's why they never play an instrument. They just instantly assume they suck. So, we wanted to create that illusion to get them over that hump and later on as they start to play on a real instrument and stuff, they'll they'll figure those things out. Yeah, I definitely wouldn't want it as the default, but Yeah, I mean, some of our guys have actually like gone through things to hook it up so they can hear themselves as well and, you know, it's certainly a fun that way, too. Yeah, my

first 6 months of playing guitar, I I sounded really really bad. You didn't want to be there. You didn't want to hear that. Uh hi. Um during the prototype phase, you mentioned that you didn't have any hardware till relatively late. So, during during that prototype phase, what did you kind of use to try and develop um solutions to the problems that you encountered? Um we basically would just buy anything we could that would work in the context of the problem we were trying to solve. So, did you have actual like input into the game with that? Yeah, we had actual input, but we had all kinds of things, you know, I mean, I even plugged a keyboard in just to check the the beat match at one point. So, you know, it was it's one of those things, right? Like you you don't have the final hardware, but you just you find things and you figure out how to figure out these problems ahead of time. Just use whatever we could. Yeah. And then for so for the Squire guitar, does it actually listen to the pitch of the note or or is there something else going on? No, it senses what your hands do. It does not It does You'll notice in those videos that um or if you It's like the

button guitar is, right? If you If you push a fret down, it knows you pushed that fret down before you pluck any string or anything like that. Um you know, there's a lot of attempts at audio analysis and all kinds of interesting things. Um that's is you get a lot of latency from that and you get a lot of a lot of other issues of accuracy and things. So, um we had we couldn't get any of those solutions to really work for us. So, is there any um Does it deal with tuning, I guess, at all in the game? It is tuning agnostic. You could have that thing tuned to the most horrible sounding thing and it would still register correctly. That's why you got to use the string mute if you're really badly out of tune. Are there So, are there songs in the game that use like outside of standard tuning that is addressed ever or no? Did you Did you ask about alternate tuning? No, no uh yeah, yeah, alternate tuning. Yeah, uh I kind of uh breezed through that in my slides, but that's one of the problems we had to solve later on was alternate tuning and we decided to pretend that you're in the same tuning that the guitar player uh played. That's just Okay. Our general philosophy on all these systems was if you're looking down at your hand, you're playing it like you're that guy. So, if you're in drop D

tuning, you know, using one finger for those of you who know what that is, you're going to play it like that guy played it. And it's your decision if you want to tune your guitar to that. It works no matter what. So, does it um tell you that though or or is it just you just like play? Does it what? Does it tell you what tuning you're in? Uh no, it doesn't It doesn't tell you what tuning in the game. You can look that up though. Okay. Thanks. Thanks. So, the decision to um allow the player to play uh um plug it into an amp and play as a real guitar, was that decided early on in the process or was that a feature you implemented later? Are there Is your question um the decision to use a real guitar? Was that early on? Uh yeah, do you know the capability to plug it into a real into an amp and play as a real guitar. that's what we wanted to to bring to the world, right? That's how we started. We wanted people to play on a real guitar. Like they don't have fantasies about playing on a more elaborate plastic guitar, they have fantasies about playing on a real guitar. Um so, you know, having a guitar that could then plug into an amp was kind of the ideal solution for us. Now,

we didn't know that we'd be able to get that to ever happen. Um so, we pursued both avenues and they both ended up with certain advantages. Sounds cool. Hey. Yo. Um when you were working out the notation and how to show these to the player and doing your external play test to see what was more intuitive, at what point did you start seeing people transition from hey, I need to learn this song beforehand and now I can play to start seeing people go oh, I can actually sight read these entire chord charts and songs and things like that. Uh yeah, I think I think for um it really depended on their comfort level and like people that played a little guitar were able to pick it up fairly quickly. If you gave them a you know, a song that was just C, G, and D chords, they wouldn't have any problem uh sort of sight reading that. Um you know, single note runs, things like that, solos, all way harder. Um but so it it sort of depends on the song selection and what their experience was. Now, for a player who never played

anything before, uh it would take a little while, but you know, we would do user testing um sessions that would last about an hour. And in almost every case, you know, you could take take the person who never played a guitar and give them a guitar and get them to play you something. I uh poorly, but you know, they'd do it. I'll say it in those short sessions, someone's not going to go from being able to be super proficient. Yeah. At the company, it took us once we had a pretty good system, uh we we could sight read pretty much all all the good guitar players could sight read on hard within like, you know, a couple weeks. You mentioned it might work the the actual guitar would work well as a good synthesizer. Could you speak up a little bit louder, please? The the the You mentioned that the guitar would work out of context as a as a MIDI guitar of sorts plugged into your regular regular synthesizer. Have you tested this, worked with it? It must have I love the idea of low latency over other synth guitars. I I fully expect that there's going to be a lot of musicians picking up a

Squire because it it has really incredibly low latency and high accuracy MIDI output. Uh just good value. I mean that could be this other thing that takes off with this for musicians. You know, it seems like it could become I can't quite hear you. I'm sorry. It sounds like it could become popular with with guitarists. Hopefully. I mean, you know, that's that's great if it does, right? Um a lot of people have spent years trying to come up with good MIDI solutions. So, if this is another one and it turns out to be popular, that's great. Yeah, I should take the time to plug both the guitars are are MIDI output, the plastic one and the string one. As is our keyboard, so we made actual real instruments this time. I'm going to try it. Like Um one of your earliest slides had a fun score as part of the user research. I was just wondering, is that something you developed internally, like the concept of a fun score and We actually Yeah, what what what's that meant to test for you? Uh we would actually just just ask the people if you had to rate it on the scale of 0 to 5 how fun it was

and then aggregated that onto that chart. Okay. And so, it's just a general sanity check. Yeah, I mean, you know, fun is I mean, is it If you've ever seen any design talk where people start talking about fun, it's kind of hard to quantify it. So, but you know, people when they they find something fun, they have some idea of how fun it is. So, we just ask them and use that sort of information to figure out whether we're on the right path or not and then try to drill down into what's tripping them up, what's causing little issues. I think early on I think we've done maybe 0 to 5 the whole time, so you can compare it against past games fun level at that certain level of the project. So, whatever you whatever standard, I guess just keep it the same so you can compare it in future games, you know. Neat. Thanks. No problem. Yeah, hi. Uh I have been playing this with with quite a lot of people and since you've done a lot of play testing, I would like to hear I'm sorry. Could you Could you pick up a little bit louder, please? Oh, sorry. If you you have done a lot of play testing and I've noticed some difference in people who are

unexperienced or have never played this and the way they are picking things up, whether they have been playing music or not. And also difference between whether they've been playing music using musical notes or just by by like rock music is usually you don't know the notes and then classical musicians are very trained in using musical notes. Do you have any experience from your play testing like around that issue? Yeah, well, we actually internally we keep a database of all our employees, what their musical experience is, um you know, how experienced they are with our games, what difficulties they play on. So that when we're testing these things, we actually can pull that type of information out. Like certainly somebody who plays guitar or even plays music has a little bit of a leg up, right? Um and we did find that there was one guy who went into the to the test, didn't go through any of the tutorials, went straight in and saw the the the chord language and just he got it right away, right? Now other people would be like, I don't know what this is, but some people just got it right away and

it you know, um certainly the the ones who play more music are are going to do that. I I should point out that there were some musicians that it actually had a little bit more of an actual learning curve for the first 5-10 minutes because they're used to a different way of thinking with tab and all that. So, some people that are non-musicians actually had a leg up in our in our final system. It depended on the person and like kind of how they thought, but generally there was about anywhere from a 5 to 40-minute learning curve of just being able to translate the dimensions of moving up, moving left on the screen, and you know, that kind of axis rotation. All right. And did you see any link between knowing to play with musical notes or not in your your this this data of what experience they had before and how long took for them to pick things up? But um between them being able to read traditional music? Yeah. Uh there's not really a correlation there because we don't display anything in traditional notation. I think uh the more important skill um for actually parsing it was being familiar with uh beat match to

some degree. And if they had read tablature, that was also useful. Um you know, a lot of players a lot of what we were trying to eliminate was players stopping and looking at their hands because we want them looking at the screen. So, we're trying to keep feedback and things like that up there so you could know what your hand was doing without having to look at it, Right. and yeah. That's it. 1 hour. We're out. Thank you. Thank you.