There has to be *something* we're all good at other than masturbatingaxel000 wrote:My god this board is like a shrine to skepticism sometimes heh
E3 - The MS Press Conference - 18:25 BST
- WiteWulf
- Sheepie Whore

- Posts: 10125
- Joined: Mon Aug 23, 2004 11:20 am 0
- PSN: WiteWulf
- Location: Loughborough(ish)
Re: E3 - The MS Press Conference - 18:25 BST
:::G:::
PSN: WiteWulf
Switch: SW-3941-2878-7584
Steam: WITEWULF66
PSN: WiteWulf
Switch: SW-3941-2878-7584
Steam: WITEWULF66
- teamonkey
- Sheepie Whore

- Posts: 12082
- Joined: Thu Jun 13, 2002 1:07 am 0
- XBL Gamertag: itsthetaste
- PSN: teamonkey_ca
- Contact:
Re: E3 - The MS Press Conference - 18:25 BST
The trouble with facade is that it consists of three or four interesting pieces of tech, but is completely broken by the usual IF obsession with text input.SiN wrote:@Flabio: I'm curious, have you guys have been in contact with the guys who did Facade? There's some overlap in the projects ... I suspect they'd love to work R&D at Lionhead perfecting their tech.
Mo
Re: E3 - The MS Press Conference - 18:25 BST
I don't know if it's really "the usual IF obsession with text input".
Using dialogue trees would defeat the purpose of the tech. But I suspect they also weren't up for writing voice recognition software. So text input is the best way to go.
Of course, if Milo were using the Facade tech, it would take the output from the voice recognition tech and feed it into the Facade engine as text (with "emotion" tags added, or something)
Cheers,
Mo
Using dialogue trees would defeat the purpose of the tech. But I suspect they also weren't up for writing voice recognition software. So text input is the best way to go.
Of course, if Milo were using the Facade tech, it would take the output from the voice recognition tech and feed it into the Facade engine as text (with "emotion" tags added, or something)
Cheers,
Mo
Re: E3 - The MS Press Conference - 18:25 BST
*suddenly feels like playing Starship Titanic again*
-
want to read eye-wateringly bad stories? No? Ok.
want to read eye-wateringly bad stories? No? Ok.
- teamonkey
- Sheepie Whore

- Posts: 12082
- Joined: Thu Jun 13, 2002 1:07 am 0
- XBL Gamertag: itsthetaste
- PSN: teamonkey_ca
- Contact:
Re: E3 - The MS Press Conference - 18:25 BST
Voice recognition would have exactly the same problem as text input, which is that you'd still be trying to hunt out the keywords that trigger a response. Getting a computer to understand the meaning of what you're saying (or writing) is way beyond us at the moment so all you've got is a set of phrase triggers. Natural language input is just broken.SiN wrote:I don't know if it's really "the usual IF obsession with text input".
Using dialogue trees would defeat the purpose of the tech. But I suspect they also weren't up for writing voice recognition software. So text input is the best way to go.
Of course, if Milo were using the Facade tech, it would take the output from the voice recognition tech and feed it into the Facade engine as text (with "emotion" tags added, or something)
Cheers,
Mo
Facade had so many interesting things under the hood that they could easily have used a different input method without losing any meaning (and without resorting to decision trees). For example, choosing a category of response (or emotion) rather than specific words.
Milo responding to emotion might work a bit better. He might not understand "go jump in the fucking lake" but he might detect what you mean. If the voice input is reliable enough.
Re: E3 - The MS Press Conference - 18:25 BST
Ah yeah, agreed. I just assumed you wanted Facade to run on voice-tech as opposed to the keyboard. Silly me.teamonkey wrote:Voice recognition would have exactly the same problem as text input, which is that you'd still be trying to hunt out the keywords that trigger a response.
But yeah, totally. Selecting an emotion category does make sense, but of course, not quite as natural. I think part of the "magic" with Milo is that you can talk to him, right? It's a natural interface. It's a broken natural interface, but the point still holds.
Still, a web/tree-like interface where you'd select an emotion and then drill down as far as you'd like would be interesting. So you'd start with a set of 7-8 basic emotions, and you'd select "Anger" and either leave it at that, or drill further down fine-tuning your feelings or giving them context ("angry because you forgot x", etc).
Molyneux talked about this, actually. You can tell Milo a joke, and he won't "get" the joke, but he'll know that you're telling one from the tone of your voice. Fascinating, really.Milo responding to emotion might work a bit better. He might not understand "go jump in the fucking lake" but he might detect what you mean. If the voice input is reliable enough.
Cheers,
Mo
