Ai Dreams Forum

Creativity => Graphics => Topic started by: 8pla.net on November 11, 2014, 04:31:29 pm

Title: Blender Render
Post by: 8pla.net on November 11, 2014, 04:31:29 pm

Special Place (http://www.youtube.com/watch?v=dueehkEWW7I#ws)

She says, "Hello!  Welcome to my Special Place."

Title: Re: Blender Render
Post by: Freddy on November 13, 2014, 09:31:53 pm

Hehe that's fun  ;D

Title: Re: Blender Render
Post by: 8pla.net on November 14, 2014, 01:17:11 am

Thanks Freddy!

Shape keys, lip-synced to Text To Speech, moving right along on some animation, in a 3D world.

Title: Re: Blender Render
Post by: 8pla.net on November 14, 2014, 03:11:10 am

Special Place. Come Inside.  (Episode 2)

Special Place. Come Inside. (Episode 2) (http://www.youtube.com/watch?v=6W3Z95pyNIw#ws)

Title: Re: Blender Render
Post by: Freddy on November 14, 2014, 10:27:01 pm

There's something about that voice that makes me chuckle  ;D

Title: Re: Blender Render
Post by: Art on November 15, 2014, 01:53:53 pm

Yeah, the voice is kind of a retro-robotic-child sort of thing. (General enough?) Hehe!!

Title: Re: Blender Render
Post by: Data on November 15, 2014, 02:41:38 pm

8pla, you somehow managed to make a cube look and sound cute, good work  :)

Title: Re: Blender Render
Post by: 8pla.net on November 16, 2014, 04:15:06 am

Quote from: Art on November 15, 2014, 01:53:53 pm

Yeah, the voice is kind of a retro-robotic-child sort of thing. (General enough?) Hehe!!

Yes, Art. 

I very much like realistic sounding text to speech.  It's just so cool!

Every now and then, I enjoy making use of the retro text to speech.

It may be an older technology, but it was very well designed TTS.

Title: Re: Blender Render
Post by: 8pla.net on November 16, 2014, 04:27:58 am

Quote from: Data on November 15, 2014, 02:41:38 pm

8pla, you somehow managed to make a cube look and sound cute, good work  :)

Thanks Data,

Rather than high quality work, it's sort of a low polygon hobby.
My goal is to make it  into a simple Blender game.  Here is screenshot of
a simple game I did today of a shuttlecraft.  I am able to fly it around in
the BGE (Blender Game Engine).   

Title: Re: Blender Render
Post by: 8pla.net on November 16, 2014, 04:53:28 am

Quote from: Freddy on November 14, 2014, 10:27:01 pm

There's something about that voice that makes me chuckle  ;D

It's gotten me into a bit of trouble, laughing out loud at it, while others found it annoying.  I tried explaining, that I was not laughing at their reaction to it, but they wouldn't believe me.
 

Title: Re: Blender Render
Post by: 8pla.net on November 16, 2014, 11:08:02 pm

Native American Squaw

Native American Squaw (http://www.youtube.com/watch?v=Ox3ddFXJ-t4#ws)

Rendered in Blender using Linux. Extremely low polygon count.  All original artwork.

Title: Re: Blender Render
Post by: Freddy on March 18, 2015, 09:18:36 pm

I missed this last one the first time around. I like these low poly things you are making and the lighting you used for the Native American.  8)

Which TTS was the cute box talking with please ?

Have I asked you before about MaryTTS ?

Title: Re: Blender Render
Post by: 8pla.net on March 18, 2015, 10:56:19 pm

These low poly counts are modeled in Blender with the web in mind. 
This way, I can program the polygons right into the source code in
an array, in just a few lines of code, instead of a complex 3D format.

That Text To Speech is eSpeak, which completely free.  eSpeak has also
been ported to JavaScript!   I have an example of eSpeak in JavaScript at
http://www.chatbots.tk (http://www.chatbots.tk)

MaryTTS is what great companies like BOT Libre (http://www.botlibre.com) are using.  Me? I'm 
just a hobbyist, so my hobby is eSpeak, as almost like a parody on
commercial virtual assistants. 

That was just a joke. Seriously, eSpeak is a powerful TTS technology. 
From what I gather, natural sounding TTS is just human voice recorded
snippets stitched together. eSpeak is like the computer's real voice.

Title: Re: Blender Render
Post by: Freddy on March 18, 2015, 11:40:03 pm

Quote

These low poly counts are modeled in Blender with the web in mind. 
This way, I can program the polygons right into the source code in
an array, in just a few lines of code, instead of a complex 3D format.

Very clever.

I think I have tried eSpeak, I must have, I've tried most of them.

I admit I don't really know how TTS works entirely. Yes I think it's about sampling real voices. With MaryTTS you can build new voices from recorded content - but it's a steep learning curve on that one. I've only found one resource with information on how to do it. I think it would take a few days to get to the bottom of it.