AI Chatbot/Avatar in store

James_H

Anybody got any thoughts on this? Looks quite interesting, in theory, but not sure I follow what exactly it can do.

Wonderland

Yes, I’m wondering too. What is this meant to be used for? Creating videos you can post online or is it just some sort of live display that requires a PC?

kwerkx

caveat: Didn't buy, just my opinion

The product video was interesting.. type a question and it responds with a pre-defined response for that question.  <gripe> calling that an AI is a stretch </gripe> and alone, that's not much of a use case.

The system they demonstrate for eye movement, mouth movement, and text to speech is promissing.  IF they follow up with a straight text to speech feature, e.g type it and the avatar says it.. THEN the product would work in the VTubing space (after all the licensing is figured out).  Add face (eye/mouth) tracking to sync user conversation with the avatar's eye/mouth movement and it would be an interesting addition to the interacactive 3D space.

Isn't there a PA product already in the store that models mouth movements to speech?

WendyLuvsCatz

this is not a product that works within DAZ studio 

it is a series of renders of phonemes that are used to verbalise text outside of DAZ studio

you could actually use literally any series of images including your own face making the correct expressions so yes any DAZ figure or render, or image hand drawn, photo whatever 

if you watch the video it clearly explains it

the price is the licensing of the use of voice and ai chatbox presumably

I have no opinion on it, am just explaining what it is

Wonderland

But what is it used for? Is it just putting an AI face to the annoying chatbots many corporate CS sites now have? I get the feeling this is not really intended for artists or making videos. 

James_H

WendyLuvsCatz said:

this is not a product that works within DAZ studio 

it is a series of renders of phonemes that are used to verbalise text outside of DAZ studio

you could actually use literally any series of images including your own face making the correct expressions so yes any DAZ figure or render, or image hand drawn, photo whatever 

if you watch the video it clearly explains it

the price is the licensing of the use of voice and ai chatbox presumably

I have no opinion on it, am just explaining what it is

Thanks for confirming what I thought. I was struggling to see how this offers much for Daz, and thought the video rather jerky as it moves between the 23? images. The lack of body movement made it seem rather limited: certainly not closing in on uncanny valley.

barbult

James_H said:

WendyLuvsCatz said:

this is not a product that works within DAZ studio 

it is a series of renders of phonemes that are used to verbalise text outside of DAZ studio

you could actually use literally any series of images including your own face making the correct expressions so yes any DAZ figure or render, or image hand drawn, photo whatever 

if you watch the video it clearly explains it

the price is the licensing of the use of voice and ai chatbox presumably

I have no opinion on it, am just explaining what it is

Thanks for confirming what I thought. I was struggling to see how this offers much for Daz, and thought the video rather jerky as it moves between the 23? images. The lack of body movement made it seem rather limited: certainly not closing in on uncanny valley.

The very jerky unconvincing video quality, and the fact that it only handles predefined question and answer pairs, makes if seem worthless to me.

WendyLuvsCatz

WendyLuvsCatz

kwerkx said:

caveat: Didn't buy, just my opinion

The product video was interesting.. type a question and it responds with a pre-defined response for that question.  <gripe> calling that an AI is a stretch </gripe> and alone, that's not much of a use case.

The system they demonstrate for eye movement, mouth movement, and text to speech is promissing.  IF they follow up with a straight text to speech feature, e.g type it and the avatar says it.. THEN the product would work in the VTubing space (after all the licensing is figured out).  Add face (eye/mouth) tracking to sync user conversation with the avatar's eye/mouth movement and it would be an interesting addition to the interacactive 3D space.

Isn't there a PA product already in the store that models mouth movements to speech?

https://www.daz3d.com/anilip-2 

or use the DAZ studio 32bit for lipsync to wav files

I use Mimic but that is no longer available

also there is

https://www.daz3d.com/poserecorder

which I also own but is not good with my lopsided face (is OK using videos of others)

Novica

I shake my head everytime I watch that Anilip 2 video- that tongue flaps up into the middle of the mouth and is SO distracting. UGH. But at least they tried.

Artini

This is just a separate application, that will act as digital assistant.

They provide for Daz Studio animation clip, that should be rendered,

and then can be used for visualization of such assistant.

They said, that this application is a fruit of 12 years of research.

Will be interesting to see their store at Daz 3D with additional voices and so.

They also talking about AI, but cannot understand what it does,

because one still need to write responses to what assistant would answer.

The positive thing is, that it uses offline speech recognition,

but there are exist such an open source, as well.

The voice from the demo at Daz 3D page is not so impressive.

I like much more the voices, that are shown on their other demos on YouTube.

 

Artini

Just found this presentation from 2013 about the company, behind this product:

 

Artini

I hope, that it is related to the version selling at Daz 3D:

 

Auroratrek

Novica said:

I shake my head everytime I watch that Anilip 2 video- that tongue flaps up into the middle of the mouth and is SO distracting. UGH. But at least they tried.

I just shake my head, period. What is wrong with Daz that it patently refuses to create a decent lip sync utility? iClone's lip sync utility is FANTASTIC (and bulit-in) while Daz can't even muster acceptable. Pathetic. Why even pretend to be a character animation solution if you can't make the characters talk??