[x3d-public] Prompt to robotic motion:

[Joseph D Williams]

➢ Self-Driving robot. 

Smarter, faster drivers?

https://www.youtube.com/watch?v=3N3Bl5AA5QU

3D and nD+1 tensors fast decompositions view directions from prompts and prompts from directions 

➢  I’m not clear on what’s directing the camera.

Either you control the camera or you give control to scene sensor intelligence provided by the author/host/collaborator
Looks like the camera is mostly keeping the robot in view while following the robot as it navigates. Generally it is easy to bind a view relative to any object in a scene, and to bind an object to the view. 

➢ Wondering about peer review at this point.

If the robot collides with you then you are automatically a peer



From: John Carlson
Sent: Monday, October 3, 2022 11:28 AM
To: Joe D Williams; X3D Graphics public mailing list
Subject: Prompt to robotic motion:

https://youtu.be/xYvJV_z5Sxc

Self-Driving robot.  Prompts for directing  the robot.

Wondering about peer review at this point.

Not animation that i know of.

I’m not clear on what’s directing the camera.


John

On Sun, Oct 2, 2022 at 11:49 PM John Carlson <yottzumm at gmail.com> wrote:
Okay!  DreamFusion takes a text prompt to 3D Model, but only allows models to move with respect to the camera.  They do have one multiple object result.

Note that this uses <model-viewer>  so it's apparently fully webbified!

https://dreamfusion3d.github.io/gallery.html

Ask and you will receive?

How about a text prompt to 3D animation, not just moving the camera around or using image interpolation.   Model 3D animation.   Text prompt to animation of a 3D model with respect to itself.  Then add interactivity, like sensors.

I think some of these below allow lights to be changed. I am not looking for light animation right now.

Step one is SHRDLU, step two is model deformation.

John



On Sat, Oct 1, 2022 at 9:05 PM John Carlson <yottzumm at gmail.com> wrote:
Now with prompt engineering, we can do text-to-image and text-to-video.   Now, how about text(description)-to-mesh, text(description)-to-shape, text(description)-to-scene, or even text(description)-to-world?

Can semantics help this?

John

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://web3d.org/pipermail/x3d-public_web3d.org/attachments/20221015/f993f6f8/attachment.html>