3D-Mobster posted at 1:41 PM Fri, 24 November 2023 - #4478100

If you look at this AI-generated movie. Based on the description it is an animator with 25 years of experience making his first AI video. Now imagine how it will be in 5-10 years, given that AI video generation is very new. But I think one could imagine how Poser could be used to help control scenes and shots etc.

AI movie

To make this movie it appears the AI is applied as 2D compost and AfterFX. Iin other words, the AI system is not building the scene in 3d in its memory, then rendering it out - it it using pixel morphing at the 2D level - aka PS perspective warp and blending. You could render scenes in Poser today, use as source data for an AI Image/movie generator and get this result.  Today the text and image interfaces necessitates enormous Trial and Error in production to have the AI LERN and produce teh desired output.  The trial and error nature of artist produced AI creations means LOTS and LOTS of attempts, therefore 3d raytraced rendering by an AI scene builder would require cloud based virtualized GPU hardware and be VERY expensive, unless the AI API option for local rendering system was used.  That would tie up your system for ages.... because AI is not as smart as the intelligent human artist at getting it (near) right the first time before hitting render.

>> Why would you need to Mocab anything if Poser could hook up to your webcam and the AI transfer your movement to the character? << This literally IS the definition of Motion Capture (MOCAP), and there are affordable hobby grade camera pose and animation capture systems available today for your 3d pipeline. Intel makes the realsense, and I still have an old Microsoft MOCAP camera some where... It was used by IClone's MOCAP. Reallusion, a competitor to DAZ/Poser actually created a DAZ to to IClone 3d pipeline exchange figure converter and a Camera Mocap system was developed to merge figures with Poses animation. So Blenders AI 2D to Pose approximator script could be applied to a Poser grouped figure in Blender and export BVH file. .BVH is the general file standard to exchange pose/animations to .FBX and Collada figures and Poser imports BVH fine now, even in a parametric grouped rig.

AI creating scenes in your Local Poser instance would be weird again, seems like automation is what you are seeking rather than Artificial Intelligence... are you looking to describe in text what you want in scene rather than grab the content from your library or are you expecting Magician AI to procure the 3D scene content on demand and charge your credit card for the scene you described... I hope accurately or $ wasted...

Recall this thread is a CONTENT DEVELOPER Wishlist for future Poser versions so coming up with a way to eliminate the need for a artist, or even the content store itself would be a bummer to most reading this. Dang we could have AI be the audience for the Gallery and have AI appreciate the AI artists,,, we could save power to our PC's and just eat left over Turkey all day.