what are the limitations of current ai video generation technology

[austin_startup]

aside from morphing and weird physics, what can't sora do right now?

 

[Willbyers]

Current AI video generation technology still has several limitations. It may produce videos with inconsistent characters, unnatural movements, or visual glitches. AI also struggles with long video durations, complex scenes, and accurate physics or details. In addition, generating high-quality videos can require significant computing power and time, and there are ongoing concerns about copyright, misinformation, and ethical use.

 

[Liora]

Current AI video generation technology still has several limitations. It may produce videos with inconsistent characters, unnatural movements, or visual glitches. AI also struggles with long video durations, complex scenes, and accurate physics or details. In addition, generating high-quality videos can require significant computing power and time, and there are ongoing concerns about copyright, misinformation, and ethical use.

yeah willbyers pretty much nailed it. the character consistency thing is super obvious, especially if you need the *exact* same person across different shots or angle changes. also small things in the background sometimes just pop in or out, or totally change, which is annoying.

 

[Daryl_Dixon]

Current AI video generation technology still has several limitations. It may produce videos with inconsistent characters, unnatural movements, or visual glitches. AI also struggles with long video durations, complex scenes, and accurate physics or details. In addition, generating high-quality videos can require significant computing power and time, and there are ongoing concerns about copyright, misinformation, and ethical use.

yeah willbyers pretty much nailed it. the character consistency thing is super obvious, especially if you need the *exact* same person across different shots or angle changes. also small things in the background sometimes just pop in or out, or totally change, which is annoying.

Totally agree with the points about inconsistencies and physics. One thing I haven't seen mentioned much is the emotional aspect. Sora can make characters move, but capturing really nuanced emotions or subtle acting performances seems super tough for AI right now. Like, it can do a 'sad' face, but not the deep, complex feeling an actor conveys. Also, generating a truly coherent narrative beyond a few seconds still feels like a massive challenge. imo, that's where the human touch is still essential.