HIGH RES VIDEO:
https://www.artstation.com/artwork/b5QzLd
- VOICE: We have fully developed AI voice modelling
- SOUND: We have fully developed sound modelling (foley studio anyone?)
- MUSIC: We have Riffusion partially developed (a model based on Stable Diffusion that produces high quality music based on spectrograms)
- VISUAL FIDELITY AND CONTROL: We have fully developed ControlNet just last month and proceeding at breakneck speed toward fidelity control
--For those of you not informed yet, ControlNet gives virtual total control over AI images created - poses, facial expressions(to come) and much more. In the hands of current users, scenes can be directed in a very pain free way - by way of reference videos and reference photos of poses in produced existing video clips and films.
Already people are able to create rapid batches of 4k images on existing GPUs and the models are being optimized for even more rapid image generation. Once visuals are so fast and so easy to make, the natural venture into full 3D fidelity will naturally broadcast to the public for pain-free use.
Video is already emerging because image generation is reaching 'diminishing returns' territories, or significant milestones that indicate a 'diminishing return' as far as humans can discern with imagery.
Meaning, in laymens terms:
- CGI artists out of jobs
- Foley artists out of jobs
- Voice actors out of jobs
- Movie directors, tv ad spot directors out of jobs
- Existing Hollywood structures and favoritism of screenwriters, directors, and producers dies a fast, slow motion death
- Virtual complete movie making ability within 3 years that a nine year old could use to make a 2hr feature of their preference
Some may say that Hollywood will adjust, but that simply isn't possible. AI video in its current early versions (automatic1111 webui) is still early versions of AI video.
Next generation open source suites will ravage the industry of elite Hollywood. Your imagination and fears are the only limitation in a time of great techno disruption.