Raw LLM Responses

I have been paying attention to the generative AI video space ever since I saw the Facebook and Google debut their first AI models in late 2022. First I wish to disabuse you of the notion that all Gen AI is just typing a prompt. I knew this would be a paradigm shifting technology, as was the introduction of Digital Video and Digital Animation. You would get a lot of pushback from purists, you’ll see wide adoption, and then it will be treated as the norm. I have since viewed a lot of work by genuinely creative people, who wouldn’t have been able to make it without it. Creators like Aze Alter, Neural Viz, Gossip Goblin, Glumlot, Above the Vortex, Voidstomper, Doopiidoo, etc. These creators are doing more than just typing a prompt (text2image, text2video). Sometimes they are taking their own illustrations and having the AI use that as a model for further generations (image2image, image2video). They are taking video of actors and having the AI use their performance as the basis for completely different characters and settings (video2video). And often these creators do all the voice work themselves and have the AI morph it into different characters (audio2audio). To the ignorant it appears as if the computer is doing all the work, when in reality the artist has more control over the outcome than they ever had before. There are people who say that it is unethical because it steals from existing artists. A number of courts have ruled that AI training is legal fair use, because the AI is not storing, creating, or distributing copies of the original work. Like human artists it is simply learning what visual patterns are associated with different concepts. They say it makes us beholden to large corporations. Which is an incredibly ironic criticism for artists to make because if they don’t want to be beholden to large corporations they’re in the wrong business. Aside from that it is not true, at least most of the time. Many Gen AI models are open source and can run on a mid tier consumer laptop. You can download ComfyUI right now and within minutes using image and video models like Stable Diffusion, Z Image, Wan, LTX, etc. Also the John Whick clip you showed at the end was not “made by a human” it was the work of a person mashing together two previously existing copyrighted properties in a crude fashion, which most likely included AI assisted masking programs to cleanly separate the live action and video game footage into composite layers, and editing them together in a way that is mildly amusing.