Gemini Omni 1.1 Flash

The clearest landing point was that video generation is starting to look commercially real in narrow workflows, especially advertising and pre-production. Several commenters working with these tools said image and video models are already changing mockups, storyboarding, and rapid concept testing because they collapse days of production work into paid API calls. Google’s angle made sense in that light. Unlike OpenAI’s more social or showcase-style launches, Google can drop video generation directly into ad tools, cloud APIs, and YouTube-adjacent workflows and charge for usage. A second theme was that this is not really about making “art” from a text prompt. People kept coming back to control. Raw generation quality is getting close to table stakes. The harder problem is steering outputs, preserving intent, and getting consistency across iterations. The product promise of cheap 360p drafts was criticized for exactly that reason. If a higher-resolution rerun changes the clip, low-res preview is not a draft in any useful production sense. That pushed the conversation toward richer control surfaces like 3D blockouts, annotated frames, video-to-video guidance, and audio sync rather than pure prompting. Commenters were also blunt about who gets hurt first. Voice actors, audiobook narrators, and other creative workers without a strong personal brand were seen as especially exposed. The discussion around unions was not really about banning AI. It was about consent, compensation, and bargaining power when a company can clone or replace performance. Even people who accepted automation as inevitable still argued that labor needs a stronger seat at the table, because the likely near-term outcome is not leisure but more output pressure with weaker negotiating leverage. At the same time, many readers still viscerally dislike the current output. Human clips were described as lifeless, badly mixed, and stuck in the uncanny valley, even by people who admitted that nonhuman footage like fish or scenery is getting hard to distinguish. That did not translate into confidence that the category will stall. The stronger view was that quality gaps matter less for ads, mockups, children’s entertainment, birthday gags, and other low-stakes uses than for prestige film or carefully directed work. The broader business read was that Google may not need the best coding model or the buzziest chatbot if it can own multimodal infrastructure where money already flows. YouTube, advertising, enterprise distribution, and pay-as-you-go APIs give it reasons to keep investing in video generation even if the startup discourse is obsessed with agentic coding. Several commenters argued that Google is playing a different game from labs chasing monthly benchmark wins. The strategic bet here is not just “make neat videos.” It is to make generated media a billable feature inside products Google already controls.