Seedance 2.0 Won't Replace Hollywood — But Disruption Has Already Begun

My Twitter feed has been flooded with Seedance 2.0 discussions these past few days, but the takes all feel familiar. The last time everyone was proclaiming the "revolutionary disruption of video production" was back in September when Sora 2 launched. Every time a new model drops, the same predictions get recycled — "the film industry is about to change forever." I've honestly grown tired of hearing it. But is it really overhyped? The thing is, these claims are always future-tense predictions. They're not about what the model can do right now — they're about what people expect once these capabilities keep improving over time.

Impressive, but Far from Industry-Ready

The model is good, no question. But "good" and "production-ready" are worlds apart. If we set the bar for AI video industrialization at 100, Sora was a 1, and Seedance 2.0 is maybe a 3. In relative terms, that's several times better. In absolute terms, it's still nowhere close.

What would industry-ready even look like? One benchmark: coherent video generation with consistent scenes and characters over meaningful durations. You could establish a standard by analyzing real films and TV shows — what's the median length of a continuous scene with the same characters? If the median is 30 seconds, then an AI model that can generate 30 seconds of coherent footage with consistent characters would already cover 50% of typical film production needs.

Disruption Is Gradual — and Already Underway

Too many people frame this as an all-or-nothing revolution. It's not. Right now, Seedance 2.0 probably covers about 5% of everyday video production needs — promotional videos, e-commerce content, and similar use cases. For those specific workflows, productivity has genuinely improved. As the model improves, that coverage expands. The previous generation of video models was already being used for novel narration videos. You don't need a perfect 100-point model for industry transformation to begin — it's a gradual process, and it's already happening.

Pushing Quality Beyond Human Limits

Here's an angle most people miss: AI video can push content quality from the current human-produced standard of 100 to something closer to 200. The bottleneck has always been human labor — you can only shoot so many takes, test so many compositions, iterate on so many concepts. AI removes those constraints. Higher output volume means more experimentation, which means better results. This could drive the entire video industry into a new era of quality.

The Next Million-Fold Leap

Think about it in terms of output volume across eras. In the pre-smartphone era, video was produced by professional cameras — output was inherently limited. The smartphone era tied production to the number of devices in people's hands, multiplying output by hundreds of thousands. The AI era ties it to compute power, potentially multiplying it by another few hundred thousand times. This is explosive growth.

And it points to something even more interesting: personalized video and entertainment. Instead of a centralized team producing content for mass audiences, AI could enable one-to-one content creation — videos and shows tailored specifically to each individual viewer.

Short-Form Drama: The Perfect Testing Ground

Short dramas might be the ideal use case for the current state of AI video. They're naturally short, switch scenes every few seconds, and have relatively low requirements for visual continuity — all of which align well with what models like Seedance 2.0 can deliver today. Design the story, generate the visuals.

I haven't personally tested this workflow for short drama production yet, and I'm genuinely curious about what problems would surface. Someone should really do a daily livestream or production diary, documenting the real experience of using AI video for short drama creation. That would be genuinely fascinating to watch.