Video created during Seedance2.0 load endurance testing.

Before creating this video, I tested the prompt's load tolerance with a high-speed battle video, and Seedance 2.0 was able to handle it. Therefore, I considered more demanding visual effects. Specifying 9 references resulted in inconsistent performance; sometimes it could handle it, and sometimes it couldn't. I think this is probably the limit of what Seedance 2.0 can handle. With a few more prompt generation attempts, a more stable result might be possible. HappyHorse seems to excel at realistic rendering, but struggles with animation. HappyHorse seems to be good at high-speed, high-quality videos. Seedance 2.0 is also fast, but I think its greatest strength lies in its inference performance, which allows it to handle extremely demanding prompts. HappyHorse → Realistic backgrounds, effects, etc. Seedance 2.0 → Animation possible, high-load video This kind of differentiation in usage is also possible. With Seedance 2.0, using the Fast or Mini version will lower the cost, making it slightly cheaper, but the generation cost is still higher than HappyHorse. HappyHorse costs 1 USD (approximately 160 yen) per 15-second 720p video. It uses a distillation model, so it seems to have limitations in certain genres. 720p 15-second videos can finally be tried with Dreamina for around 300 yen per video (even cheaper with an annual plan during half-price campaigns), while Lumina costs around 500-700 yen per video. 1080p videos cost around 1000-2000 yen. Generating 100 videos would cost over 30,000 yen, and you still have to rely on the generation lottery, so it's not yet truly unlimited for everyday use.