After agonizing for over a dozen hours, I went ahead and bought DaVinci Resolve… Hope I can actually use it properly.
I was planning to buy from the official store to prioritize the refund policy, but the list price is ¥49,800 — with PayPal's ~4% spread that comes to ¥52,000 (granted, that's because I was paying in USD) — versus ¥45,600 on Amazon. 45,600 / 52,000 = Δ13%, or 45,600 / 49,800 = Δ8.5%. The Amazon version also includes a physical package, so I could assign resale value to it even secondhand. Given that trade-off, I figured it was better to go with Amazon and give up the refund option.
For the record, this is for MV/music video production.
I'd also considered Filmora's AI auto-montage, but that's cloud-based with credit consumption, and — the biggest dealbreaker — it's mainland Chinese software. CapCut is out for the same reason. I researched basically every other AI subscription tool too, but none of them are specifically suited for full MV production (like OpenArt, etc.). I even tried Google Vids, but it's too bare-bones.
You can't really know until you try, but apparently most operations can be scripted via Python/API, and the 2-week refund window was a bit short for that kind of evaluation anyway.
I'd spent about three days consulting Perplexity and Grok before landing on the conclusion that full automation isn't possible, but DaVinci is still the right call.
Worth noting: unlike FL Studio, Blackmagic Design's DaVinci apparently does not contractually guarantee perpetual version upgrades. So from a game-theory perspective, the right move is to hedge against the convex uncertainty of future first-mover opportunity costs — hence buying now.