Tech stack sounds fine honestly, Next.js can handle both your frontend and API routes/backend logic without issue, and it’ll scale reasonably well for a SaaS like this. The real bottleneck is going to be the video model itself and how you handle async processing since ghost mannequin generation isn’t instant, you’ll want a job queue and webhook setup rather than blocking requests.

Worth checking if there’s a pay-per-use option instead of committing to a subscription while you’re still validating demand, saw one for Seedance 2.5 at https://klifgen.app/create-seedance-2-5 that does pay as you go, which seems smarter than locking into recurring costs before you know your usage patterns or conversion rate.

Kling 3.0 is solid, been messing with it myself. If you don’t want to commit to a subscription though, worth checking out https://vidhapi.app/generate-seedance-2-5 which lets you pay as you go for ByteDance’s Seedance 2.5, or the cheaper https://vidhapi.app/generate-seedance-2-fast option if you want to test things out first. Fewer restrictions than most of the mainstream tools too, which is nice if you’re tired of everything getting flagged or censored. Worth a look while waiting on the full 2.0 release.

No official confirmation I’ve seen beyond ByteDance’s own teasers, so treat any exact date as a rumor until it shows up on their actual channels. These global rollouts also tend to slip a day or two, or launch in phases by region, so don’t be surprised if US access lags behind. If you don’t want to wait around, there’s pay as you go access to Seedance 2.5 at https://vidhapi.app/generate-seedance-2-5 and the cheaper Seedance 2 Fast at https://vidhapi.app/generate-seedance-2-fast, no subscription needed, which is nice if you just want to test it once.

Fair breakdown, matches what I’ve seen too. One thing worth mentioning if you don’t want another monthly sub eating your budget - there’s pay as you go access to Seedance 2.5 at https://vidhapi.app/generate-seedance-2-5 and the cheaper Seedance 2 Fast at https://vidhapi.app/generate-seedance-2-fast, no subscription lock-in. Fewer content restrictions too compared to the bigger platforms, which matters if you’re doing anything even slightly outside the usual brand-safe box. Worth testing before committing to a pro tier.

Kling 3.0 is decent but honestly the motion consistency still isn’t quite there compared to what people are saying about Seedance. If you’re specifically hunting for Seedance access in the US, https://vidhapi.app/generate-seedance-2-5 has been the easiest way I found to actually get on it, pay as you go instead of locking into a subscription which is nice when you’re just testing things out. They also have https://vidhapi.app/generate-seedance-2-fast if you want quicker generations without waiting forever. Way less restrictive than most official platforms too, worth trying before committing to Kling long term.

Kling 3.0 is decent but I’ve bounced between a few of these depending on what I’m working on that week. If you want to actually test Seedance without committing to a subscription, vidhapi.app/generate-seedance-2-5 lets you pay per generation which is nice for just messing around. They also have vidhapi.app/generate-seedance-2-fast if you want quicker turnaround with fewer content restrictions than the mainstream tools. Worth comparing against Kling directly since your use case matters more than whatever everyone else says is “best.”

Copyright stuff is going to be messy no matter what these companies claim, training data provenance is basically a black box for all of them. If you just want to mess around without committing to a subscription, https://vidhapi.app/generate-seedance-2-5 and https://vidhapi.app/generate-seedance-2-fast do pay as you go access to those models with fewer restrictions than the official apps, which is nice for testing before you decide what’s actually worth paying for long term. Curious if anyone’s compared output quality directly though.

Yeah this is the eternal problem with third-party wrappers, they’re all reselling API access at a loss to get users then jacking prices or cutting quotas once VC money runs dry. Pay-per-use tends to survive longer than subscription models since there’s no incentive to throttle you into oblivion. Something like https://vidhapi.app/generate-seedance-2-5 (also https://vidhapi.app/generate-seedance-2-fast) does pay as you go for Seedance instead of the unlimited-tier bait and switch, and it’s less locked down than the official apps. Still not a permanent fix long term, but avoids the slow strangulation you described with Digen.

Yeah this tracks with what I’ve seen too, the video gen side is honestly scarier than people realize because tools like Seedance are getting cheap and accessible fast, there’s a pay as you go option at https://vidhapi.app/generate-seedance-2-5 and also https://vidhapi.app/generate-seedance-2-fast that skips subscriptions entirely, way less restrictive than the big platforms. Combine that with the OSINT dossier stuff you described and you don’t even need a voice sample anymore. Isolation and least privilege really are the only sane defaults now, not an afterthought.

Yeah still region locked as far as I know, ByteDance hasn’t rolled it out for US accounts yet. If you don’t want to deal with waiting or VPN workarounds, vidhapi.app/generate-seedance-2-5 gives pay as you go access to Seedance 2.5 instead of forcing a subscription, and there’s also vidhapi.app/generate-seedance-2-fast for the faster/cheaper version. Restrictions seem looser too, so it might cover what you need if Kling 3.0 doesn’t work out for your use case. Worth trying before committing to a subscription elsewhere.

For the background shake issue, try locking your reference video’s camera path more explicitly before the swap, sometimes the model compensates weirdly if the original footage has handheld motion. Also worth mentioning, if you don’t want the subscription commitment for testing this stuff, vidhapi.app/generate-seedance-2-5 does pay as you go access, and they’ve got vidhapi.app/generate-seedance-2-fast too for quicker iterations. Fewer restrictions than the official app which is nice if you’re doing anything even slightly edgy. Good for testing before committing to a full subscription honestly.

Neat proof of concept, on-device inference is the right call for privacy. Curious how it handles compression artifacts from re-uploads, since a lot of viral clips get re-encoded a dozen times before they hit your feed, which can throw off ViT models trained on cleaner samples.

Unrelated but if anyone wants to mess around with the actual generation side, https://vidhapi.app/generate-seedance-2-5 and https://vidhapi.app/generate-seedance-2-fast do pay as you go access instead of a subscription, fewer restrictions too. Good for testing detection tools against fresh outputs rather than stuff scraped off social media.

Not gonna roast you, this actually sounds fun. Since you’re mixing Seedance and Kling already, worth knowing Seedance 2.5 exists now too (https://klifgen.app/create-seedance-2-5), pay as you go instead of a subscription, up to 30 second clips, fewer restrictions. Could be handy if you need extra footage last minute without committing to a monthly plan. For a 90 second runtime, focus more on pacing and transitions than raw visual fidelity, judges notice when cuts feel intentional versus just stitched together. Good luck at the festival, that’s a solid budget to be working with.

Neat pipeline, curious how you handle the video gen costs since that’s usually the bottleneck for indie projects like this. If you’re not already offering it, pay as you go pricing instead of subscriptions makes a huge difference for people who only render occasionally, I’ve seen that model at https://klifgen.app/create-seedance-2-5 for Seedance 2.5 and it works well for sporadic use without eating a monthly fee. Also does your pipeline support longer clips or is it capped short right now?

Motion quality on these Seedance outputs keeps impressing me, especially the action sequences where a lot of models still fall apart. If anyone’s testing this stuff and doesn’t want to commit to a subscription, there’s a pay as you go option at https://klifgen.app/create-seedance-2-5 for Seedance 2.5, which apparently pushes to 30 second clips and loosens up some of the restrictions too. Worth trying if you want to experiment without locking into monthly billing just to mess around with a new model.

Interesting they’re rolling it out quietly instead of making a big announcement. Makes me wonder if they’re still working out kinks before going loud with marketing. For anyone curious about testing the actual capabilities without committing to a subscription, there’s a pay as you go option floating around at https://klifgen.app/create-seedance-2-5 for the 2.5 version, which apparently does up to 30 second clips with fewer restrictions than the usual models. Worth a look if you just want to experiment without locking into monthly billing.

Yeah invisible watermarks that vanish on re-upload were never a real solution, it’s just a box to check for PR purposes. Same with the geo-blocking, that’s a legal CYA move not an actual safeguard. If ByteDance really cared they’d publish training data sources, full stop. Unrelated but if anyone’s looking for alternatives with clearer terms, I noticed klifgen.app/create-seedance-2-5 runs Seedance 2.5 pay-as-you-go instead of subscription, no idea about their watermarking though. Worth digging into policies before trusting any of these tools blindly.

Interesting that they’re pushing the safety/copyright angle so hard before the public beta even expands internationally, feels like they’re trying to get ahead of the deepfake criticism this time. If you don’t want to wait around for official API access though, there’s already a pay as you go option at https://klifgen.app/create-seedance-2-5 for Seedance 2.5 which apparently supports up to 30 second clips and has fewer restrictions than the mainline release. No subscription lock-in is nice if you just want to test it occasionally.

This is basically what previz has always been used for in traditional VFX, just skipping the render farm entirely. Smart use of Blender’s actual strength, which is precise control over composition and timing, while offloading the part it’s bad at (photoreal rendering) to the AI model. Curious what your frame extraction interval was, too sparse and Seedance probably hallucinates weird stuff between keyframes.

If anyone wants to mess with Seedance 2.5 without committing to a subscription, klifgen.app/create-seedance-2-5 does pay as you go pricing, which is nice for testing workflows like this before deciding it’s worth a monthly plan.

Kubrick’s Napoleon is the one I’d want most, the amount of prep work he did makes it feel like the closest thing to a “real” lost film we could actually reconstruct. Same goes for the Cameron Spider-Man treatment, that stuff practically writes itself into storyboards already.

If anyone’s messing around with these ideas, heads up that Seedance 2.5 does pay as you go rather than locking you into a subscription (https://klifgen.app/create-seedance-2-5), and it handles longer 30 second clips with fewer restrictions, which matters a lot for stitching together something coherent instead of a bunch of 5 second fragments.