Kling, Seedance, they all bug

Hi Exuberant_experience5022,

I pulled your account logs to see what's been happening with those models.

I checked the logs for your account and found the pattern. With Kling, most of your failures on Aug 10 were safety flags on the generated video itself (13 out of 15), plus a couple prompts that hit the 2,500 character limit. Seedance had similar safety filtering, but also flagged 5 of your requests for using reference images or videos with faces in them, that's a hard blocker there.

The safety rejections are happening after the models generate, which means the content itself isn't matching platform requirements. The face detection and prompt length issues are straightforward to work around (shorter prompts, reference images without people), but the generated-video safety flags are harder to predict without trying different prompt angles.
Since these two tools aren't reliable for what you're trying to do, they're not going to be the right fit for your workflow.

If you want to share what you're trying to create, I can point you toward better options. 

Hajar
[Moved to questions]