@AikiBoy My summary for Kling 3.0 after all the testing I have done.
Kling 3.0 has multi-shot which is the main highlight, similar to what Sora 2 has right now, and it also comes with a storyboard shots UI.
It does work with 3x3 storyboard grids too.
Video-3.0 now can use Elements to maintain consistencies for characters/environment, etc.
And Elements can include a voice-over for characters.
Video-3.0-Omni has better success in output stability, than Video-3.0.
So use more reference-to-video, as opposed to image-to-video with start-end frames.
2.6 model cannot use the end-frame with native audio, but you can use the end-frame for 3.0 with native audio enabled.
Also 2.6 models only have 5 or 10 seconds output options.
3.0 models can have a variable choice between 3 to 15 seconds.
Not forgetting Image-3.0 and Image-3.0-Omni.
Certain image topics have upgraded visual qualities.
It is cheaper to generate without native audio.
720p at 3 seconds is 18 credits, while for 1080p is 24 credits.
Motion quality also got upgraded, to the point that certain motions only need 3 seconds, instead of 5 seconds previously.
So it is better to think in terms of 3 seconds minimum, not 5 seconds.
For multi-shot, it can be as low as 1 second each, but for 3 shots.
Native audio now supports Cantonese, Korean, Japanese and Spanish languages, with accents and voice tone.
Not 100% perfect at launch, but it will get better over time.
It has less hype than Seedance 2.0, that is for sure.