I know I'm likely shouting into the void here but I'd love to talk about prompting best practices for SeeDance 2.5 because I've been learning a lot. It's a great idea, even with other models, to bone up on your camera terminology. For framing try Close-Ups for more emotional scenes, Medium Wide for establishing shots, and when you have more than one character present in a frame, try stating it as a two-shot/three-shot to help with consistency. It works the same for placement! Profile is good when you want your character facing the side, straight-on is better for again, simple subject-in-frame shots, and three-quarters is actually the sleeper; it gives shots the most dimension and helps fight the flat/ai 'look' I've seen plenty of in standard text-to-video generation. Angles are also important. Low angle helps your subjects feel larger than life or creates a moody image. High Angle makes your subjects appear small or, again, helps establish space. And more to the high-angle, consider MOTION! Handheld gives you some nice shake and jitter, tracking shot means that the camera is moving WITH the subject (and you can even give it as specific item/body part to track with. Faces are good when you want a lot of motion but want to make sure your subject is in frame. Whip-Pans create that tense 'what is going on' feel with fast dramatic camera swings. And finally, the PRO MOVES. Depth of Field. There are lots of different types of depth of field. Soft is more subtle, bokeh creates those nice warm coronas around light sources, and describing depth of field as 'creamy' creates interesting results. Focus RACKING is also very fun and thanks to Seedance's longer video generation run-times, you can really work depth of field. Try describing a focus rack from subject A to Subject B when you're doing over-the shoulder conversational shots. Just some fun little tips for anyone reading this. Build your prompts as much around emotion as technical terms and you'd be surprised what you can make!