Seedance 2.5 Turbo: draft fast, then decide what deserves the full run
Seedance 2.5 Reference to Video Turbo caps out at 1080p and starts cheaper. When to draft with it, when to switch to the standard model, and what is the same.
Seedance 2.5 Reference to Video Turbo takes the same kind of input as its bigger sibling and runs lighter. The useful question is not which one is better. It is which clips deserve which run. This is a short decision guide, built around one habit: draft on the turbo model, finish on the standard one.
What is identical
- The inputs: up to 30 reference pictures, up to 10 reference videos and up to 10 reference audio files, with the videos and the audio each capped at 30 seconds in total.
- The prompt style: name each file as @Image 1, @Video 1 or @Audio 1 in upload order and say what it is for.
- Duration from 4 to 30 seconds, native audio on by default, and the same aspect ratios: 16:9, 9:16, 4:3, 3:4, 1:1 and 21:9.
- Text-to-video behaviour when you attach nothing.
What is different
Resolution is the visible difference. Turbo offers 720p and 1080p. The standard model also offers 480p and 4K. The starting cost on the turbo page is lower, and results come back sooner. If a delivery needs 4K, turbo is simply not the tool. If it needs a quick look at an idea, the standard model is more than you need.
A drafting loop that wastes little
- Pick three to five shots you might want. Write each prompt once.
- Run all of them on turbo at 720p and the short end of the duration range, five seconds or so.
- Cut the list. Drop shots where the character drifts or the move fights the references, and fix their prompts or their reference set.
- Rerun the survivors on turbo until each reads clearly. Same inputs, one change at a time.
- Take the one or two keepers to the standard model at the delivery size, with the same inputs and the same wording.
Prompts that suit quick drafts
@Image 1 is the product, a matte black thermos with a copper lid, keep the shape and the lid exactly. A cold alpine morning, the thermos on a flat rock, steam rising as a hand unscrews the lid. Slow push in from the side, soft blue light, shallow focus on the lid.@Image 1 is the host, @Image 2 is the set. The host walks into frame, turns to the camera and waves. Locked wide shot, bright even studio light, warm wood and plants in the background.When turbo is enough on its own
- Social clips that will be watched on a phone, where 1080p is already more than the screen shows.
- Animatics and pitch boards, where the motion matters more than the pixels.
- Variations: ten versions of one shot to find the angle you like.
When it is not enough: anything that will be projected, cropped hard in the edit, or delivered at 4K. There the standard model is the one to finish on. The step-by-step version of that finish is in the standard model walkthrough, and the wider choice of video models is laid out in choosing an AI video model.
Limits that apply to both
Long clips still drift, busy hands are still the weak spot and a prompt that disagrees with its own references loses to the references. A faster model does not change that. It only lets you find out sooner.
Questions
What does the turbo model give up?
The extreme resolutions. It offers 720p and 1080p, where the standard model also offers 480p and 4K. The inputs and the prompt style are the same.
Will a draft look the same on the standard model?
Close, not identical. Use the same references, order and wording, and expect small differences in detail and motion.
Should I always draft on turbo?
For anything you may throw away, yes. For a single shot you already know is right, go straight to the model you will deliver on.
