Replace Sora SKU shots with Seedance
SKU identity that has to hold maps to Seedance 2.0 at 69 credits, 4–15s, 1080p native audio, or Fast R2V at 25 credits, 720p.
SKU shots that used to run on Sora — the pack has to stay the pack, multi-ref if you have it, lettering you can still read — map to Seedance 2.0: 69 credits, 4–15 seconds, 1080p, native audio, text-to-video and reference-to-video. Draft the same identity on Seedance 2.0 Fast Reference to Video at 25 credits, 720p, 4–15s, native audio. The SKU must hold. That is the job.
Sora 2 Text to Video is 10 credits, 720p, native audio, 4 / 8 / 12 / 16 / 20s. Sora 2 Pro Storyboard is 30 credits, I2V, 10 / 15 / 25s, native audio. Both sit on the API that OpenAI’s Help Centre dates to 24 September 2026. A Sora prompt that says “our bottle” without a still is how the label becomes a cousin. Seedance 2.0 is the Versely row whose categories include reference-to-video.
This is not a Sora versus Seedance taste test. It is the identity route.
Why SKU work is not a Sora T2V prompt
A SKU shot has a legal object: this label, this cap, this carton die-line. Text-to-video will invent a near-pack. Near-packs fail review.
Sora 2 T2V never required an image. Pro storyboard did (image-to-video). If your Sora SKU workflow was already still-led, you are closer to Seedance R2V than you think. If it was prompt-led, the sunset is the moment to attach the JPEG instead of mourning the model.
Seedance 2.0’s catalog card: audio true, categories text-to-video and reference-to-video, durations every second from 4s through 15s, 1080p, 69 credits. The description is cinematic video from text with native audio-visual sync. The R2V category is why it belongs on SKU work. Fast R2V is the lower-cost ref row: 25 credits, 720p, same 4–15s, native audio, discounted on the card.
Rules that keep the pack:
- The still is the brief. Front of pack, back if the turn needs it, in-hand if the hand is part of the ad. Prompt around the still. Do not re-describe the artwork.
- 15 seconds is the 2.0 ceiling. A 16s or 20s Sora orbit does not fit. Cut the move, or take the long hold to Seedance 2.5 at 47 credits, 4–30s, 720p, native audio — knowing 2.5’s published category is text-to-video, 720p, not 1080p R2V.
- 1080p vs 720p is a pick. Full 2.0 is 1080p at 69 credits. Fast R2V is 720p at 25. Sora T2V was 720p. If you were already shipping 720p SKU, Fast can be the production row, not only a draft. If the insertion order says 1080p, Fast is the look-dev row.
Do not replace a SKU Sora take with Veo 3.1 T2V just because both have native audio. Veo T2V invents the frame. Veo R2V (64 credits, 8s) is the Google identity door, and it is one 8-second tick. Seedance 2.0 holds 4–15s at 1080p.
Full 2.0 versus Fast R2V
Use Fast R2V (25 credits, 720p) when:
- You are checking whether the pack holds under a camera move before you spend 69.
- The delivery is a 720p social cut anyway.
- You have references and you need 4–15s with native audio without the 1080p sticker.
Use Seedance 2.0 (69 credits, 1080p) when:
- The stills already worked on Fast and you need the 1080p plate.
- Review will pixel-peep lettering.
- You might generate from text or from refs on the same family without changing vendors.
Do not run both on every SKU “to see.” Fast exists so you do not spend 69 credits learning that the cap was the wrong still. Once Fast holds, either ship 720p or pay 69 once.
Native audio is on for both. Write the room and whether anyone speaks. A SKU orbit often wants no speech, only foley. Say so. A native-audio row will invent a voice if you leave a hole.
If you only have one still and a first-to-last move, Flux 3 first-last is 9 credits, 5–20s, native audio. That can beat Seedance on length (20s vs 15s) when two frames define the travel. It is not multi-ref. If you have a stack of pack angles, stay on Seedance R2V.
Wan 2.7 Reference to Video is 18 credits, 2–10s, 1080p, requires refs. Audio is null on that catalog row. Use it when you accept silence or a later track. Do not use it as a native-audio SKU twin of Seedance.
Moving a Sora SKU library across
Export Sora stills and keepers first. The export path is OpenAI’s Help Centre URL, not a screenshot folder you hope is complete.
Sort SKU files into: hero stills, approved motion, rejected cousins. Heroes go on Fast R2V. Approved motion can still ship as-is; you do not have to regenerate a pack that already cleared. Rejected cousins are not references. Feeding a bad Sora take as a “style ref” teaches the next model the wrong bottle.
Rewrite the prompt as camera and verb only: slow table turn, three-quarter, stop on the barcode, no new label text, no extra flavor banner. If Sora was inventing condensation and you need condensation, photograph it or paint it on the still. Do not ask the video row to redesign the SKU.
Duration map: Sora 4s and 8s fit Seedance 2.0 easily. Sora 12s fits. Sora 16s and 20s do not. Cut the orbit or use Seedance 2.5 / Flux 3 for the extra seconds, then check the pack again. Length that wrecks lettering is not a win.
Caption after identity holds. A muted UGC cut still needs burned-in captions if there is a line. The UGC video generator does not replace R2V.
If New York consumers will see a photoreal pack-plus-person who is not a real identifiable performer, that is a different statute than SKU lettering. This page is the pack. Do not mix the legal objects.
When not to use Seedance for a Sora leftover
The leftover is a talking presenter and the pack is a blur. That is the Veo 3.1 dialogue job, 40 credits, 4 / 6 / 8s, 4K.
The leftover is a cheap composition test. Kling 3 Turbo, 12 credits, silent. Do not spend 69 credits on a layout you will throw away.
You need 25 seconds from Pro storyboard at 1080p. Seedance 2.0 cannot. Seedance 2.5 can at 720p. Movie-mode can chain. See Sora 2 Pro storyboard after the API.
You have no still and no pack photo. Seedance T2V will invent. Go shoot or render the pack first. The sunset is not an excuse to prompt a label from memory.
Run the identity row from Versely.
FAQ
Is Seedance 2.0 Fast the same model at a lower bill?
It is the Fast reference-to-video card: 25 credits, 720p, 4–15s, native audio. Full Seedance 2.0 is 69 credits, 1080p, T2V+R2V, 4–15s. Fast is the draft/ref row. It is not a 1080p discount on the same file.
Can I keep 20-second Sora product orbits on Seedance 2.0?
No. Seedance 2.0 stops at 15s. Cut the move, use Seedance 2.5 (4–30s, 720p, 47 credits), or use Flux 3 T2V / first-last up to 20s and re-check the SKU.
Why not Veo 3.1 for product?
Veo 3.1 T2V is a short 4K native-audio generate that invents the frame. Veo R2V is 8s at 64 credits. SKU work that needs 4–15s and a reference stack belongs on Seedance 2.0 or Fast R2V unless the brief is specifically Veo’s 8-second ref clip.