# Seedance 2.5 frequently asked questions

## Where can I try Seedance 2.5 online?

Visit seedance-ai.dev to create Seedance 2.5 Turbo videos from text or a starting image in your browser. Select a mode, describe your scene, choose 4–30 seconds, and review the credit estimate. No software installation is required.

Sign in with Google to generate using your available credits. My videos shows task progress and completed downloads. Preparing an image or typing a prompt does not spend credits; submitting a generation does.

[Explore the studio](/#studio)

## What is Seedance 2.5?

Seedance 2.5 is an audio-video generation model developed by ByteDance. Its published capabilities include clips up to 30 seconds in a single generation, reference-based creation, and video editing. These are model-level capabilities; the controls available in a particular application can differ.

Our website runs the Turbo model for text-to-video and image-to-video. Both modes support 720p and 1080p, 4–30 seconds, and optional generated audio. Editing, video extensions, and multiple-reference workflows are not available in this studio.

[ByteDance’s model overview](https://seed.bytedance.com/en/seedance2_5)

## Is video generation available on this website now?

Yes. Turbo text-to-video and image-to-video generation are available to signed-in users with enough credits. Choose a mode in the studio, write your prompt, review the total, and select Generate video.

Google sign-in, credit balances, and task history are available. Buy a one-time credit pack to fund your account before generating. Selecting an image or typing a prompt alone does not start a generation.

[Check studio availability](/#studio)

## Can I generate AI videos for free or without signing up?

You can explore this website, read the prompt examples, and preview a starting image without an account or payment. This free preview does not include a generated video. We are not currently offering anonymous generation.

New accounts start with zero credits, and registration does not include free credits or a free generated video. The minimum generation is four seconds at 720p and costs 80 credits. You need a sufficient credit balance to generate; credit packs are available as one-time purchases.

[Pricing](/#pricing)

## How much does AI video generation cost?

Generation costs 20 credits per second at 720p and 22 credits per second at 1080p. The dollar cost per credit depends on the credit pack you buy. Both text-to-video and image-to-video use the same rates, with or without generated audio. Durations are whole seconds from 4 to 30.

A five-second video costs 100 credits at 720p or 110 credits at 1080p. An eight-second video costs 160 or 176 credits respectively. The studio confirms your total before submission. Confirmed failed generations are refunded; submitting another successful take uses additional credits. Credit packs are available as one-time purchases.

[Read the pricing guide](/pricing.md)

## What images can I use for image-to-video generation?

The studio accepts one JPG, PNG, or WebP image up to 20 MB. Choose a local file, drag it into the image area, or paste a direct public HTTPS image URL and select Use image. The URL should point to the image itself rather than a webpage containing it.

Check that the intended image appears before writing your motion prompt. A remote host may block image loading; if that happens, upload a permitted local copy instead. Only use content you have the right to share, and avoid confidential or sensitive images.

[Choose a starting image](/#studio)

## Can I generate a video from text alone?

Yes. Select Text to video in the studio and describe the subject, setting, action, camera movement, and sound. No image is required. Choose an aspect ratio in Advanced Settings: 16:9, 9:16, 1:1, 3:4, 4:3, or 21:9.

For Image to video, provide one JPG, PNG, or WebP and a motion prompt. The image guides the framing, so a separate aspect-ratio control is not shown. Both modes use Turbo and share the same resolution, duration, audio options, and per-second prices.

## How do I write a good AI video prompt?

Write a shot description rather than a list of visual adjectives. Identify the subject and setting, give the subject one main action, and explain the camera movement. State what changes during the scene and how the shot should end. For an image-led scene, focus on movement instead of repeatedly describing everything already visible.

In your motion prompt, add specific sound direction when it matters: footsteps on stone, leaves moving in the wind, or a short spoken line. Say which details should remain consistent, such as a face, costume, or product shape. Avoid asking for unrelated locations, rapid transformations, and several competing camera moves in the same short shot.

[Read the prompt examples](/#examples)

## Can I generate videos with sound?

Yes. Enable Generate audio in Advanced Settings to create audio with your video. Audio is on by default and does not change the listed credit rate. Turn it off when you want a silent clip.

Describe ambience, dialogue, or music as deliberately as the visuals. Review the finished sound for synchronization, clarity, and unwanted details. A prompt is creative direction, not a guarantee of perfect speech or timing.

[Model capabilities](https://seed.bytedance.com/en/seedance2_5)

## What does Seedance 2.5 change from Seedance 2.0?

ByteDance describes this release as extending single-generation storytelling to 30 seconds, broadening multimodal references, and adding finer control over audio-video editing. Its release materials also discuss improvements to continuity and reference-guided creation.

These are the developer’s published capabilities, not a benchmark conducted by this website. This studio supports Turbo text-to-video and image-to-video at 720p or 1080p. Choose a model and interface based on the scene you need to make, the controls actually available, and the result you can obtain.

[ByteDance’s release announcement](https://seed.bytedance.com/en/blog/one-take-creation-flexible-referencing-introducing-seedance-2-5)

## How does Seedance 2.5 compare with other AI video models?

This model is one option; there is no single best model for every shot. Compare the exact versions and tools you can access using the same starting image and scene direction. Look for consistent faces and objects, believable movement, correct camera behavior, usable sound, and a result that follows your instructions.

We have not run a controlled comparison that establishes an overall winner. A dramatic showcase is useful inspiration, but it does not predict how every prompt will perform. Check current availability, total generation cost, and usage terms before choosing a service for production work.

## Can I create scenes for short films or product campaigns?

You can plan individual scenes for a short film, a product reveal, or a social campaign. Start with a shot list and a consistent reference style. Keep each scene focused, then assess continuity, pacing, and audio across the edit rather than expecting one prompt to replace an entire production workflow.

For commercial work, confirm that you have rights to the images, logos, likenesses, music, and other material involved. Review our service terms before generating. AI output can include visual errors or unintended details, so inspect each clip before publication.

## Where are my generated videos saved, and for how long?

Sign in with Google to open the My videos dashboard. Each submitted task appears there with its progress, credit cost, and completed output. The page refreshes task status automatically while generation is running.

Generated video files are kept for only 5 days after completion. Download each finished clip as soon as it is ready to keep a copy. Task history is not permanent media storage. Historical media from the retired service may be unavailable.

[Open My videos](/dashboard/)

## Are the homepage examples generated with Seedance 2.5?

The homepage shows actual Turbo image-to-video results: two eight-second 720p clips with generated audio. Each example links to its AI-generated starting image and includes the motion and sound prompt in English translation.

The workflow panels are illustrative guides, while the playable gallery clips are real model outputs. Output quality varies, and a selected example should not be treated as a guarantee for every image or instruction.

[Watch the video examples](/#examples)

## Is seedance-ai.dev the official Seedance website?

No. seedance-ai.dev is an independent website and is not affiliated with or endorsed by ByteDance. The model name identifies the technology this studio uses; the name does not imply that we developed it.

Use this website for our studio availability, pricing, and service policies. Consult ByteDance’s own model information for the developer’s technical announcements. Features announced for the model may not all be supported by this service.

[About this service](/about.md)