# Alors on Danse AI Video model and output

Updated: 2026-10-09. Website: https://alorsondanseaivideo.net/

## Generation model

The website uses ByteDance Seedance 2.5 through KIE, with model identifier `bytedance/seedance-2-5`. There is one fixed production preset, not a menu of interchangeable models.

Two uploaded photos guide the appearance of the two subjects. A site-owned reference video guides the white-studio composition, keyboard-player and expressive-partner positions, movement and camera timing. The first photo takes the left keyboard-player role; the second takes the right partner role. People and pets are supported inputs. Pets are asked to retain species features while performing in an upright pose.

Standard output: 16-second, 720p, 16:9 landscape MP4. Users choose the two photos; the scene, duration, format and audio are fixed. There are no custom prompts, alternate scenes, vertical-video controls or music-upload options in this product.

## Audio and content checking

Model-generated audio is disabled (`generate_audio=false`). The supplied reference track is merged into the generated video through FAL's audio/video composition service. This preserves the chosen original track rather than creating a new song or relying on generated sound.

The KIE request enables `nsfw_checker=true`. This setting is part of the generation workflow; it is not a claim that every possible unsafe or inaccurate result can be detected. Use photos you have permission to upload and get the consent of depicted people.

## Actual samples and limits

The published golden-retriever-and-human video is a verified 16-second 480p validation sample made through this website. Its two synthetic input portraits are shown with it. The separately labeled human-duo clip was supplied by the site owner and is used as the scene/audio reference. It is not represented as an output generated through this website.

The 480p sample is evidence of the tested workflow, not a 720p quality comparison. The production preset is 720p. AI may change faces, fur, clothing, hands/paws, anatomy and movement. Exact likeness and frame-for-frame fidelity are not guaranteed. Queue and processing determine completion time; there is no fixed minute guarantee.

The historical Stromae and Jamel Debbouze comedy sketch comes from Made in Jamel (2010): https://www.youtube.com/watch?v=v-wdfQyUSb0 . This independent site is not an official product of Stromae, Jamel Debbouze, ByteDance, KIE or FAL.

## Privacy and help

Account-owned photos and results are not publicly listed. Photos are retained for up to 7 days and finished videos for 30 days; playback and download require account ownership. Rendering uses the named third-party services as described in the privacy policy: https://alorsondanseaivideo.net/privacy .

Examples: https://alorsondanseaivideo.net/examples
Photo and generation tutorial: https://alorsondanseaivideo.net/guides/how-to-make-alors-on-danse-ai-video
Support: support@alorsondanseaivideo.net

## 中文摘要

本站通过 KIE 使用 Seedance 2.5。两张照片分别扮演左侧键盘手与右侧搭档，固定参考片引导白棚场景、动作和镜头。正式规格为 16 秒、720p、16:9 横屏 MP4。模型生成声音关闭，成品通过音视频合成保留指定原音轨；KIE 请求开启 NSFW checker。

公开金毛与人物样例是本站实际生成的 16 秒 480p 验证结果；另一个人物双人片段是站方提供的场景参考素材。两者来源和规格分别标注。AI 可能改变细节，不承诺外貌完全一致、动作逐帧相同或固定完成分钟数。