Skip to content

ByteDance Releases Seedream 5.5 for More Stable Image Editing

Baoyu · XRelayed· @dotey·· 3 days agoPicks

4.9K views · 34 likes · 4 reposts (at collection)

Original
Who
ByteDance
What
released the Seedream 5.5 image generation model
When
Quote
字节跳动 Seed 团队 11 月 9 日发布 Seedream 5.5。

What happened

ByteDance’s Seed team released the Seedream 5.5 image model on November 9, with editing that can keep unchanged areas pixel-level stable and generate images with transparent backgrounds. The post reports a severe text error rate of 0.4%, down from 2.8% in the previous generation, and says editing realism trails Nano Banana 2.1 by about 6%. Seedream 5.5 is rolling out across Jimeng, Doubao, and Doubao Work; developers can try Doubao-Seedream-5.5 on Volcano Engine Ark.

Why it matters

The post compares Seedream 5.5 with two named image models and reports both its text error rate and a limitation in image editing realism, giving readers concrete points of comparison.

Read next

See the full storyByteDance Releases Seedream 5.5 for Image Generation and Editing

Reported by 3 sources

    Body · English edition pending (original content)

    字节发布图像模型 Seedream 5.5:同一张图连改 100 轮不走样,能直接出透明底素材,即梦、豆包陆续上线

    字节跳动 Seed 团队 11 月 9 日发布 Seedream 5.5。用 AI 改图最烦的是改着改着别处也变了:换个戒指,脸和颜色跟着偏,几轮下来图越来越糊。

    新版没要求改的区域能做到像素级不变,官方演示里同一个人物连改 100 轮,细节还能对上。它还能直接生成背景透明的图,羽毛这类难抠的边缘都保留,拿来就能叠进海报和设计稿,省掉抠图。照片质感和小字也更准,严重文字错误率从上一代的 2.8% 降到 0.4%。

    字节拿它和 OpenAI 的 GPT Image 2.5 Sunburst、Google 的 Nano Banana 2.1 做了对比,文字错误率最低,凭文字生图的真实感也领先。对 Nano Banana 2.1 的优势不大,拿一张图去修改时的真实感还落后约 6%。

    模型正陆续在即梦、豆包、豆包工作上线,生图时选 Seedream 5.5 即可;开发者可以在火山方舟试用,模型名 Doubao-Seedream-5.5。定价方面,Seedream 5.5 API 延续“输入图 + 输出图”两段计费结构,文本 Token 不产生费用。以2K方图编辑为例,单张成本在 0.28 元左右,不到 Seedream 5.0 Pro 一半。

    我自己也有测试,稳定性和背景透明相当不错,在游戏、电商改图、视频制作等实用场景都有价值,完整的测试见文章。

    官方博客:seed.bytedance.com/zh/blog/mo…

    Quoted postBaoyu@dotey

    x.com/i/article/2119819226856…

    View the quoted post on X

    The story so far

    ByteDance released Seedream 5.5, an image generation and editing model that is rolling out through Jimeng, Doubao, Doubao Work, and Volcengine Ark; developers can try it on Ark as Doubao-Seedream-5.5. ByteDance says the model can preserve unchanged image areas at the pixel level through up to 100 editing rounds and create native RGBA images with transparent backgrounds. A Pro version is planned. In his tests, Baoyu found that eight of 10 rounds of edits to a crowded tourist photo left untouched areas pixel-identical to the prior image, and all 11 edits to a talking cartoon avatar met his pixel-difference criteria. He also reported mixed results for multilingual text and opaque glass in a transparent-background image. ByteDance reports a severe text error rate of 0.4%, down from 2.8% in the previous generation, and says editing realism trails Nano Banana 2.1 by about 6%. For 2K square-image editing, the reported cost is about 0.28 yuan per image, less than half the Seedream 5.0 Pro cost. Under the API pricing model, text tokens are not charged.

    How the story developed · 2

    View full event
    1. 11/10 00:47 (UTC+8)
      Testing Seedream 5.5 for Stable Image Editing and a Talking AvatarBaoyu
    2. 11/10 00:47 (UTC+8)

      ByteDance Releases Seedream 5.5 for More Stable Image Editing This report

    This story is in 11/10 AI DailyRead the whole issue