[ByteDance Releases Seedream 5.5: Up to 100 Rounds of Continuous Editing of a Single Image, and Direct Generation of Transparent Backgrounds] Beating AI News: ByteDance has released its new generation image model, Seedream 5.5, which supports up to 100 rounds of continuous editing of the same image and can directly generate transparent background materials. ByteDance showcased a demonstration of 100 rounds of continuous editing. Focusing on the same character, each round uses the previous image as a base, constantly changing clothing, glasses, rings, and desktop items. After repeated modifications, the character retains its original features, and the model can accurately adjust small objects such as rings and glasses frames. The official statement claims that the new version can maintain pixel-level consistency in unmodified areas, avoiding the problems of increasing distortion and color deviation with each modification. Another new feature is native RGBA transparent material generation. For example, if the AI draws a snowy owl with outstretched wings, the generated result will have a built-in transparent background, even handling the tiny gaps between the feathers. Stickers, artistic fonts, and product materials can be directly added to posters or design drafts, eliminating the need for manual image cutout. The generated results have also been improved. The new version reduces issues such as overly smooth skin, oily images, and localized overexposure in portraits, and can generate more complex Chinese characters and densely packed small characters. According to ByteDance's official tests, the rate of severe text corruption has been reduced to 0.4%, and the rate of minor text corruption is 1.7%. The usability of generated results for film and office scenes is more than double that of the previous generation Seedream 5.0 Pro. Seedream 5.5 is gradually being rolled out on Jimeng, Doubao, Doubao Work, and Volcano Ark. A Pro version will be released later. [Original Link]