AI图像界暴击! 谷歌香蕉模型Gemini Flash横空出世:角色一致性封神

B站影视 欧美电影 2025-08-30 17:10 2

摘要:8月27日,谷歌正式揭晓代号‘香蕉’的Gemini 2.5 Flash Image,不仅以200万票提前锁定用户认可,更在文生图与图像编辑双榜登顶!

8月27日,谷歌正式揭晓代号‘香蕉’的Gemini 2.5 Flash Image,不仅以200万票提前锁定用户认可,更在文生图与图像编辑双榜登顶!

看看下面我的使用测试看看香蕉模型的效果如何

首先试试看外面传疯了的手办图

turn this photo into a character figure. Behind it, place a box with the character’s image printed on it, and a computer showing the Blender modeling process on its screen. In front of the box, add a round plastic base with the character figure standing on it. set the scene indoors if possible

Create a LOGO icon for a tech company with the text 'small-JJ', using a futuristic sans-serif font. The design should have a geometric abstract style (integrating hexagons with data flow lines) and feature a blue-white gradient paired with a silver metallic texture color scheme. The graphic and text should blend naturally, with clean lines suitable for clear recognition at different sizes, conveying a sense of cutting-edge technology and innovation.

漫画分格

Storyboard in 热血少年 manga style, single-page 4-panel layout, referencing the dynamic expressiveness of *Naruto* Foreground: Golden-haired ninja, orange forehead protector, throwing shuriken, focused gaze, mid-shot following the camera Background: Forest battlefield, flying shuriken, explosion smoke, dynamically blurred leaves, medium depth of field Dialogue box: Vertical rectangular frame, text "忍法·螺旋丸!", yellow handwritten font, bottom left corner Lighting: Top lighting, strong light-dark contrast, predominantly warm tones, creating an intense battle atmosphere

这个功能让我可以联想到做视频分镜的时候可以用到,一致保持得非常好

换装/图像编辑

一个女模特穿着图像上的服饰

改变模特的姿势

把场景转换成室内,调整光线和阴影以匹配环境

把女人头上的帽子去除

把logo添加到衣服上面

━ ━ ━ ━ ━

通过上面的测试可以看出,谷歌的香蕉模型的确是非常强大,一致性和图像编辑方面是现在当之无悔的王者!

虽然还是有缺点,在图像转绘方面是不行的,但那在需要商用的玩家来看,不值一提!

━ ━ ━ ━ ━

这期的分享就到这里了,我们下期再见!下期我们一起来看看怎样才能使用香蕉模型

关注订阅的都是真爱!

来源:ai小健健

相关推荐