Logan Kilpatrick
@OfficialLoganK · 1周前公告:我们的团队全天候在线,帮助客户使用 Gemini 3 Pro 和 Nano Banana Pro 进行扩展,请告诉我们您需要什么(包括更高的 API 速率限制)!我的邮箱是 Lkilpatrick@google.com
262
102
2,461
646
542
公告:我们的团队全天候在线,帮助客户使用 Gemini 3 Pro 和 Nano Banana Pro 进行扩展,请告诉我们您需要什么(包括更高的 API 速率限制)!我的邮箱是 Lkilpatrick@google.com
ICYMI:上周我们开始了推出我们全新的最先进的图像生成和编辑模型,Nano Banana Pro。
了解它如何将视觉效果提升至新高度 🧵
🍌 nano banana promt
This prompt creates a Pixar-style 3D animated scene showing a cheerful, personality-driven group selfie moment, featuring characters with expressive poses and culturally-inspired clothing. Users customize the characters, background, and details to vividly represent their unique personalities and cultural contexts.
(Please change the names )
---
A vibrant Pixar-style 3D animated scene depicting a joyful group selfie moment featuring <group of characters> in a <culturally representative environment>.
At the center, <main character> confidently holds a selfie stick topped with an iPhone, wearing an expression that clearly reflects their <distinctive personality trait> and exudes <leadership or core presence>.
To the left, <character A> adopts a pose or action reflective of their <distinct personality trait>, showcasing an expressive face that vividly captures their <personality description>.
On the right side, <character B> strikes a playful/humorous/cute pose, holding a characteristic item (<character B’s representative object>), with an exaggerated, lively facial expression highlighting their <distinctive personality trait>.
Additional characters (optional):
Nearby, <character C> performs an action or posture aligned with their personality, bearing an expressive facial expression that encapsulates their unique personality traits.
All characters wear bright, cheerful, and adorably rounded outfits styled in a contemporary fusion of traditional and modern attire representative of their cultural or historical backgrounds.
The scene is warmly lit, colorful, and filled with dynamic expressions and lively poses.
The background features a setting emblematic of the characters' cultural identities or personalities—such as cherry blossoms, lakes, mountains, historic architecture, or fantasy-like natural landscapes—rendered in the adorable, cinematic style characteristic of Pixar animations.
The overall composition exudes energy, humor, and heartwarming joy, capturing the essence of each character through their selfie expressions and postures.
---
Names: [Frodo, Sam, Aragorn, Gandalf, Legolas, Gimli]




🍌 nana banana pro Prompt
(optional: add your city/language/date at the bottom)
----
Please create a cute, stylish calendar illustration in a vertical (9:16) layout featuring a fresh, bright, hand-drawn style:
Illustration Requirements:
- The main character is a young, fashionable female with a cute and lively watercolor or hand-drawn texture, vibrant yet soft colors.
- Character features include large eyes, rounded rosy cheeks, and bold, fashionable accessories (e.g., sunglasses, hoop earrings, headscarves, headbands, bows, knit hats, etc.). Clothing should be bright, with dynamic and playful poses. Proportions may be slightly exaggerated (e.g., larger head, slender waist).
- Outfit and accessories must reflect seasonal elements, holidays, recommended activities ("auspicious items"), or distinctive local characteristics based on the user's location and input. Outfit description: [{Character Outfit Description}]
- Character positioned centrally or slightly right-aligned to leave ample whitespace for textual content.
- Pure white, minimalist background without additional decorative elements, emphasizing the character and text.
Calendar Layout:
- Prominent position at the top center: Gregorian date number [{Gregorian Date Number}] (large and eye-catching).
- Below the date number, display the English month [{English Month}].
- Below the English month, display the year [{Year Number}].
- Symmetrical layout left and right of the date: weekday in both local language [{Weekday in Local Language}] and English [{Weekday in English}], along with the lunar date and local holiday [{Lunar or Local Calendar Date}] [{Local Holiday}], ensuring clear, elegant fonts.
"Recommended Activities" and Inspirational Quote:
- Vertically aligned on the left side in bold handwriting: [{Recommended Activities}], using brush calligraphy for Chinese users and complementary handwriting style for other languages, slightly larger and vertically arranged.
- To the right of "Recommended Activities," arrange vertically an inspirational and comforting quote [{Inspirational Quote}] in slightly smaller matching handwriting.
Localized Elements:
- Incorporate distinct local cultural elements or landmarks based on the user's current location or input into the character's outfit, accessories, or details (e.g., city landmarks, climate characteristics, local cultural motifs).
General Guidelines:
- All elements must be neatly arranged with balanced whitespace.
- Ensure text readability without overlapping or obscuring the illustration.
- Replace all placeholder content with information dynamically generated based on user input or system-provided user data.




SAM 3D 正在帮助推进康复的未来。
了解 @CarnegieMellon 的研究人员如何使用 SAM 3D 捕捉和分析临床环境中的人体运动,为实现个性化、数据驱动的康复方案奠定基础。
🔗 了解更多关于 SAM 3D 的信息:go.meta.me/305985
Why are the images generated by Nano Banana blurry? To get the high-definition version, please hover your mouse over the image and click the "Download fullsize" button in the top-right corner. By default, Nano Banana Pro generates a lower-resolution preview; it will only generate the full high-quality version when you initiate the download.
用Gemini里的Nano Banana Pro出图有时会出现图片质量下降很严重的现象,简单来说就是图片变「糊」,这个情况你们有遇到过吗?是怎么解决的?(我是Ultra会员
“学术界尤其是在 AI 中的角色,在过去十年中发生了很大的变化。
5 或 10 年前,你真的可以在实验室里用几个 GPUs 训练出顶尖模型……[现在]你不能。
这意味着围绕我们作为学术界人士应该做什么的期望发生了一些变化。
它不应该是关于尝试训练最大的模型和扩大最大的东西。它应该是关于尝试古怪的想法和新的想法以及疯狂的想法。”
Justin 阐述了由于 AI 领域发生的创新,研究人员的目标需要如何改变。
@jcjohnss
Fei-Fei Li 教授在斯坦福任教多年,并构建了 ImageNet——一个催化深度学习革命的里程碑项目。现在,她与之前的学生 Justin Johnson 共同创立 World Labs,All in 空间智能,致力于成为 AI 的下一个前沿。
在 Latent Space 的播客中,我们与 Fei-Fei 和 Justin 坐下来,深入探讨世界模型的“AlexNet 时刻”会是什么样?为什么 transformers 实际上是集合模型而不是序列模型?高斯溅射、物理先验和新的硬件原语如何重新定义 GPUs 之外的堆栈?在前沿实验室时代,学术界面临的角色转变是什么?将 AI 植根于物理现实如何永远改变机器人技术、医学和创意设计?以及他们押注具有空间感知能力的世界模型,将是从今天的仅文本 LLMs 到实际生活并在 3D 空间中行动的代理的桥梁。
@drfeifei @jcjohnss @swyx @FanaHOVA
@CrowdStrike @PayPal @Synopsys 看团队如何将开放模型打造成领域特定 AI 代理,从而提高准确性、速度和投资回报率 → nvda.ws/47T25Hk