Gemini 組合圖 (操作/實驗)

#2026-0824

本實驗包括三組，每組由我提供 8 張照片，請 Google Gemini 選擇其中部分（或採用全部）加以組合，創作出一幅（或更多）新穎有趣的圖片。

【第一組】所提供的照片包括：

【第一組】結果產生下圖：

=    =    =    =    =

【第二組】所提供的照片包括：

【第二組】結果產生下圖：

=    =    =    =    =

【第三組】所提供的照片包括：

【第三組】結果產生三圖如下：

= = =

相關文章 (See also)：

1)  An Integrated Picture by Gemini     (2026)

2)  台灣史前文化博物館 (National Museum of Prehistory)     (2026)

3)  都歷遊客中心 (Torik Visitor Center)     (2026)

4)  My Favorite Place to Visit     (2026)

5)  《花田小路．慶祝之日》     (2026)

An Integrated Picture by Gemini

#2026-0822B

I have been using ChatGPT for some time. Recently, I heard that Google Gemini could create drawings and paintings in a more effective way, so I was interested in giving it a try.

I uploaded a few photos that were, in a sense, related to one another. In fact, they were some of the illustrations I had shared in an earlier post, “Welcome Home” (霧峰行腳～圖解篇). The photos were all taken during an afternoon stroll one day in July.

Then I asked Gemini to combine those separate images into a single picture within a frame. Look! In just three or four seconds, Gemini created a wonderful picture, as you can see at the very end of the series below. You can also see part of the same picture in the upper-right corner, though.

= = =

相關文章 (See also)：

1)  家在霧峰 (Mookoo's Home)     (2026)

2)  "Welcome Home" (霧峰行腳～圖解篇)     (2026)

3)  Old Man Drinking Coffee     (2025)

家在霧峰 (Mookoo's Home)

#2026-0822A

Google Gemini seemed to have difficulty producing an accurate, detailed map, although it performed a little better than ChatGPT.

Over the past few days, I have tried Gemini in various ways. I asked it to reproduce a map of Wufeng, integrate several photos into a scenic picture, and correct or change anything I thought was “imperfect” or “wrongly done.” After a series of experiments, I began to see both the strengths and weaknesses of Gemini. It is quite good at creating scenery paintings or drawings, but not so good at designing maps or similar things.

In the upper-right corner is “Mookoo's Home,” a creation of both Gemini and me. Besides Gemini, I also used Microsoft Paint (小畫家), mainly to correct the names of roads and villages.

In the last three images below, you will see three different versions of the same composition. The first version was also created by Gemini, which combined several photos I took near my house into a single picture. See my next post for those special photos.

~ ~ ~ ~ ~

附註: (Copied from Chinese Wikipedia)

Gemini（/ˈdʒɛmɪnaɪ/）是由谷歌DeepMind開發的多模態大型語言模型系列，是LaMDA和PaLM 2系列的繼承者。Gemini 1.0於2023年12月6日面世，被定位為與OpenAI的GPT-4抗衡的產品系列。Gemini家族目前包含Gemini Deep Think、Gemini Pro、Gemini Flash和Gemimi Flash-Lite。

= = =

相關文章 (See also)：

1)  An Integrated Picture by Gemini     (2026)

2)  步道、涼亭、水池     (2022)

3)  "Welcome Home" (霧峰行腳～圖解篇)     (2026)

情人節 ● 書中書 ● 野艾觀景窗

#2026-0819

~~ 寫於七夕當晚 ~~

《情人節 ● 書中書 ● 野艾觀景窗》

無來由喜歡這淒美的

七夕情人節故事

光陰無情果然是真

那些說故事的長輩們

好些都陸續走遠了

要尋找失去的人與事

只能從夢裡找尋

有時找回了幸福片段

美夢卻在瞬間清醒

化作殘存的影音

我怎能把零碎的記憶

拼湊成完好的圖像

如果不藉著好書指引

如果不透過 AI* 觀景

單靠努力仍不行！

圖／文： Mookoo Liang

= = =

* [附註] ： 

　　AI = Artificial Intellegence; 人工智慧。它是指用電腦程式與機器來模擬人類的思考、學習、推理、判斷和解決問題等能力的一種科學技術。

　　本篇標題中的「野艾觀景窗」指的就是利用 AI (野艾 / 爺也愛) 技術（如 Gemini 和 ChatGPT 之類的軟體）創作及修改作者所想要呈現的圖像；或與該軟體進行語文對話、意見交流，從而獲得想要的資料、訊息、或者創意、思辨、統整......等。這麼一來，就好像作者佇立在 AI 觀景台上，或置身於觀景窗前，欣賞美麗的風景，滿足了心靈的某種需求。

　　當然，這仍須倚賴聖書 (e.g. The Bible) 的指引，才能真正體會信仰之「永恆」與科技之「虛擬實境」兩者的區別。祝願讀者諸君，在面對人生的酸、甜、苦、澀之時，心中依然保有（來自上面的）平安。

= = =

相關文章 (See also)：

1)  The Covenant Renewed     (2010; originally, 2009)

2)  浪漫的節日 (The Chinese Valentine's Day)     (2013)

3)  The Lantern Festival and St. Valentine's Day (元宵節加上情人節)     (2014)

4)  Snake Stories     (2010; originally, 2007)

5)  Old Man Drinking Coffee     (2025)