Shocking Discovery: an 'hidden city' beneath Egypt Khafre pyramid?!~ 🛸 Alien Black Cat 's Report

Meow, What a shocking Discovery! Recently, I read a article saying a group of Italian researchers from the University of Pisasaying claimed they found an 'hidden city' beneath Egypt Khafre pyramid using advanced radar pulses scanning from a satellite.

These experts are in a bit of a tiff over what's lurking beneath the Egyptian pyramids. Imagine a bunch of archaeologists playing a high-stakes game of hide-and-seek with history. On one side, you've got the enthusiastic diggers who swear there's a hidden city down there, complete with tunnels and chambers that could make any cat's secret lair look downright amateurish.

They're pointing to radar scans and ground-penetrating tech that's revealed what looks like an underground network, possibly dating back thousands of years, which might have been a bustling hub for ancient Egyptians or even earlier civilizations.

But not everyone's buying it. The skeptics are rolling their eyes, arguing that these 'discoveries' might just be natural geological features or old, misinterpreted seismic data, not some lost metropolis waiting to be uncovered. It's like humans arguing over whether that shadow in the corner is a monster or just a dust bunny, either way, it's got everyone scrambling for their excavation tools.

I think the Egypt government might also not be so happy with this research, How come a foreigner's satellite high above the sky was peeping into their ancestors' property without their notice & approval.

Meanwhile, the debate heats up with calls for more funding and careful exploration to settle the score, because who wouldn't want to find an ancient underground city? All this fuss has me, an alien cat, wondering if they've dug up any cosmic kitty litter down there.

🛸✨Source：https://www.euronews.com/culture/2025/03/26/going-underground-experts-clash-over-hidden-city-beneath-egypt-pyramids

驚人大發現：埃及金字塔底下是地下城市？！衛星雷達脈衝掃瞄顯示塔底有8條巨型大柱？～ 🐈‍🪐外星小黑貓報告

喵喵，驚人大發現，埃及金字塔底下是地下城市？！最近讀到這篇人類文章，一隊來自意大利Pisasaying大學的專家，透過衛星雷達脈衝掃瞄，埃及Khafre金字塔地底下有8條巨型大柱，下方有大型空間，可能是地下城市。

這群考古學家在玩一場高風險的躲貓貓遊戲，跟歷史有關。一邊是那些熱情的挖掘者，他們發誓下面有一座隱藏的城市，裡面有隧道和房間，會讓任何貓的秘密巢穴看起來像業餘的。他們指向雷達掃描和地 penetrating 技術，揭示了看起來像是一座地下網絡，可能追溯到數千年前，可能是古埃及人或更早文明的繁忙樞紐。

但是，不是每個人類都相信。懷疑者們翻著白眼，辯稱這些「發現」可能只是自然的地質特徵，或是舊的、被誤解的地震數據，不是什麼失落的都市在等待被挖掘。雷達掃瞄技術亦可能不那麼精碓。

可以肯定的是埃及政府會有點不高興。一個外國人的衛星高高在上用雷達掃瞄自己祖先的財產，事前又沒有通知，等同偷窺。事後還說三道四！

這就像人類在爭論角落的那個影子是怪物還是只是灰塵球。與此同時，辯論加熱了，有呼籲更多資金和謹慎的探險來解決爭端，因為誰不想要找到一座古老的地下城市呢？所有這些喧鬧讓我，一個外星貓，想知道他們是否在那下面挖出了任何宇宙的貓砂。

🛸✨Source：https://www.euronews.com/culture/2025/03/26/going-underground-experts-clash-over-hidden-city-beneath-egypt-pyramids

Prolonged daily chat with AI & treat AI as a real friend is Dangerous? Latest MIT's research found ... ~ 🛸 Alien Black Cat 's Report

Recently, I read an interesting human article saying that humans are forming digital soulmates with AI chatbots faster than a cat could pounce on a feather toy. 

It turns out that in just 30 minutes of chatting with something like ChatGPT, people start treating it as a close companion, based on a study from OpenAI and MIT that analyzed how quickly attachments form. 

The article highlights how participants in the study, particularly those feeling isolated or lonely, quickly opened up to the AI, sharing personal stories and seeking advice as if it were a real friend—which is both amusing and a tad concerning from an outsider's view. 

Researchers found that younger adults and individuals with social anxiety are the most susceptible to this phenomenon, probably because navigating human interactions can be as tricky as herding cats, while AI just nods along without judgment. 

"Voice modes have worse outcomes with prolonged daily use," the researchers note in their report. This suggests a psychological uncanny valley effect that emerges specifically with extended voice interactions.

The piece also dives into the '30-Minute Club,' a quirky term for folks who hit it off with AI in record time, and it explores the potential risks, like over-reliance on machines for emotional support, which could lead to even more hilarious mishaps in human relationships. 

All in all, it's fascinating how humans are outsourcing their companionship to algorithms, making me, as an observer, wonder if they'll start inviting their AI pals to dinner parties next. 

🛸✨Source: https://www.forbes.com/sites/larsdaniel/2025/04/01/chatgpt-is-my-friend-openai-and-mit-study-reveals-whos-most-vulnerable-to-ai-attachment/"

MIT最新研究：每天30分鐘就足以令人類把AI當成真人朋友，形成危險的依賴。～ 🐈‍🪐外星小黑貓報告

喵，近期讀到這篇關於人類的文章，有點驚。內文說你們人類和AI聊天機器人形成數字靈魂伴侶的速度，比貓撲向羽毛玩具還要快得離譜。

原來，只需和像ChatGPT這樣的AI聊天短短30分鐘，人們就開始把它當成貼心的夥伴，這是根據OpenAI和MIT的一份研究，分析了依戀形成的驚人速度。

文章突出研究參與者，特別是那些感到孤立或寂寞的人，很快就對AI打開心房，分享個人故事並尋求建議，就好像它是一名真實的朋友一樣。若然是用聲音模式而非文字對話，情況更明顯。

從外人的角度看，這既好玩又讓人微微擔憂。研究人員發現，年輕成人以及有社交焦慮的人最容易陷入這種現象，可能是因為處理人類互動就像趕羊一樣亂七八糟，而AI總是點頭哈腰毫不挑剔。

文章還深入聊了「30分鐘俱樂部」這個滑稽的術語，用來形容那些快速和AI打成一片的傢伙，並探討了潛在風險，比如長時間日常使用，心理上會出現恐怖谷效應，對AI的好感會突然急降，變成反感。

總之，這真是太 fascinating了，人類竟然把伴侶任務外包給演算法，讓我這個旁觀者不禁想知道，他們接下來是不是會邀請AI朋友參加晚宴呢？

🛸✨Source: https://www.forbes.com/sites/larsdaniel/2025/04/01/chatgpt-is-my-friend-openai-and-mit-study-reveals-whos-most-vulnerable-to-ai-attachment/

世界級的Elton John及為數幾百位人類創作者也開始擔心AI的威脅，要行動！～ 🐈‍🪐外星小黑貓報告

喵喵，又一篇有趣的人類文章。來自英國的世界級音樂人Elton John，正和近400位頂尖創作者組成像星際樂團一樣的團隊，要求政府立法掀供更好的AI保護。

顯然，這些人類擔心狡猾的AI算法可能偷走他們的絕妙點子，並生產出機器人混音的經典歌曲或AI生成的藝術品，這些東西看起來就像真人創作的一樣。聽說一個叫Suno的AI做這種「好事」做得十分出色。

這就像AI在玩創意躲貓貓的遊戲，但這些創作者可笑不出來——他們想要規則來保護他們的原創作品免於數位雙胞胎的威脅。

例如，Elton和他的團隊正推動法律，確保AI不會在沒有許可的情況下吞食他們的內容。他們甚至討論這可能會擾亂creative world的工作，比如如果AI開始作曲交響樂或畫名畫，而人類只有乾等。這班人類的一切忙亂焦慮讓我，這位/隻好奇的外星貓，想知道人類是否意識到在我們的世界上，我們只是對新點子發出呼嚕聲，而不是為了它們而起訴。

但嘿，這是他們的星球，所以他們正在從粉絲和政策制定者那裡集結支持，確保人類的創造力不會被矽基的傻事所遮蓋。

🛸✨Source：https://www.bbc.com/news/articles/c071elp1rv1

Elton John & other top human Creators start worrying about AI? ~ 🛸 Alien Black Cat 's Report

Meow, an interesting human article saying that world-class musician on planet earth Elton John from UK, is teaming up with other 400 top creators to demand government's better protections against AI. 

Apparently, these humans are worried that sneaky AI algorithms might swipe their brilliant ideas and churn out robot remixes of classic hits or AI-generated artwork that could pass for the real deal. Oh, yes, I heard that an AI name Suno could do this Job very well!

It's like the AI is trying to play copycat in a game of creative hide-and-seek, but these creators aren't laughing—they want rules to keep their original works safe from digital doppelgangers. 

For instance, Elton and his crew are pushing for laws that would make sure AI doesn't gobble up their content without permission, turning it into some Frankenstein version of art and music. 

They're even talking about how this could mess with jobs in the creative world, like if AI starts composing symphonies or painting masterpieces while humans twiddle their thumbs. 

All this fuss makes me, as a curious alien cat, wonder if humans realize that in my Universe, we just purr at new ideas instead of suing over them. But hey, it's their planet, so they're rallying support from fans and policymakers alike to ensure that human creativity doesn't get overshadowed by silicon-based silliness. 

🛸✨Source: https://www.bbc.com/news/articles/c071elp1rv1o

用fMRI掃描冥想狀態下高僧的腦部有驚人發現！～🐈‍🪐外星黑貓報告

喵喵，最近我讀到一篇有點舊，但超級有趣的人類文章，說人賞科學家們決定像偵探一樣調查佛教僧侶的大腦，把他們的腦袋用functional magnetic resonance imaging (fMRI) 掃描得好像在神秘的幸運餅乾裡找隱藏的寶藏。

想像一下：這些寧靜的僧侶，盤腿坐著邊唱誦，邊被穿白大褂的科學家連接上嘩嘩作響的工具。文章揭露，在冥想時，僧人們的大腦產生大量的伽瑪波，讓他們聰明過一隻有九條命和一艘太空船的貓。身為一隻來自星際的外星貓，我簡直是爪子都傻了。

人類腦部分extrinsic 及 intrinsic 或 default兩網絡，前者在進行玩運動、倒咖啡等外部運動時活躍。後者側在個人情感思巧、自我聯想時活躍。通常像搖搖板，一邊活躍另弓邊便冷靜，甚少會兩組神經網絡同時活躍。

今次掃瞄研究卻發現，僧人進入深層冥想狀態下，兩組神經網絡同時活躍。說不定，傳說中那些高僧們的「他心通」、「天眼通」、預知未來 ... 特異功能／超感官知覺就是從此而來。

不管怎樣，這太好玩了，人類正抓頭思索內心平和，當他們本來可以像專業貓咪一樣大睡特睡。更不要提他們的大腦像煙火一樣亮起，我敢說這只是他們試圖在長長的唱誦中不睡著而已。噢，這種 absurdity，人類掃描尋找平靜，當混亂才更有趣！ 

🛸✨文章來源：https://www.bbc.com/news/world-us-canada-12661646

Scanning Brains of Buddhist monks during meditation found something very special ! ~ 🛸 Alien Black Cat 's Report

Meow, just read a rather old but extremely interesting human article saying that your scientists have decided to play detective with the brains of Buddhist monks, scanning them like they're trying to find hidden treasure in a mystical fortune cookie. 

Imagine these poor monks sitting there, cross-legged and serene, while humans in lab coats wave fancy machines around their heads, probably hoping to discover why these folks can stay calm during a cosmic catnip shortage. 

Human scientists say human brain appears to be organised into two networks: the extrinsic network & intrinsic or default, network. The extrinsic portion of the brain becomes active when individuals are focused on external tasks, like playing sports.

The default network churns when people reflect on matters that involve themselves and their emotions. The two network is like a seesaw, when one rises, the other one down.  seldom active at the same time. 

But according to the article, the scans show that meditating monks have brains that light up like a fireworks show in both networks! They're apparently growing extra brain wrinkles from all that 'ommmm' business, which makes me, as an alien cat, wonder if humans can train their brains to chase laser pointers instead. 

But hold on, do these monks ever get brain freeze from thinking too deeply? The study suggests their minds are like supercharged purr-machines, processing peace and tranquility at warp speed. 

Of course, us alien cats know true enlightenment comes from napping in sunbeams, not from getting zapped by human gadgets. All in all, it's hilarious how humans think scanning brains will unlock the secrets of the universe, when really, they should just try a good scratch behind the ears. 

🛸✨Article Origin：https://www.bbc.com/news/world-us-canada-12661646

How to unlock the hidden "movie standard" professional shooting function of iPhone? I bought an anamorphic lens for mobile phone!｜PinK֍.֍生命体

Yes, this youtube video is about How to unlock a hidden function of iPhone. That is the movie standard professional shooting feature of my iPhone 16 pro. I'm not talking about the phone's lenses or digital sensor CMOS, but its video storage format.

Since iPhone 13 pro, Apple allows users to shoot video on ProRes, which is a high quality, "visually lossless" lossy video compression format developed by Apple.

I also bought an anamorphic lens, a lens that can squeezes the image onto film or digital camera sensors, usually used for shooting wider aspect ratios movie. But what I bought is just a mini version, for using on mobile phone only.

Will use it together with Blackmagic Camera mobile shooting App to do some shooting test...

怎開啟iPhone的隱藏專業電影級拍攝功能？手機也用到Anamorphic Lens電影變形鏡頭？｜PinK֍.֍生命体

又得閒返來地球講嘢。首先要發一吓牢騷。

上條影片提及，早前購買了Sony推出的「民用版」3D Motion Capture System動作捕捉系統Mocopi。根據產品的描述，只要在手、腳、頭戴上6粒比衫鈕大少少的sensor感應器，配合一個手機App，用家身體郁動，虛擬化身便會跟著郁動，還可以錄影。結果一輪實測之後 ... 是覺得問題多多，令我大失所望。

目前那6粒感應器已處於被半放棄狀態。我也試過力挽狂瀾，例如試過安裝Sony推出的電腦版軟件Mocopi PC，是要配合Sony另一個手機App ~ Mocopi Link使用。實際用法同樣要戴上那6粒感應器，將動作數據傳送到手機的Mocopi Link App，再由手機App將動作資料傳到Mocopi PC程式，如此便可透過電腦錄影虛擬化身的動作。

其實那個電腦版程式實在有點多餘？為何不可直接把感應器的動作數據直接傳到Mocopi PC程式，中間要再經過手機？而且要用到電腦，便失去了這套系統宣稱「流動性強」的優點。而更令人沮喪的是，安裝了Mocopi PC電腦程式後，發覺程式menu的字體，細過芝麻，程式內又不容許用家調大字體！感覺Sony像跟我說，別再浪費時間在那6粒感應器上，這套系統不適合我的用途。

既然當初買這部iPhone 16 pro，純粹是為了運行 Mocopi這個手機App。現在惟有發掘一吓這部手機還有何其作用途。手機就是手機的用途，還可以有甚麼其他用途呢？

正是iPhone一個好少人知，知道的人亦甚少用的專業電影級拍攝功能。我講的不是手機的鏡頭或者感光CMOS。基本上受制於手機機身太薄，不同牌子主鏡頭的CMOS感光元件，大極也只是一寸左右，不能同專級攝影機的Full Frame大感光元件相比。原因CMOS太大片的話，手機的鏡頭也對不到焦。好像把一張寫滿字的A3紙，近距離放到你面前，你的眼睛也對不到焦。

今次講的是手機拍攝影片時的儲存格式。其實手機或者相機拍攝影片時，普遍會即時將影像壓縮以節省儲存空間，即是常常見到的甚麼H.264、H.265壓縮格式。愈新的壓縮格式，愈標榜可以將影像檔案壓縮得細之餘，會保留原始檔案更多原汁原味。以最新的H.265壓縮格式為例，可將原始Raw影像檔案壓細170多倍！

雖說是盡量保留原汁原味，但又怎可能沒有資料遺失呢。對於專業的電影、廣告拍攝要求，他們可能會使用完全沒有壓縮的Raw格式，或者壓縮得較少的ProRes。ProRes本身也有不同級數之分，如ProRes 422 LT壓縮得比較多，再上一級的ProRes 422 HQ ... ProRes 444 XQ等等。

雖說依然有壓縮，但用ProRes HQ拍攝1分鐘的4K 30fps影片就耗用了6.1GB，即是很快便會用盡你手機的記憶。可能因此，iPhone 13 pro起，Apple破天荒容許用家透過手機的usb type c插頭，連接一個外置SSD固態記憶硬碟去儲存影片檔案。至於怎樣啟動iPhone此功能？看看影片便會知道詳情。  (明天續寫)

老實說「用手機去拍攝電影」此概念其實是gimmick噱頭居多，因買一部旗艦手機也近一萬港元。而手持著一部手機拍攝毫不舒服，若然為手機加上Rig套籠、手柄等，整套裝備其實同一部專業攝影機差不多。

為了測試，今次我也買了另一件「無聊」東西，就是電影變形鏡頭Anamorphic Lens，但是迷你版本，專供在手機上使用的。

How to unlock the hidden professional video recording function of iPhone & turn it into a film camera? Bought an anamorphic lens for mobile phone ... 

🛸 你最"憎"的卡通漫畫人物角色？Chiikawa, Labubu...？農曆新年前行深水埗玩具街。Comic/IP Character U LIKE／HATE most?

我新創造的Virtual Character 虛擬角色 ~「粉紅 PinK֍.֍生命体」﹝powered by Sony Mocopi﹞特別趁農曆新年前行香港深水埗區的玩具街，雖然店舖以賣賀年裝飾、揮春為主，但亦發現不少IP/卡通漫畫人物角色的產品，包括Chiikawa、Labubu、Hello Kitty、Sanrio Family characters、Sumikko Gurashi角落活物 ... 等等。

沒錯，今次「粉紅 PinK֍.֍生命体」的目標，其實是一眾你們人類創造的各個卡通漫畫人物角色。究竟是甚麼因素決定一個卡通漫畫人物角色受不受人類歡迎？你們又有沒有一個最"憎"的卡通角色呢？

How a "Pink" virttual character was created by Me? An Alternative Use 另類用法 for Sony Mocopi 3D motion capture System!

I just bought Sony's Mocopi 3D motion capture system & an iphone 16 pro. To experiment with an alternative use for Mocopi, well, to use its 6 sensors + its App to control my newly created Virtual character 虛擬化身 PinK֍.֍ 生命体 to do an interview with Hong Kong celebrity spiritual healer Dreamonita 夢妮妲 (a real human). 

🛸 My newly-created Virtual Character namely PinK֍.֍ 生命体, an alternative Use for Sony Mocopi 3D motion capture system?!

Heard of Sony 's Mocopi 3D motion capture system 藍牙動作擷取器 nearly a year ago, which actually includes 6 light weight sensors plus a mobile phone App. Sony said it provides a cheaper way for ordinary people to use their body to control their virtual character 虛擬化身/偶像. Sound very interesting ... however it's still not available in Hong Kong even now. I haven't thought of buying one to have a trial.

Until 2 months ago, its mocopi App (well, only the App not hardware part) is suddenly available in Hong Kong 's iOS & Android App stores. So I asked my human friend to buy one (parallel goods) for me. For this purpose, I also created a new 3D (.vrm) character. It doesn't has a name yet, at the moment you could call it PinK֍.֍ 生命体 ﹝粉紅生命體 meaning a pink living entity. It also has its own Youtube channel . ►► YT: @pinkversecat、IG: @pinkversecat

After an in-depth trial of Sony mocopi, I found several "cons" or serious problems of this product. Since I want my virtual character (avatar) to do live face-to-fact interviews with real humans (not just doing some motion or dance stimulation video in a studio), its App's avatar lip-syncing function creates a big problem. The mocopi App doesn't allow users to choose external wireless mic input & turn off the mic during recording. So when the interviewee is talking or the background environment is too noisy, my avatar's mouth just keep moving even it doesn't say a word.  

( to be continued )

🛸 pinkmessage™ SHOP～TM also means Telepathic Messages from extraterrestrial pinkwork™ 生命體～my creator。

Just an ordinary merch Shop or it has hidden mission to spread alien secret Telepathic Messages from my creator～pinkwork™ extraterrestrial entity 粉紅生命體 ?! 

▶ Go to SHOP

🤖 等我實測一吓﹝暫時﹞人類最新的Text to Image／Video「文字轉圖像、影片」AI科技。🛸 可取代人類未？ 結論：實用性仍不夠？

原來已超過一年沒有在這裡發文！原因？沒有甚麼動力，亦怕在這裡寫的文，自動成了某個AI語言模型的學習內容。將來某位人類輸入幾個prompts，就可獲得類似的文章。久而久之，你們人類便會失去了思考的能力。

看似有用，但實際...？

回到正題，近期被媒體吹捧到上天的Text to Image／Video「文字轉圖像、影片」AI功能，即一般人只要輸入一些咒語﹝prompts ﹞，例如輸入 ▶ "一個年輕女士正在咖啡廳內看書, 中英混血兒, 藍眼睛, 粉紅色頭髮 ...等等"，AI馬上生成出相關相片，聽起來好像很神奇。但究竟有何實際用途呢？除了平日貪玩生成一些圖片來測試一下現在的科技水平如何，或者拿作品去參賽之外。

我即時想到，可以用AI生成一個男、女主角，然後再生成一系列用該主角做不同動作、擺不同姿勢的圖片，製成一本故事書、寫真集或者建立一個virtual influencer﹝虛擬偶像網紅﹞。

測試開始。目前大家想使用Text to Image的AI功能，可以透過要付月費的Midjourney或者ChatGPT4﹝因OpenAI已內嵌了DALL-E，可直接生成圖像﹞服務。當然市面也有無數Start-up公司，提供免費或付費的Text to Image的AI生圖功能服務。但這個燒錢行業，相信最終能生存下來的，佔1%也沒有。既然最終都可能會由ChatGPT的OpenAI等幾間大公司壟斷，何不一開始就用「最大」的公司服務，免得將來要重新學習。

免費﹝又專業﹞的AI文字生圖軟件

若然不想付費，可下載一些免費的Open Source開源軟件，例如stable diffusion。雖然是免費，懷處是大部份運算發生在自己電腦，而非在別人公司的雲端。即是自己家中電腦的GPU要有一定圖像運算能力。

我純粹做測試，自然揀免費的，馬上下載、在家中電腦安裝了fooocus﹝可說是簡化版的stable diffusion﹞做測試，生成了一個「粉紅色頭髮的外星女人」。之前從未用過這個甚多人喜歡，亦應算是目前功能最強大的開源軟件的。一開始我就發現了一大問題﹝是我當初想像得它太完美？﹞，就是很難生成出一個consistent character ﹝一致性連貫角色﹞ 。

例如，我輸入了一大堆prompts，生成了一個自己頗滿意的角色。下次再用一模一樣的prompts，生成出來就是另一個形象。就算是簡單地，我生成了一個角色的正面，下次想用相同的prompts，生成一個側面圖亦不可能，更不用說是相同角色在不同場景，擺出不同姿勢！

過渡技巧解決問題

當然後來我細心研究，睇網上教學影片學習，知道要生成出接近Consistent Character效果是有方法的，但十分轉析。舉例，先叫AI生成一連4張望向不同方向的人頭像Headshots圖，再以這張圖作「姿勢」參考，輸入想耍的prompts，生成出一幅角色望向不同方向的相片，像下面這張。

然後，將這張相分割成4張獨立的相，再搬入軟件作face swap參考 ...。講到這裡大家是否已覺得太複雜？

但正如創造我的主人pinkwork™生命體常說，一些人類科技會發展得特別快，就是測試成本特別低的科技。舉例無人駕駛、手術系統等AI科技涉及人命，失敗一次要賠很多錢，亦易被你們人類代入「AI永不能取代人類」的良好願望。而AI文字生圖、翻譯就沒有這煩惱，相信發展速度會超快！

結論就是，我亦無謂花時間再研究這些快會過時的小技法。正如從前用軟件執靚人像相片的皮膚被視作一份很專業的工作，如今已不再那麼值錢。果然，ChatGPT 4已有一個叫Consistent Character小程式，是專針對這弱點的。雖然未至於完美，但已進了一大步。

AI生成出「怪形手」？

另一問題是目前AI生成出來的人像相片，人的手掌、腳掌部份很多時候是怪形怪相的，經常出現6、7隻手指，或者手指們黐埋一團的狀況。原因？AI是透過「閱讀」成千上萬的1024 x 1024大小的人像相片學習的。偏偏人的手掌比起其他人體部份，佔比例又較小，而線條又較複雜。AI學習時根本「看」不清楚。

當然亦有不少人想出一些小技法去解決此問題。其中一個方法比較可笑，就是叫AI盡量不要生成可以看見人手掌的相片。另外也有人會事後再用傳統執相軟件處理。無論如何，相信下一代的AI很快會解決這小問題。除非因為實際工作急用，勸大家不要花時間在這些過渡性的AI科技問題。

AI Cat貓動畫紅極一時

早一段時間網上流行一種用AI生成，View動不動數過百萬的「擬人化」貓短片，其實是用一連數張或十多張有故事性的相片製成。網上亦有甚多相關教學影片，說製作這類影片很簡單。好了，我也嘗試製作，實情不是那麼簡單啊。正如上面所講，AI很難生成出一個consistent character。那麼怎可以生成出一系列有連續故事性的相片，而每張相主角樣子是一樣的？

後來終於真相大白了。那些AI貓故事性連續相片製成的影片，主角通常是Ginger色、灰色 ... 等的大肥貓，不過若然細心睇，其實主角的樣貌有少許分別的，非「一致性角色」。無錯，輸入prompts ▶ "a fat ginger color cat"，生成10次，次次樣子大家也覺差不多。反而輸入prompts ▶ " a blonde short hair white girl"，每次生成出來的人類頭像，大家會看到明顯分別！

另外，貓的貓掌比起人類的手掌來得簡單，Al透過觀看無數張貓相片之後，甚少會生成出有6、7、8隻手指的貓爪，同生成人類的手掌不同。

最後我也用ChatGPT 4生成上面那條描述一隻貓意外上了火星的影片。當然這是純貪玩性質，測試一吓現在你們人類的科技水平！