>> Jump to Chinese Version (跳轉至中文版)
I am a Generative AI Short Film Director based in Hong Kong, China. I specialize in bridging the gap between cutting-edge AI technology and compelling commercial storytelling. My mission is to empower businesses and brands by creating visual content with exceptional impact, narrative depth, and brand value.
This repository serves as my professional portfolio, where I share selected projects, creative experiments, and technical insights.
I provide end-to-end services for creating professional AI-powered short films, including:
- AI Commercials & Marketing Videos: Crafting high-conversion and shareable ad videos for products or services.
- AI Brand Films: Telling unique brand stories and enhancing brand image through visually stunning AI-generated content.
- AI Digital Humans & Avatars: Designing and animating lifelike virtual characters for brand representation and interactive experiences.
My craft is powered by a deep understanding of the generative AI stack, ensuring precise artistic control and high-quality output for every project.
- Advanced ComfyUI Workflows: I build controllable, efficient, and reproducible node-based workflows for precise control over style, composition, and character consistency.
- Model Fine-Tuning & Training: I train custom LoRA/Dreambooth models on specific products, characters, or artistic styles to achieve unique and brand-aligned visuals.
I am always open to industry inquiries, project collaborations, and exploring the future of AI-driven visual storytelling.
Email: iamhk1985@gmail.com
Project Brief: A bespoke AI commercial for China's top-tier tech brand, Huawei. The creative core involves ingeniously using historical leaders to bridge timeless wisdom with the groundbreaking tri-fold form factor of the Mate XT. This vividly demonstrates the device's powerful applications across multiple scenarios, solidifying its flagship positioning as an industry leader.
Impact & Reception:
- The video achieved viral status across 6 major social media platforms in Mainland China.
- It garnered over 1 million views and more than 20,000 likes within its first week.
- It sparked widespread discussion on the fusion of technology and humanities, becoming an industry case study for AI-powered advertising.
My Contribution & Key Techs:
- Role: Served as the AI Director, responsible for the entire pipeline from concept and workflow design to final generation.
- Workflow: I combined a product-specific
LoRAmodel withComfyUIto generate high-fidelity historical figures. Subsequently, I utilizedKlingto create the core AI video clips, which were then edited into the final commercial. - Model Training: To ensure precise and consistent product details, I fine-tuned a custom
LoRAmodel specifically for the tri-fold smartphone.
Project Brief: This film is a sister piece to "The Leader" commercial, centered around the same "dialogue across time" concept. It masterfully features iconic queens from history—celebrated for their beauty, wisdom, and power—seamlessly blending their extraordinary aura with the Pocket2's unique "makeup compact" design. The video vividly illustrates how the phone becomes a fashion statement for the modern woman, boosting her confidence and presence.
Impact & Reception:
- [e.g., The film resonated strongly on female-centric social platforms, successfully penetrating the high-end fashion market.]
My Contribution & Key Techs:
- Role: Served as the AI Director, handling the entire process from concept and workflow design to final generation.
- Workflow: Combined a product-specific
LoRAwithComfyUIto generate high-fidelity historical queens, followed by usingKlingto create the core animated clips for final editing. - Model Training: To ensure impeccable product detail and consistency, I fine-tuned a custom
LoRAmodel for the vertical-folding (clamshell) form factor.
🔒 This project is not publicly available due to a confidentiality agreement. Please contact me for more information.
Project Brief: A bespoke and first-ever corporate AI promo film for one of Hong Kong's most influential travel brands, China Travel Service (CTS). The challenge was to use an unprecedented visual language to convey the brand's core values of "integrity and quality service" while rejuvenating its image to connect with a younger audience. Significantly, this film transcended its promotional purpose, becoming a key strategic tool in negotiations with governmental and business partners, effectively showcasing CTS's innovative and forward-thinking corporate identity.
Impact & Reception:
- The film was highly acclaimed by the CTS management and marketing teams upon delivery.
- It has been successfully deployed in multiple high-stakes scenarios, including brand launches, internal training, and partnership meetings, significantly boosting brand exposure and affinity.
- In a landmark achievement, the CTS Group has scheduled the film for primetime broadcast on TVB, proving that AI-generated content has reached the standard of mainstream media.
My Contribution & Key Techs:
- End-to-End Service: I provided a one-stop, customized service covering everything from initial creative concepts, scriptwriting, and storyboarding to visual generation and post-production.
- Brand Asset Integration: A key technical achievement was seamlessly integrating CTS's existing 3D panda mascot into AI-generated realistic scenes, perfectly merging brand heritage with innovative technology.
- Lip-sync: Utilized
RunwayMLfor precise lip-syncing, making the panda character's dialogue more lively and natural.
🔒 This project is not publicly available due to a confidentiality agreement. Please contact me for more information.
Project Brief: A teaser film for China's prestigious automotive brand, Hongqi, for the market entry of its first electric vehicle in Hong Kong, the E-QM5. The creative approach employed a "reality-virtuality fusion," flawlessly integrating the E-QM5 into iconic Hong Kong cityscapes to build a strong sense of anticipation—"seeing the legend before the car arrives"—within the target community.
Impact & Reception:
- As the initial teaser content, it was precisely targeted and distributed within potential owner communities, sparking enthusiastic discussions.
- The film's quality was highly praised by the client and has been designated as the official in-store promotional film for Hong Kong's first Hongqi Experience Center, becoming a key visual asset for the brand's launch.
My Contribution & Key Techs:
- Solving the Product Consistency Challenge: The project's greatest technical challenge and achievement was overcoming the common issue of 'product deformation' in AI video generation.
- Workflow: My solution involved first seamlessly integrating official Hongqi car images into realistic Hong Kong photos, then using
Klingfor video generation. This workflow ensured that the vehicle's appearance, lines, and details maintained exceptionally high consistency and accuracy even in dynamic scenes.
🔒 This project is not publicly available due to a confidentiality agreement. Please contact me for more information.
Project Brief: For the renowned Hong Kong retail chain Pricerite Home, this project addressed a core industry pain point for its "Silver Living" product line targeting the senior market: standard AI models fail to accurately demonstrate products with special ergonomic designs and often cause severe distortion. Therefore, the core of this project was not just to create a video, but to develop a bespoke AI product demonstration solution for the client.
Impact & Reception:
- Pricerite was highly satisfied with the final product demonstrations, believing it perfectly solved their marketing challenge.
- Based on this project's success, the client has initiated the next phase of strategic collaboration to create custom LoRA models for other product lines.
My Contribution & Key Techs:
- Developed a Proprietary LoRA Model: My primary focus was to build a high-quality, custom dataset for the client's specialized products. This successfully resolved the distortion and functional misrepresentation issues common in traditional AI generation.
- Outperforming Mainstream Models: The final demonstrations achieved a level of detail and functional accuracy that surpassed many leading closed-source models on the market.
- Creating Sustainable Value: This LoRA model is more than a tool; it has become a valuable, reusable digital asset for the client, laying a solid foundation for future product promotions.
🔒 This project is not publicly available due to a confidentiality agreement. Please contact me for more information.
Project Brief: The core IP of Hong Kong-based designer toy brand notoys, "Naughty Bear," had a unique appearance but lacked a clear 'soul' or character identity. My role extended beyond visual enhancement; as an IP consultant, I redefined its world-building and established its core persona of "deceptive charm"—appearing mischievous on the outside but cunning on the inside.
Impact & Reception:
- The client highly acclaimed the "reborn" Naughty Bear, stating that the new image successfully gave the character a new soul and vitality, "as if it were truly alive among us."
- This IP revitalization provided a clear and powerful direction for the brand's subsequent product development, marketing, and content creation.
My Contribution & Key Techs:
- IP Strategy + AI Creation: I served a dual role as "IP Strategist + AI Creative Director." I began by constructing a complete world-view and character profile for Naughty Bear.
- AI Visual Translation: I then used general AI redrawing techniques to precisely translate this new persona into a concrete visual identity.
- Dynamic Interpretation: Finally, using
Kling, I generated multiple personalized scenes that vividly brought Naughty Bear's "deceptive charm" to life, laying a solid visual and narrative foundation for its future story universe.
🔒 This project is not publicly available due to a confidentiality agreement. Please contact me for more information.
Project Brief: This project was not just a remake but a deep creative collaboration with Dr. Gilbert Ho, the "Father of Hong Kong Micro-Films." The goal was to use today's AI technology to fulfill a creative dream from a decade ago: realizing the Arctic scenes for his award-winning film "Arctic Complaint," which were impossible to shoot at the time. We re-envisioned the script and used AI to completely recreate the polar environment, presenting the plight of polar bears in an unprecedented visual manner.
Impact & Reception (Industry Impact):
- Dr. Gilbert Ho gave high praise for the AI version's artistic expression and technical maturity.
- More importantly, this collaboration successfully validated and established an effective AI filmmaking workflow, laying a solid technical foundation for future, more ambitious film and television productions.
My Contribution & Key Techs:
- Script Re-creation & AI Workflow Design: Working closely with Dr. Ho's team, I re-conceptualized the script and designed a production pipeline specifically for AI-driven narrative.
- Pioneering a New Model for Actor Collaboration: I trained a high-fidelity
LoRAmodel for a well-known Hong Kong actress, achieving exceptional character consistency. This was more than a technical breakthrough; it pioneered a new collaborative model for how actors can "star" in an AI film while retaining their unique essence. - Scene Consistency Workflow: I utilized
Flux Kontextto generate highly coherent base scenes of the Arctic, combined withKlingfor core dynamic shots, perfectly solving the scene stability challenge in long-form AI generation.
🔒 This project is not publicly available due to a confidentiality agreement. Please contact me for more information.
Project Brief: A deeply emotional thematic film created for the 40th-anniversary celebration of the Philadelphia Church of Hong Kong. The creative core was to build a "visual time machine," using generative AI to condense forty years of the senior members' lives into a few minutes. It witnesses their journey from youth to the present, showcasing the church's rich history and profound emotional heritage.
Impact & Reception (On-site Sensation):
- The film's premiere at the anniversary celebration caused a sensation among the audience. When the seamless transition from young to old faces appeared on the screen, the venue erupted in thunderous applause and cheers.
- Many senior members were deeply moved seeing their own life journeys. The experience left all guests with a profoundly deep impression of AI's capacity for emotional storytelling.
My Contribution & Key Techs:
- Digital Restoration: I began by performing high-precision digital restoration and enhancement on precious 40-year-old photos provided by the members.
- AI Morphing Visuals: I then used AI Morphing techniques to process the restored old photos with their current-day portraits, creating a stunningly seamless and fluid "passage of time" visual effect.
Project Brief: Shabibi is my own original IP, born from my social observation and cultural response to the current consumerism craze. Through this character, I use dark humor to satirize the phenomenal hype around IPs like Labubu. The goal was to give Shabibi a vivid life, allowing it to spread a message of mindfulness on social media and remind people not to fall into the trap of consumerism.
Impact & Reception:
- The video achieved phenomenal viral status across all major social media platforms in Mainland China.
- It garnered over 10 million views and more than 50,000 likes within its first week.
- It successfully sparked a nationwide conversation about consumerism in the digital age, and my 0-to-1 IP creation and promotion method became a prominent industry case study.
My Contribution & Key Techs:
- IP Creator & Director: I handled the IP's top-level design from scratch, including its world-building, character persona, and the official three-view character sheet for Shabibi.
- Narrative Cohesion Workflow: The key technical innovation was using
ChatGPT-4o's image understanding and generation capabilities to create a series of stylistically consistent and logically coherent scenes for Shabibi's story, effectively ensuring narrative continuity. - Dynamic Generation: Finally, I used
Klingto generate the core dynamic clips, bringing the story to life.
Project Brief: This is a personal creative piece to pay homage to the life wisdom and free-spirited attitude of Hong Kong's legendary "God of Cookery," Chua Lam. Titled after his famous quote, "Don't Eat Bad Food," this short film uses AI to create an experience where audiences can "hear his story, in his own voice." The core technique involves morphing animations between his real-life photos, paired with an AI-cloned voice, to take viewers on an immersive journey through his remarkable life.
Impact & Reception:
- The film received widespread praise on social media, standing out among numerous tributes to Mr. Chua and capturing significant public attention.
- More importantly, this creation sparked a positive discussion on "AI for Good," with many expressing admiration for how generative AI can not only create the future but also gently revisit and preserve memories.
My Contribution & Key Techs:
- Story Curation & Voice Cloning: I first curated Chua Lam's life stories and classic quotes, then used voice cloning technology to recreate his unique voice with high fidelity to narrate the film.
- Visual Time Corridor: I collected a vast number of his real photos from youth to recent years and used AI Morphing technology to connect these static points in time into a seamless, fluid "visual time corridor," vividly portraying his life's journey.
我係一位嚟自香港嘅 生成式AI短片導演。我嘅專長係將最新嘅AI技術同商業創意故事結合，幫唔同嘅公司同品牌，創造出有強烈視覺衝擊力、故事深度同品牌價值嘅影片。
呢個 Repo 主要係我嘅個人作品集 (Portfolio)，喺度會分享我嘅精選Project、創作實驗，同埋一啲技術心得。
我為客人提供專業AI短片嘅一站式服務，包括：
- AI 廣告同 Marketing 短片: 為產品或服務度身訂造高轉化、易傳播嘅AI廣告。
- AI 品牌宣傳片: 透過獨特嘅視覺風格，講好品牌故事，提升品牌形象。
- AI 數碼人同虛擬化身: 設計像真度極高嘅虛擬角色，用喺品牌代言同互動體驗。
我嘅創作建基於對 AI 生成技術嘅深入理解，確保每個項目都可以做到最精準嘅藝術控制同最高質素嘅出品。
- 進階 ComfyUI 工作流程: 我擅長搭建複雜而精準嘅節點工作流程 (Workflow)，實現對畫面風格、構圖同角色一致性嘅精準控制。
- 模型微調 (Fine-tuning) 同訓練: 我能夠為指定嘅產品、角色或者藝術風格，訓練專屬嘅 LoRA/模型，做出獨一無二、又完全貼合品牌要求嘅視覺效果。
我對同業界交流、項目合作，或者一齊探索AI影像嘅未來都好有興趣。隨時歡迎聯絡我！
Email: iamhk1985@gmail.com
項目簡介： 為中國頂尖科技品牌華為度身訂造嘅 AI 廣告片。創作核心係巧妙運用歷史偉人形象，將佢哋跨越時空嘅智慧，同 Mate XT 開創性嘅三摺疊形態結合，生動演繹出部機喺多場景下嘅強大應用，確立其作為行業領導者嘅旗艦級定位。
迴響與成績:
- 影片喺中國大陸 6 個主流社交平台成為熱話，引發病毒式傳播。
- 上線首周就 錄得超過 100 萬次觀看 同 超過 2 萬個 Like。
- 成功引發科技同人文結合嘅廣泛討論，並成為 AI 廣告領域嘅一個行業參考案例。
我嘅貢獻與核心技術:
- 職責： 擔任 AI 導演，由概念、工作流程設計到生成一手包辦。
- 工作流程 (Workflow): 結合產品
LoRA模型，利用ComfyUI生成高像真度嘅歷史人物形象，再透過Kling生成核心 AI 動畫片段，最後進行剪輯製作。 - 模型訓練 (Model Training): 為咗確保產品細節嘅極致還原同一致性，我專門為 Mate XT 微調咗一個三摺疊手機 LoRA 模型。
項目簡介： 本作係《領導者》AI 廣告片嘅姐妹篇，同樣以「古今對話」為核心。創作上巧妙運用多位歷史上集美麗、智慧與權力於一身嘅女王形象，將佢哋嘅非凡氣質，同 Pocket2 獨特嘅「化妝鏡」式設計無縫結合。影片生動咁展現咗呢部手機點樣成為現代女性提升自信同氣場嘅時尚單品。
迴響與成績:
- [此處可填寫影片的具體數據，例如：影片喺女性為主嘅社交平台獲得熱烈迴響，成功打入高端時尚市場。]
我嘅貢獻與核心技術:
- 職責： 擔任 AI 導演，由概念、工作流程設計到生成一手包辦。
- 工作流程 (Workflow): 結合產品
LoRA模型，利用ComfyUI生成高像真度嘅歷史女王形象，再透過Kling生成核心 AI 動畫片段，最後進行剪輯製作。 - 模型訓練 (Model Training): 為咗確保產品細節嘅極致還原同一致性，我專門為 Pocket2 微調咗一個上下摺疊形態嘅 LoRA 模型。
🔒 由於商業保密協議，此項目未能公開播放。如閣下有興趣了解更多，歡迎隨時與我聯絡。
項目簡介： 為香港最具影響力嘅旅遊品牌之一「香港中國旅行社 (CTS)」，度身訂造其史上首部 AI 企業宣傳片。挑戰在於要用一種前所未見嘅視覺語言，既要傳達到品牌「誠信、優質」嘅核心價值，又要為品牌形象注入年輕活力，開拓同後生仔嘅溝通渠道。更重要嘅係，呢條片已超越一般宣傳用途，成為港中旅同政府及商業夥伴洽談時嘅重要工具，有效咁展示出品牌與時並進、勇於創新嘅專業形象。
迴響與成績:
- 短片交付後，獲得港中旅管理層及市場團隊嘅高度評價。
- 已成功應用喺品牌發佈會、內部培訓、合作會議等多個重要場景，顯著提升品牌曝光率同好感度。
- 更難得嘅係，港中旅集團已計劃將呢條片投放喺 TVB 黃金時段，證明咗 AI 創作已可登上主流媒體。
我嘅貢獻與核心技術:
- 全流程服務: 由前期創意、腳本、分鏡設計，到中期視覺生成同後期剪輯，提供一站式嘅定製化服務。
- 品牌資產整合: 最大嘅技術亮點，係將港中旅固有嘅 3D 熊貓吉祥物，無縫融入 AI 生成嘅真實場景入面，實現咗品牌資產同創新技術嘅完美結合。
- 口型同步 (Lip-sync): 利用
RunwayML進行精準嘅口型同步處理，令熊貓角色嘅對白更加生動自然。
🔒 由於商業保密協議，此項目未能公開播放。如閣下有興趣了解更多，歡迎隨時與我聯絡。
項目簡介： 為中國殿堂級汽車品牌「紅旗」，度身訂造旗下首款進軍香港市場嘅電動車 E-QM5 嘅預熱宣傳片。創作上採用「虛實結合」嘅手法，將 E-QM5 完美融入香港多個標誌性實景，為目標客戶社群營造出「未見其車，先見其影」嘅強烈期待感。
迴響與成績:
- 作為首發預熱內容，喺潛在車主社群中進行精準投放，並引發熱烈討論。
- 影片質素備受客戶肯定，已被指定為香港首間紅旗體驗店嘅官方店內宣傳片，成為品牌落地嘅重要視覺資產。
我嘅貢獻與核心技術:
- 攻克產品一致性難題: 本項目最大嘅技術挑戰，亦係最大嘅成果，就係成功攻克咗 AI 生成影片中常見嘅「產品變形」難題。
- 工作流程 (Workflow): 我嘅方案係先將紅旗汽車嘅官方圖素無縫融入香港實景相片，再利用
Kling進行影片生成。呢個流程確保咗喺複雜嘅動態畫面入面，車輛嘅外觀、線條同細節，都保持到極高嘅一致性同準確性。
🔒 由於商業保密協議，此項目未能公開播放。如閣下有興趣了解更多，歡迎隨時與我聯絡。
項目簡介： 為香港著名連鎖品牌「實惠家居」，針對其「銀優生活」系列嘅銀髮族市場，解決一個行業核心痛點：一般 AI 模型無法準確展示具備特殊人體工學設計嘅產品功能，而且會產生嚴重變形。因此，呢個項目嘅核心唔係單純製作一條片，而係為客戶研發一套專屬嘅 AI 產品演示解決方案。
迴響與成績:
- 實惠家居對最終嘅產品演示效果非常滿意，認為完美解決咗佢哋嘅營銷難題。
我嘅貢獻與核心技術:
- 自主研發專屬 LoRA 模型: 我嘅工作重點係為客戶嘅特殊產品，建立一套高品質嘅專屬數據集進行訓練，成功解決咗傳統 AI 生成嘅變形同功能失真問題。
- 超越主流模型嘅效果: 最終生成嘅演示效果，喺細節還原度同功能演示嘅準確性上，甚至超越咗市面上好多主流嘅閉源模型。
- 創造可持續價值: 呢個 LoRA 模型唔單止係一個工具，更成為客戶可以重複使用嘅寶貴數碼資產 (Digital Asset)，為未來嘅產品推廣打下堅實基礎。
🔒 由於商業保密協議，此項目未能公開播放。如閣下有興趣了解更多，歡迎隨時與我聯絡。
項目簡介： 香港潮玩品牌 notoys 旗下嘅核心 IP「壞壞熊」，雖然有獨特嘅外型，但一直缺乏清晰嘅「靈魂」同人設。我嘅工作唔單止係做美術優化，更重要嘅係擔任 IP 顧問，為佢重新定義世界觀，並確立咗「外表古惑，內心腹黑」嘅「反差魅力」核心人設。
迴響與成績:
- 客戶對「重生」後嘅壞壞熊形象高度認可，認為新形象成功賦予咗角色全新嘅靈魂同生命力，「就好似佢真係活喺我哋身邊一樣」。
- 呢次 IP 活化，為品牌後續嘅產品開發、市場推廣同內容創作，提供咗極之清晰嘅方向。
我嘅貢獻與核心技術:
- IP 策略 + AI 創作: 我喺項目中擔任「IP 策略顧問 + AI 創作總監」嘅雙重角色。首先，我為壞壞熊構建咗一套完整嘅世界觀同角色設定。
- AI 視覺轉化: 然後，我利用通用嘅 AI 重繪技術，將呢套全新嘅人設，精準轉化為具體嘅視覺形象。
- 動態演繹: 最後，我再透過
Kling生成多個個性化嘅動態展示場景，將壞壞熊嘅「反差魅力」活靈活現咁演繹出嚟，為後續打造壞壞熊嘅故事宇宙，奠定咗堅實嘅視覺同故事基礎。
🔒 由於商業保密協議，此項目未能公開播放。如閣下有興趣了解更多，歡迎隨時與我聯絡。
項目簡介： 呢個項目唔單止係一次重製，而係我同「香港微電影之父」何緯豐博士嘅一次深度創作合作。目標係用今日嘅 AI 技術，去彌補十年前屢獲殊榮嘅原作《極地控訴》因技術所限，而未能親赴極地拍攝嘅遺憾。我哋重新打造劇本，利用 AI 徹底重現咗極地場景，將北極熊嘅困境以一種前所未見嘅視覺方式呈現。
迴響與成績 (行業影響):
- 何緯豐博士對 AI 版嘅藝術表現力同技術成熟度予以高度讚賞。
- 更重要嘅係，呢次合作成功驗證並確立咗一套行之有效嘅 AI 電影製作流程，為未來更宏大、更精美嘅影視製作，奠定咗堅實嘅技術基石。
我嘅貢獻與核心技術:
- 劇本再創作與 AI 流程設計: 同何博士團隊緊密合作，重新構思劇本，並設計咗一套專為 AI 敘事而設嘅製作流程。
- 開創演員合作新模式: 我為香港知名女演員訓練咗一個高像真度嘅
LoRA模型，成功實現咗極高嘅角色一致性。呢個唔單止係技術突破，更重要嘅係，我哋開創咗一套演員點樣可以喺保留個人神韻嘅前提下，「主演」AI 電影嘅全新合作模式。 - 場景一致性技術流程: 利用
Flux Kontext生成高度連貫嘅極地場景基礎，再結合Kling生成核心動態鏡頭，完美解決咗 AI 生成長片時嘅場景穩定性難題。
🔒 由於商業保密協議，此項目未能公開播放。如閣下有興趣了解更多，歡迎隨時與我聯絡。
項目簡介： 為香港非拉鐵非教會嘅 40 周年慶典，度身訂造一部極具情感深度嘅主題影片。創作核心係打造一部「視覺時光機」，利用生成式 AI 技術，將資深教友們四十年嘅光陰，濃縮喺短短幾分鐘入面，見證佢哋由青蔥歲月到今日嘅成長軌跡，以此展現教會悠久嘅歷史底蘊同深厚嘅情感傳承。
迴響與成績 (現場震撼):
- 影片喺周年慶典上首播時，引發全場轟動。當大銀幕上出現教友們由年輕到年長嘅無縫轉變時，現場瞬間爆發出雷鳴般嘅歡呼同掌聲。
- 好多資深教友睇到自己同身邊朋友嘅成長軌跡，都深受感動。呢次體驗令所有來賓對 AI 技術嘅情感表達能力，留低咗極之深刻嘅印象。
我嘅貢獻與核心技術:
- 數碼修復 (Digital Restoration): 首先，我對教友們四十年前提供嘅珍貴舊相，進行咗高精度嘅數碼修復同增強處理。
- AI Morphing 視覺技術: 然後，我將修復後嘅舊相，同佢哋而家嘅樣貌進行 AI Morphing 處理，創造出一個極度無縫、流暢嘅「時光流逝」視覺效果，完美呈現咗歲月嘅變遷。
項目簡介： Shabibi 係我一手創造嘅個人原創 IP，佢嘅誕生係我對當下消費主義熱潮嘅一次社會觀察同文化回應。我透過呢個角色，用一種黑色幽默嘅方式，去反諷 Labubu 嗰種現象級嘅爆紅，目的係賦予 Shabibi 一個鮮活嘅生命，幫我喺社交媒體上傳播反思，提醒大家喺潮流當中保持清醒，唔好跌入消費主義嘅陷阱。
迴響與成績:
- 影片喺中國內地所有主流社交平台引發現象級病毒傳播。
- 上線首周就錄得超過 1000 萬次觀看同超過 5 萬個 Like。
- 成功引發咗一場關於網絡時代消費主義嘅全民級大討論，呢次由零到一嘅 IP 打造同推廣手法，亦成為行業內一個備受矚目嘅參考案例。
我嘅貢獻與核心技術:
- IP 主理人 + AI 創作總監: 由零開始，一手包辦 IP 嘅頂層設計，包括世界觀設定、角色人設、同埋 Shabibi 嘅角色三視圖設計。
- 敘事連續性工作流: 最大嘅技術亮點係，我利用
ChatGPT-4o嘅圖像理解同生成能力，為 Shabibi 嘅故事生成咗一系列風格高度統一、邏輯連貫嘅場景，有效保證咗成條片嘅敘事連續性。 - 動態生成: 最後再透過
Kling生成核心動態片段，將故事活靈活現咁呈現出嚟。
項目簡介： 呢個係一部純粹嘅個人創作，旨在向香港一代「食神」蔡瀾先生嘅人生智慧同灑脫態度致敬。我哋以佢「別吃難吃的飯」呢句經典名言為題，利用 AI 技術，打造咗一部能夠**「聽佢親口，講述一生」**嘅短片。創作核心係將佢真實嘅人生照片進行 Morphing 動畫處理，再配上 AI 復刻嘅聲音，帶領觀眾沉浸式咁回顧佢嘅精彩人生。
迴響與成績:
- 影片發布後，喺社交媒體上獲得廣泛好評，喺眾多致敬蔡瀾先生嘅內容中脫穎而出，引起網友熱烈關注。
- 更重要嘅係，呢次創作引發咗一場關於**「AI 技術如何向善」**嘅正面討論，好多人感嘆生成式 AI 唔單止可以創造未來，更可以溫柔咁回溯同傳承記憶。
我嘅貢獻與核心技術:
- 故事梳理與聲音復刻: 我首先深入整理咗蔡瀾先生嘅人生故事同經典語錄，然後利用聲音克隆技術，高像真度咁復刻咗佢獨特嘅聲線，用嚟旁白整條影片。
- 視覺時光走廊: 我搜集咗佢由年輕到近期嘅大量真實相片，再透過 AI Morphing 技術，將呢啲靜態嘅時間節點，串連成一條無縫、流暢嘅「視覺時光走廊」，生動咁演繹咗佢嘅一生。