Project Name: The Origin of Flower Knows Garden
1. What Inspired Me
Our inspiration stems from two core sources:
- The core proposition of the beauty brand Flower Knows — "break free from constraints and embrace diverse colors". Its brand story advocates breaking through a single aesthetic and encouraging individuals to bravely showcase their unique charm, which highly aligns with Gen Z's demand for personalized expression.
- The practical requirements of the "Cross-Media" course: We aim to make the proposition "Every color can freely bloom" more intuitively perceived through a short film.
The brand's precise positioning (targeting Cosplay enthusiasts, beauty bloggers, etc.) also led us to determine a "fantasy and inspirational" story tone. We hope to convey the brand's core spirit of "courage + color" through the adventure of the young girl Xiao Lai guarding colorful flower seeds.
2. What I Learned
1. Cross-media content creation logic: Mastered the complete link of "story narrative → multi-platform content adaptation → user interaction", and understood how to strengthen the communication of brand concepts through short films, UGC activities, and offline installations.
2. Integrated application of AI tools: Proficient in using:
- Visual asset generation: Midjourney, LTX, Weavy (characters, scenes)
- Audio production: Eleven Labs (dubbing, sound effects)
- Script optimization: Kimi
- Video editing: Jianying (CapCut), LTX, Midjourney
Learned to collaborate with multiple tools to improve creation efficiency.
3. Brand narrative and user resonance: Applied the "Hero's Journey" narrative theory to design Xiao Lai's emotional arc:
Fear and restraint → Firm resistance → Confidence and blooming
Mastered how to create emotional resonance between the brand story and the target audience's demand for "breaking aesthetic anxiety".
4. Team collaboration and division of labor: Clarified responsibility boundaries (content creation, visual generation, technical implementation) in cross-role tasks, and resolved cross-linkage issues through effective communication to ensure overall project progress.
3. How I Built the Project
We completed the project in 5 key steps following the logic: Brand Analysis → Content Creation → AI Production → Cross-Platform Optimization → Project Completion
3.1 Preliminary Preparation: Brand and Audience Analysis
- Conducted in-depth analysis of:
- Flower Knows' core brand concepts
- Target audiences (15–25-year-old Gen Z females, Cosplay enthusiasts, etc.)
- Communication objectives
Clarified that the short film must convey the core value of "diverse colors + courage".
- Determined the narrative framework: Adopted a "three-act structure" with "Xiao Lai guarding colorful flower seeds" as the main line, reflecting the brand proposition of "breaking a single aesthetic".
3.2 Content Creation: AI-Assisted Script and Storyboard
Text creation:
- Used ChatGPT and Doubao to draft the initial story text.
- Optimized it into a 3-minute short film script via Kimi, clarifying:
- Core characters (Xiao Lai, the Gardener, the Elderly Woman)
- Key plots
- Emotional turning points
Visual design:
- Based on the brand's "dreamy and colorful" style, used Midjourney, LTX, and Photoshop to draw storyboards.
- Defined visual specifications: Character images, scene tone transitions (gray wasteland → colorful flower sea).
3.3 Video Production: AI Tool Collaborative Implementation
Material generation:
- Generated character/scene video clips: Midjourney, Doubao
- Produced dubbing: Eleven Labs (e.g., gentle voice for the Elderly Woman, stern tone for the Gardener)
- Matched sound effects and background music
Post-production integration:
- Audio-visual synchronization: LTX
- Lip-sync effects: Weavy
- Clip editing & rhythm adjustment: Jianying (CapCut)
- Video quality repair: Wink (ensured 1080P clarity)
- Output: Formats suitable for multiple platforms
3.4 Team Collaboration: Clear Division of Labor
Wang Meiyan:
Brand preliminary research → Story script writing → Narration text creation → Character voice generation → Special effects sound/background music production → Shot design → Video image generation → Auxiliary video ## editing
Tao Xingyi:
Brand preliminary research → Story content creation → Narration text writing → Character voice generation → Special effects sound/background music production → Shot design → Video editing → Auxiliary video production
4. The Challenges I Faced & Solutions
4.1 Core Challenge 1: Unstable AI-Generated Character Style
Problem:
The image of Xiao Lai generated by Midjourney had inconsistent styles across scenes (e.g., facial features, costume details), which damaged the short film's coherence.
Solution:
- Established a unified visual specification document (clarified hairstyle, costume colors, scene tones).
- Embedded specification keywords into Midjourney prompts.
- Uniformly adjusted material styles with Photoshop to ensure visual consistency.
4.2 Core Challenge 2: Incoherent Story Plot Transitions
Problem:
Due to initial script time constraints, key plot transitions (e.g., the crowd shifting from "silence" to "united to protect flowers") were abrupt, failing to arouse emotional resonance.
Solution:
- Returned to brand communication objectives to optimize narrative rhythm.
- Added transitional plots: "Xiao Lai persisting in planting flowers despite injury", "Young people secretly 1. helping water plants at night".
- Iterated the script repeatedly via Kimi to strengthen foreshadowing for emotional turning points.
4.3 Core Challenge 3: Technical Implementation Issues (Lip-Sync & Video Quality)
Problem:
- AI-generated dubbing was out of sync with characters' lip movements.
- Some video clips had blurred image quality, affecting viewing experience.
Solution:
1.Optimized lip-sync effects with Weavy and manually adjusted audio-video timelines. 2.Repaired blurred clips using Wink's "HD Quality Restoration" function. 3.Ensured the final output met 1080P clarity standards.
Summary and Outlook
Through the combination of AI tools and cross-media thinking, the project successfully transformed Flower Knows' brand philosophy into perceivable short film content.
Future optimization directions:
- Enrich the short film plot to refine characters and visuals.
- Develop multi-language dubbing versions to adapt to cross-regional communication.
Built With
- chat-gpt
- doubao
- elevenlabs
- english
- jianying
- jimeng
- ltx-studio
- meitu-xiuxiu
- midjouney
- weavy-ai
- wink
Log in or sign up for Devpost to join the conversation.