阿里妈妈丨Lookie「AI试衣黑科技」
Ali Mama – Lookie "AI Try-On Tech" Film Project
DESIGN DESCRIPTION.
ROLE
AIGC DESIGNER & AIGC EIGNEER
DESCRIPTION
-
Integrated digital matte painting with KRITA-AI into the production pipeline to create accurate front and side perspective background images based on pre - shot materials.
-
Developed various image - to - illustration style workflows to ensure compliance with specific requirements, such as maintaining the consistency of character faces while changing their clothes and background, keeping character faces and hairstyles unchanged while altering their clothing and background, and achieving an overall stylistic transformation.
-
Achieved face swapping effects with consistency in face swapping, dynamic movement, and expression control.
YEAR
2024
AIGC TECHNOLOGY
KRITA-AI + Digital Matte Painting
Partial Video Redrawing
Stylistic Consistency
Character Consistency
PLATFORM
PROCESS 1
REQUIREMENT 1 ：
AI-Driven Scenario Construction Framework
WORKFLOW
REQUIREMENT 2 ：
Style-Consistent AIGC Scene Generationork
The framework requires stylistic consistency preservation, element repetition avoidance via contextual substitution
(e.g., replacing swimming pools with ocean elements while retaining fuchsia wall backgrounds)
Accurate perspective projection.
Structural coherence is maintained through core element retention
(e.g., preserving cabinets and magenta neon lighting when substituting furniture with cyberpunk window views).
WORKFLOW
Video Tool for Directional Generation of Viewpoint Offset with Spatial Transformation
Wall Surface Micro-Element Generation and Style Transfer with Localized Re-rendering to Specified Object Scenarios
Artifact Generation Perception Correction through Adversarial Neural Rendering
PROCESS 2
REQUIREMENT 1 ：
AIGC Multi-Style Cross-Style Translation
REQUIREMENT 2 ：
AIGC Multi-Style Cross-Style Translation
Retain the female facial features while altering garments and background elements. Specifically, implement facial mask generation and AI-driven synthesis for clothing and environmental components.
Require preservation of facial and capillary characteristics while modifying apparel and scenery. This necessitates AI-powered facial/capillary semantic segmentation with mask preservation, coupled with neural network-based garment and background synthesis.