Sri Lanka's inaugural deepfake-driven campaign at Magic Mango.
Our mission was to contribute to Keells' brand equity through a captivating sub-campaign under the overarching theme of "Priceless Moments." The goal was ambitious yet heartwarming — to recreate those moments that people couldn't capture in their lives using the power of GenAi And deepfake technology.
The journey began with an invitation to the public to share their untold priceless stories. We received an overwhelming response, with 400 heartfelt narratives flooding in through Facebook. From this treasure trove, we meticulously handpicked eight stories to weave our magic.
The campaign began with targeted ads and organic social media engagement, encouraging our audience to share their untold stories. Narrowing down from 400+ submissions to just eight was a challenging task. Selected applicants were then asked to provide three images per person and additional photos of those featured in their stories.
With a curated collection of character images, we dived into the realm of Generative Generative Ai. Stable diffusion became our canvas for creating human figures, while Roop and Faceswap Lab played key roles in crafting lifelike faces. Though the execution seemed straightforward, the process was marked by extensive trial and error. We seamlessly blended existing community models with our proprietary image model to bring these stories to life.
Our mission was to contribute to Keells' brand equity through a captivating sub-campaign under the overarching theme of "Priceless Moments." The goal was ambitious yet heartwarming — to recreate those moments that people couldn't capture in their lives using the power of GenAi And deepfake technology.
The journey began with an invitation to the public to share their untold priceless stories. We received an overwhelming response, with 400 heartfelt narratives flooding in through Facebook. From this treasure trove, we meticulously handpicked eight stories to weave our magic.
The campaign began with targeted ads and organic social media engagement, encouraging our audience to share their untold stories. Narrowing down from 400+ submissions to just eight was a challenging task. Selected applicants were then asked to provide three images per person and additional photos of those featured in their stories.
With a curated collection of character images, we dived into the realm of Generative Generative Ai. Stable diffusion became our canvas for creating human figures, while Roop and Faceswap Lab played key roles in crafting lifelike faces. Though the execution seemed straightforward, the process was marked by extensive trial and error. We seamlessly blended existing community models with our proprietary image model to bring these stories to life.
