Dianthus uses cutting edge AI technology to grow D2C brands – applying hyper-personalized marketing tactics and new AI-centric workflows and processes, Dianthus is the world’s only AI-first ecommerce company.

Client
Dianthus is the world’s only AI-first ecommerce compan
Industry
eCommerce

THE OPPORTUNITY

“As an AI-first company, Dianthus seeks out fellow deep learning researchers who can offer unique experience and perspective in applying and implementing cutting-edge solutions in AI and ML to partner with our in-house AI teams.”– Dianthus Executive

“Custom AI development for new and innovative products can be a risky endeavor without a disciplined approach to experimentation, broad domain knowledge and experience implementing the latest research and architectures. Neu.ro’s combination of PhD-level AI expertise and an extremely robust foundation in MLOps gave us the confidence that our project would be handled both innovatively and responsibly.”– Dianthus Executive

“Neu.ro went above and beyond in terms of identifying and implementing cutting edge deep learning techniques to create a genuinely innovative solution. We look forward to continuing to advance the state of the art in the area of AI-generated visual marketing with them.”– Dianthus Executive

In the new post-pandemic, work from anywhere world, home fitness is booming. Peloton is a famous growth leader, with paid digital subscribers increasing 176% y/y in their last quarter. Apple recently announced a subscription fitness service providing guided classes via their Apple TV product and incorporating input from their Apple Watch sensors.

Altis aims to go one step further by incorporating AI vision technology using multiple camera sensors to accurately track and analyze movement and provide guided fitness instruction and form coaching for fitness and sports applications.

Challenge
In order to achieve believable AI-generated digital influencer product shots, a unique data pipeline had to be created which incorporated background generation, identity generation, 3d rendering, human positioning, product positioning, harmonization of elements and then various stages of refinement to result in a finished result photo attractive and natural looking enough to share on social media.

A major challenge is that different approaches were required for background generation, human face generation, human pose generation, influencer placement, product placement and camera angle selection.

Finding optimal approaches for each of these elements required an iterative process of experimentation and optimization

The Challenge

Scaling D2C brands requires a deep understanding of a products’ audience, their customer journey and the available channels for reaching them. Smart marketers know to test a wide array of variables in terms of audience segmentation, messaging and creative to support growth. Assets must further be aligned across various communication channels, managed for individual customer groups and tracked over time. Dianthus has been a leader in terms of applying advanced deep learning techniques to these issues. A major bottleneck in this process, however, has been the cumbersome and manual creation of unique visual marketing assets in ecommerce product marketing.

A proposed solution was to create a sophisticated AI computer vision system that would be able to generate unique photographs including computer generated human or animal models, against naturalistic backgrounds and incorporating the D2C product.

Solution
AI-Generated “influencer” product shots for social sharing

The ability to programmatically create compelling visual advertising assets for specific products and audience segments appropriate for social media channels is a holy grail for efficiently scaling brands
Results
Approaches tested during experimental phase:
- CLIP guided optimization
- ESRGAN-based refinement
- CycleGAN
- MUNIT
- Diffusion model reprojection
- StyleGAN3 reprojection
- StyleGAN3 face recovery
- Face swap
- Contrastive Unpaired Translation

The Solution

Neu.ro set to work conducting research of possibly applicable techniques for dynamically composing unique images composed of people products and backgrounds.

Steps included generating the background images, appropriately scaling the person image, and selecting the best place to overlay, as well as various techniques for enhancing the resulting image.

Real image datasets from commercially available sources were combined with synthetically generated images via Synthesis API with different camera views and configurations. Various rendering technologies were tested for quality, stability across platforms and ability to be automated at scale.

Additional research was conducted on reduction of manual input, including automated avatar and wardrobe creation and automated scene selection with product placement.

Furthermore, various personas were created and tested, including male, female, gen z and millennial figures.

The resulting solution incorporates technologies for human positioning, camera positioning, appropriate sizing of foreground and background elements, and product harmonization improvements, as well as numerous optimizations for speeding up the pipeline and limiting required human input into the process.