The Technology Behind the Viral Prisma Photo App

The new Prisma mobile app that transforms your photos into a work of art has been downloaded nearly 17 million times since it was released last month on iOS…

Brad Nemire
1 min readintermediate
--
View Original

Overview

The Prisma mobile app, which transforms photos into artistic styles, has gained immense popularity with nearly 17 million downloads in its first month. Utilizing deep learning technology based on research by Leon A. Gatys, Prisma applies artistic styles using NVIDIA GPUs and the cuDNN-accelerated Torch framework.

What You'll Learn

1

How to utilize deep learning for image processing applications

2

Why leveraging GPU acceleration is crucial for performance in mobile apps

3

When to apply artistic filters versus deep learning techniques in photo editing

Prerequisites & Requirements

  • Understanding of deep learning concepts
  • Familiarity with NVIDIA GPUs and cuDNN framework(optional)

Key Questions Answered

How does Prisma use deep learning to transform photos?
Prisma employs deep learning to analyze an image's visual characteristics and then applies artistic styles from famous artists like Van Gogh and Picasso. This process is not just a simple filter application but involves sophisticated algorithms that reinterpret images based on learned artistic features.
What technology underpins the Prisma app's functionality?
The technology behind Prisma is based on research by Leon A. Gatys titled 'A Neural Algorithm of Artistic Style'. Prisma has further developed this technology using NVIDIA GPUs and the cuDNN-accelerated Torch framework for training and inference.
What are the download statistics for the Prisma app?
Since its release, the Prisma app has been downloaded nearly 17 million times on iOS devices and is now surpassing 2 million downloads per day after its launch on Android. This rapid growth highlights the app's viral popularity.
What future developments are planned for the Prisma app?
Prisma is working on a new tool that will allow users to apply similar artistic effects to video clips, which is expected to be rolled out in the coming months. This expansion indicates the app's commitment to enhancing user creativity.

Key Statistics & Figures

Total downloads on iOS
17 million
Downloads since the app's release last month
Daily downloads after Android launch
2 million
Daily downloads following the app's availability on Android

Technologies & Tools

Hardware
Nvidia Gpus
Used for training deep learning models and inference processes
Software
Cudnn
Accelerated the Torch framework for deep learning tasks
Software
Torch
Framework used for training deep learning models

Key Actionable Insights

1
Explore the integration of deep learning in mobile applications to enhance user experience.
As seen with Prisma, leveraging deep learning can significantly improve the functionality of mobile apps, making them more engaging and innovative.
2
Consider the benefits of GPU acceleration in processing-intensive applications.
Using NVIDIA GPUs for deep learning tasks can drastically reduce processing time, allowing for real-time image transformations that enhance user satisfaction.
3
Stay updated on advancements in AI/ML technologies for creative applications.
The rapid evolution of AI/ML tools can provide new opportunities for developers to create unique and compelling applications, similar to what Prisma has achieved.