Big Data Processing

Big Data infrastructure is a challenging foundation. It calls for automatic and advanced workflow, scalable storage index and distributed approach to retrieve results near real-time. We handle billions of creative texts, images and videos every day, and mine valuable information for our customers. We are working on the vertical issues, including cloud storage and computing architecture, popular parallel processing framework, major applications and optimization of MapReduce.

Storyline-based 3D-Video Creation

Too boring and time-consuming editing video frame by frame? Try our first "event-based storyline editing tool" in the world wide. This is quite a easy and unique way to express your idea for even a complicated story - just several drag-and-drop and it’s done!

3D-Avatar Cloud Modeling

Imagine that there is a vividly digital version of you, which can chat with others on social network, play game or even do anything you can come up with. Just a few photos of you needed with our advanced computer vision and artificial intelligence technology, you can get your own 3D virtual avatar!

2D Auto Rigging

Rig 2D Characters from Images, The 2D Bone Editor has all the tools to structure sophisticated bone rigs. Users could use pins to constrain areas to selected bones, and optimize subdivision topology for smoother bending effects. Custom Templates for Free Bone Actors, Now you can create new bone structures for unique creature species, and allow them to share the same structured motion assets.

video2animation algorithms

How do you create amazing animations from videos? After we have built the framework of bone, weighting and IK restraint, and then extracted character’s motion from video, our algorithms could reconstruct 2d-3d mapping and generate new animations fitting the motion trajectory.

Voice Style Transfer

What if you could imitate a famous celebrity's voice or sing like a famous singer? We present deep-voice model with a goal to convert someone's voice to a specific target voice. So called, it's voice style transfer. We implemented a recurrent neural network (RNN) to achieve that. Unlike traditional models for speech synthesis, Deep-voice Model learns to produce audio directly from your voice.

Audio Avatar Modeling

With only big corpus data provided, our voice style transfer technology can accurately mimic the voice of anyone. Imagine that, your home robot is talking with mother’s voice, isn’t that baby’s favorite? As popular as live video is concerned, if you can switch your voice freely between a funny star and a cartoon character, will your videos go viral online? There are also chances that you may become an virtual influencer and have a huge number of followers.

Virtual & Real Synthesis

We have used planar tracking technology. This is used to drive tracking, ROTO &masking and object removal. And we have applied 3D camera solver.The fast and easy-to-use solution is ideal for set extensions, 3D text and particle tracking. Additionally the 3D solver can be used to assist other 3D tracking applications on difficult shots with low detail or significant foreground occlusions.

User-Sponsored Distributed Crawler

We are building a distributed crawler system by using residential proxy network. This system could allow users to voluntarily offer their own computing and bandwidth resources towards crawling web pages. We have use such sponsored framework to do overall data analysis and monitoring.

Audio FX Mixer

With assistant of artificial intelligence technology, the Audio FX Mixer allows you to fuel your creativity. Those terms, such as spectral, formant, granular and so on, can be totally ignored. Guitar, cello, piano and other instruments, Audio FX Mixer can easily mix one sound with another. The database with massive sound effect is your source of inspiration that will never end, with which let you easily shape any sound.

Multi-facts Predictor

The multi-factor strategy for quantifying transactions can be described as a long-established classic old-fashioned routine, and it is still the most frequently used and most skilled strategy. The biggest feature Recurrent Neural Network (RNN) is that they hook up with time, which is different from feedforward neural networks, says the results of the next time are not only affected by the input of the next time, but also influenced by the output of the previous time, and further to say that information actually has a lasting impact. With ability to RNN, we use Long short-term memories (LSTM) structure to obtain good accuracy and significant levels before parameters are optimized.

Animation Live Capture

In order to make virtual character(2D or 3D) move or do some action, you should have the animation resource that you want, or you have to make it frame by frame. With our animation live capture technology, you can either use low-price mocap device or even use just a web camera on your computer or in your iPhone to create any animation fast!

GPU-based Virtual Rendering Cloud

The traditional way of rendering video requires a grid of really powerful workstation-level machines with heavy software installed. Now by using our GPU-based cloud infrastructure, anything you need to do is just a web browser and click that all. Create your own storyline, click generating button and all the rendering work is counting on us!

Cloud-based Game Maker

Do you have a wonderful and innovative idea about making some interesting game which will attract lots of people to engage and have fun? Don’t worry about those tough parts like preparing resource, developing and publishing. With our cloud-based game maker, the only thing that we need is your idea!