LogInPhygital+'s CEO Oleg Yusupov told us about the company's innovative approach to AI which combines neural networks in a user-friendly way, spoke about their products and how they help game developers save content creation costs, and discussed game studios' readiness to implement AI workflows.
Introduction
Hey there! I'm Oleg Yusupov, CEO and co-founder of Phygital+. This product is an extension of Phygitalism, a B2B company that's been helping global brands like Volkswagen, Samsung, and Kaspersky implement cutting-edge technologies such as VR, AR, and AI for seven years. Our team's hard work and innovation have been recognized with over 60 international awards, including an Emmy nomination, Cannes Lions, and Webby awards.
At Phygital+, we have a team of talented researchers, developers, and creative masterminds who are all dedicated to pushing the boundaries of tech even further. We've conducted extensive visual research on machine vision and developed a unique approach to 3D ML that uses neural networks and ML algorithms to work with 3D data. Our mission is to make technology accessible and user-friendly for everyone, and we're excited to continue pushing the envelope in the years to come.
Phygital+
Our projects have always included a hands-on approach to new technologies, and after years of working in the ML and CG market, the team noticed a gap – there were no tools that combined neural networks in a user-friendly way. Existing makeshift solutions required serious machine learning expertise, closing the doors to AI for many creatives.
Our "a-ha!" moment came during a project that required a long DeepFake video based on a single image. Back then there was no DreamBooth, and existing tools could not deliver the desired high-quality production. We set out to create a custom solution, combining several ML approaches to solve this problem.
With success in challenging projects like this, we began developing our own product, Phygital+, that could potentially solve incredibly hard AI tasks yet feel native to creative professionals. Our unique feature is a node-based interface, akin to Blender and Houdini, but AI-powered. The Phygital+ team currently consists of 10 professionals spread across the globe, and together we are building a future workspace for creatives, ourselves included.
Phygital+ AI Workspace & AI Library
Phygital+ offers two complementary solutions: Phygital+ AI workspace and AI Library.
Image: Two complementary products by Phygital+: AI Library (left) and AI Workspace (right). Source: Phygital+
AI Library is a one-stop-shop for all your Generative AI needs: from collabs and public spaces on HuggingFace to ready-to-use solutions such as mobile and web apps. With over 1000 tools and 200 educational materials, smart filters, and semantic search, you can easily find the perfect AI tool for your project. AI Library is our contribution to the community: it is free, updated daily, and grows organically by tens of thousands of users monthly.
Phygital+ AI workspace is our main product that saves gamedev clients up to 80% of their content creation costs. Our clients now create digital assets up to 25 times faster, with neural networks being a part of their workflow. Phygital+ is a node-based tool, where each neural network is just like a LEGO block that could be used separately or assembled into a unique pipeline.
Currently, Phygital+ combines 25 different neural networks, all ready to use for creative professionals and gamedev studios. The most popular are Stable Diffusion with 20+ custom models and styles, DreamBooth training, ControlNet, and Upscale, among many others. Most importantly, gamedev companies can train models on their own style and create custom AI pipelines – saving time and money and bringing endless possibilities to the world of content creation.
Images: Game characters, assets, and locations created in Phygital+. Source: Phygital+
Image: Currently available features in Phygital+. Source: Phygital+
Node-Based System
Image: Generating separate images and combining them in a new node. Source: Phygital+
In order to bring real value, a solution should be native and intuitive to its users. Node-based systems are familiar to CG specialists and allow for flexible, non-destructive experimentation. Implementing the node-based system is a challenge, but we believe it is well worth the effort. With this approach, every neural network becomes a building block, like LEGO, that can be assembled into a unique pipeline.
With an ever-growing AI landscape, it is very difficult to create a one-size-fits-all solution, and nodes provide exactly that. Neural networks thrive in combination with each other – together they can solve a variety of tasks, from creating images to reconstructing dynamic 3D environments.
Implementing AI Workflows in Game Studios
Disrupting an established workflow is surely a challenging task – but not doing so is a costly mistake in the long run, and game studios are well aware. The real question is how to integrate AI into their workflow most effectively and conveniently. Current solutions in the B2C market such as Midjourney and DreamStudio (which our clients usually try at first) don’t consider the needs and production aspects specific to the gamedev industry. We work closely with each studio to ensure that the AI workflow optimizes the pipeline and maximizes the benefits.
One of the biggest obstacles is understanding how a neural network works and how to write efficient prompts (description of your desired output for a neural network). To help novice users, we create tutorials and have recently published the Prompt Guide.
Image: game development pipeline with and without Phygital+. Purple elements indicate areas that can be accelerated with AI. Source: Phygital+
Overcoming the Challenge of Keeping Up with Neural Network Releases
Generally, staying on the edge and offering the best possible solution can be difficult. However, with our AI Library, we successfully monitor the trends and news, so adding a new buzzing neural network becomes a matter of code availability. It takes us one day to add a new AI tool to the workspace, so our users can enjoy the latest tools without worrying about research, installation, storage, or hardware.
Phygital+'s Roadmap
We envision Phygital+ becoming a central hub for content creation, with AI-powered tools and workflows seamlessly integrated into every step of the process. We plan to build a workspace for creators, developers, and tech enthusiasts where they can exchange experience and knowledge and collaborate using the latest tools for content creation. That being said, we want to refine the UI, add more neural networks, and develop our own ML algorithms for working with video and 3D data.
Ultimately, we see Phygital+ as a human-computer interface, a next-generation AI assistant that helps creators unlock their full potential and pushes the boundaries of what's possible in the phygital space.