
Learn for free: “Neural networks. Practical Course"
Learn moreCreation Idea
In May 2024, the project where I worked as a marketer ended. In recent months, I have been working diligently, devoting almost all my time to work. This was a unique experience for me, since I had previously worked in the same place for six years and had no interview skills. Now I have to explore new horizons and find a suitable job in marketing.
In recent days, I have been facing serious stress and decided to distract myself by starting to play the popular game Water Sort. The essence of the game is to pour liquids of different colors between flasks, avoiding the shades mixing. In a week, I completed more than 500 levels, while simultaneously searching for a job. This game not only helps me relax but also develops logical thinking and attentiveness.
While searching for my calling, I realized that I didn't want to dedicate myself entirely to work. I started living life to the fullest: I play, read books, enjoy cakes every day, and live in the present moment. I even decided to become a certified diver. For the first time since school, my summers are filled with carefree joy again.
I've now completed over 1,500 levels in Water Sort, and the lengthy pouring animations have begun to tire me. This inspired me to create a game with similar mechanics, but without the drawn-out animations. Since I don't have programming or drawing skills, I decided to test whether neural networks could replace these skills. This is how the idea for Lo-fi Sort was born, and it took me a little over two weeks to bring it to fruition. This game offers a unique experience, allowing players to enjoy the exciting process of sorting liquids without unnecessary delays.

Code
I hadn't done any programming for almost five years, so I entrusted this task to the Claude chatbot. Claude has a Projects mode, which allows you to create workspaces with unique databases. For example, you can upload your project code to Claude so that the neural network can analyze this code when generating responses. This significantly simplifies the process of working with code and increases its efficiency.
The workflow with Claude is a consistent and organized process that ensures effective interaction. First, users formulate their requests, taking into account the specifics of the task. Then Claude analyzes the information provided and generates a response based on a deep understanding of the context. Importantly, during the interaction, users can clarify their requests and provide feedback, which helps improve the results. The final stage involves evaluating the received response and possibly making changes to achieve the best result. This approach ensures maximum productivity and user satisfaction.
- Think about the result you want to get from the neural network and describe your wishes in detail.
- Copy the code and check if it works as you intended.
- If everything works correctly, then add the code to the Claude knowledge base and continue generating code.

I chose Unity as my game engine because Claude generated code efficiently for this platform. Unity offers a wide range of game development capabilities, including a powerful toolkit and support for multiple platforms. Using this engine allows you to create high-quality game projects with attractive graphics and interactive gameplay.

Read also:
7 neural networks for programmers that will help you write code faster and better
Modern technologies are actively Artificial intelligence is being implemented in various fields, including programming. Neural networks are becoming indispensable assistants for developers, helping to speed up the coding process and improve its quality. In this article, we will look at seven neural networks that will help programmers increase their productivity and code quality.
The first neural network worth paying attention to is GitHub Copilot. This tool, built on OpenAI Codex, offers intelligent suggestions and code autocompletion in various programming languages. It significantly saves developers time, allowing them to focus on more complex tasks.
The second neural network is TabNine. It uses machine learning to predict the next piece of code based on the context of your project. TabNine supports many languages and integrates with popular code editors, making it a convenient tool for programmers.
The third neural network is DeepCode. It analyzes code for errors and vulnerabilities, offering recommendations for improvement. DeepCode helps developers not only find problems but also learn from their mistakes, which contributes to the growth of professional skills.
The fourth neural network is Kite. This tool provides intelligent hints and documentation in real time, which speeds up the development process. Kite supports many languages and integrates with various editors, making it a universal assistant for programmers.
The fifth neural network is Codeium. It offers developers automatic code hints and advanced auto-completion features, which helps reduce software writing time and improve its structure.
The sixth neural network is Sourcery. This tool analyzes code and suggests improvements based on programming best practices. Sourcery helps developers write cleaner and more efficient code, which is especially important in large projects.
The seventh neural network is OpenAI Codex. This is a powerful tool capable of generating code based on text descriptions. OpenAI Codex can significantly speed up the development process, allowing programmers to quickly implement ideas.
The use of neural networks in programming is becoming increasingly popular, and these tools can significantly improve the efficiency of developers. Integrating neural networks into the coding process not only reduces time but also improves the quality of the final product.
Currently, I use Cursor, a development environment with integrated neural networks, for developing various projects. Cursor's interface is similar to Visual Studio Code, which eases the transition for those already familiar with this environment. One of the key advantages is that code generation takes into account the context of the entire project. This significantly simplifies the workflow, as it eliminates the need for constant copying and pasting of code. The neural network can make changes directly to the project, rather than just generating solutions in the form of text responses. Using Cursor increases development efficiency and saves time, making it an excellent choice for modern developers.
Graphics and Animation
Creating character graphics and animation is one of the most challenging stages in digital art. In this process, neural networks can add unimportant details and remove important elements. To address this issue, I developed my own algorithm that helps optimize the generation process. In the future, I will share its main principles and approaches.

To create a 3D character, you first need to develop an image, which will then be fed to a neural network to generate a 3D model. The key is the character's pose in the image; it is recommended to use poses resembling the letters "A" or "T". This will significantly simplify the animation process and make it more efficient. The correct pose at the image creation stage will allow you to achieve higher quality and more realistic animation later.
I created characters for my game using Midjourney, using the following prompt:
A 3D model of [character description] in the [style] style, with open arms and prepared for animation in a T pose, is presented in a frontal view against a simple blue background. This model is ideal for use in a variety of projects, including games and animation.
Midjourney provides four image options from which you can choose the most suitable one. After selecting one, it is recommended to upscale the selected image and, if necessary, edit certain areas using the Vary (Region) function. This tool allows you to, for example, change the position of the character's hands, which helps create a more precise and expressive image.

Reworked text:
Discover new horizons with our recommendations. We offer you unique materials and useful tips that will help you deepen your knowledge and improve your skills. Our content is tailored to your interests and needs, making it as relevant and informative as possible. Don't miss the opportunity to learn about the latest ideas and trends in your field. Dive into a world of new discoveries and stay one step ahead.
Read also:
Top 10 Neural Networks for Image Creation and Editing
Modern AI technologies have significantly simplified the process of creating and editing images. Neural networks are becoming an essential tool for designers, artists, and marketers. This review presents the ten best neural networks that will help you create unique visual materials and edit photos.
The first neural network on our list is DALL-E, developed by OpenAI. It allows you to create images from text descriptions, making it ideal for generating original images. Second is Midjourney, which also focuses on image generation and offers users a variety of settings to achieve the desired result.
Third place belongs to Stable Diffusion, known for its ability to generate high-quality images from text cues. The fourth neural network is Artbreeder, which allows you to blend images and create new variations based on existing ones.
Fifth on the list is Runway ML, a platform that offers a wide range of video and image editing tools using neural network technologies. Sixth is DeepArt, which transforms your photos into stylized works of art.
In seventh place is Deep Dream, known for its unique approach to generating dream-like images. Eighth place is occupied by Let's Enhance, a service that improves image quality and increases their resolution without losing detail.
The ninth neural network is Remove.bg, a specialized tool for removing backgrounds from images, making it indispensable for designers. Pixlr, a cloud-based image editing platform offering a variety of tools and filters, rounds out our list.
Each of these neural networks has its own unique features and can be useful in different areas. Use them to enhance your visual projects and open up new creative possibilities.
To convert images into 3D models, I used the HyperHuman service. The process is extremely simple: upload the generated image, and you will receive a finished 3D model in return. The service offers a choice of level of detail, ranging from 3,000 to 20,000 polygons. This allows you to adapt the model to different needs and requirements, providing the optimal combination of quality and performance. HyperHuman is a convenient tool for creating high-quality 3D models from your images.
For creating simple objects, using concepts is not necessary. HyperHuman has a built-in neural network capable of generating 3D models from scratch. This is especially useful for developing environment objects, furniture, and props. This makes the model creation process faster and more efficient, allowing you to focus on other aspects of the project.

The service has its drawbacks. For example, the neural network doesn't always successfully generate UV mapping. A HyperHuman subscription costs $30 per month. For this amount, you can generate 30 models, and each model can be edited for free up to 20 times.

Read also:
Prompts for neural networks: how to effectively formulate queries to ChatGPT and other neural networks
Properly formulating queries to neural networks like ChatGPT is key to receiving high-quality and relevant answers. To achieve the best results, it's important to consider several aspects when creating prompts.
First, use clear and specific language. The more precisely you formulate your request, the more satisfying the answer you'll receive. Avoid vague phrases and try to be specific in your questions.
Second, provide context. Neural networks understand your requests better if they are accompanied by the necessary information. Clarify exactly what information you expect and what format you need it in.
It's also helpful to experiment with different wording. Try different approaches and query styles to see which ones produce the best results. This will help you better tailor your prompts to the specific workings of a particular neural network.
Don't forget to test and refine. After receiving a response, analyze it for completeness and relevance. If you're not satisfied with the result, adjust your request and try again.
These recommendations will help you create more effective prompts for working with neural networks, which in turn will improve the quality of the information obtained and simplify interaction with tools like ChatGPT.
I animated 3D models using the Mootion service, using it through Discord. This service has an official server where, to generate animations, you need to send the /motion command with a description in one of the chats called Creation. Then, select a suitable character from the list. The neural network will then create an FBX file and send it to the chat. The resulting animation can be integrated with the character's 3D model in Blender. My YouTube channel features a detailed tutorial where I demonstrate the process of creating an animated character step by step.

Music
Suno offers an excellent solution for generating soundtracks. This tool lets you create both instrumental compositions and full-length songs. To start generating, simply select the desired genre and enter your search query. The free plan provides all the necessary options for experimenting and finding the perfect track. Suno is suitable for both professional musicians and beginners who want to create a unique musical atmosphere.

Read also:
Suno AI neural network: creating musical hits in minutes
Suno AI is a powerful A tool that allows you to create music hits in just a few clicks. Using advanced algorithms and artificial intelligence, this neural network analyzes musical trends and suggests unique compositions. Users can easily generate melodies, lyrics, and arrangements, making the music creation process accessible to everyone. With Suno AI, you can easily create high-quality tracks that will attract listeners and help you stand out in the music world. Try Suno AI today and discover new horizons in musical creativity.
Marketing
On social media, you've probably come across unusual commercials where the main character finds himself in a difficult situation. Most often, to help him get out, he is offered to click a link and download a mobile game. These promotional videos can also be created using neural network technologies. Generating these videos using artificial intelligence opens up new horizons for advertisers, allowing them to create unique content that attracts user attention and increases interest in the product.
First, let's create several images for the advertising campaign in Midjourney. It's important to describe the characters and their actions in the images in detail in the prompt. Two to four high-quality frames are sufficient for effective promotion.
To create a short video, you need to animate the image using the Luma AI service and increase its resolution using Video Quality Enhancer on the Vmake AI platform. This will result in high-quality and engaging video content.
Background music is required for creating the video. It can be easily generated using the Suno service. After that, all that remains is to combine all the elements and add sound effects to enhance the overall experience of the video.

Find out more:
Top 8 Neural Networks for Music Creation
With the development of artificial intelligence technologies, neural networks are becoming increasingly popular in the music industry. They are capable of generating melodies, arrangements, and even lyrics. In this review, we present the eight best neural networks for creating music that can help both professional and amateur musicians.
The first neural network worth mentioning is OpenAI Jukedeck. It allows users to create unique music tracks based on specified parameters. The second is Amper Music, which offers an intuitive interface for generating music perfect for videos and other media formats.
The third neural network is AIVA (Artificial Intelligence Virtual Artist). It is designed for composing music in different genres and can be used to create original works. The fourth is DADABOTS, which is capable of generating endless streams of music, making it ideal for creating background music.
The fifth neural network is Soundraw, which helps create music tracks tailored to the specific needs of users. Sixth is Ecrett Music, which provides the ability to choose mood and genre, which simplifies the music creation process.
Seventh is Google Magenta, a project aimed at using machine learning to generate music and art. And finally, the eighth neural network is Alysia, which not only generates music but also helps in writing lyrics.
These neural networks open new horizons in musical creativity, allowing you to create original works with minimal time and effort.
This video will take approximately two hours to create. This content can be effectively used as part of your advertising campaign.
Copyright
I released the game Lo-fi Sort on the App Store. The entire game, including music, 3D models, and code, was completely generated. However, Apple delayed the app's release due to concerns that the music and graphics might infringe third-party copyrights. This underscores the importance of respecting copyright in game development, even when content is created using modern technology.
This problem can be resolved using a neural network. Simply send a text message from Apple to ChatGPT and request recommendations for action to help you resolve the issue. The neural network will provide helpful tips and steps for further action, significantly simplifying the resolution process.
In my case, I needed to provide Apple with links to the license agreements for the services I used, as well as proof of active subscriptions to them. Many services only transfer rights to generated content when you subscribe. This is important to consider in order to comply with copyright and ensure the legality of content use.

Read also:
Neural networks have become an important tool in the creation of game art, 3D models and animations. This review presents 20 neural networks that significantly simplify the development process and enable artists and developers to create high-quality visuals. These tools use modern machine learning algorithms to generate unique and engaging images, models, and animations. Using neural networks in game design not only speeds up the process but also opens up new creative possibilities, enabling the creation of content that was previously impossible or required significant time and resources. Let's take a closer look at each of the neural networks and their key features that can be useful for game developers and artists.
Bottom Line
- A simple game can be created entirely using neural networks. There are services for generating code, 3D graphics, animations, and music.
- This method is excellent for prototyping. If previously a test version of a project required several months of development and the efforts of several teams, then with the help of neural networks it can be completed in a couple of weeks.
- App stores may have questions about copyright compliance, but if you provide links to license agreements, the administration will approve the release of the game.

