princessafia Apr 24 2023 at 21:14

An introduction to the world of AI for designers

Easy

5 min

1.5K

Every day a new neural network appears and every day more opportunities are opened to designers to simplify their workflow. Someone fundamentally refuses to use them, because “there is no life in machinex and technologies”, and someone is only happy to find a way to reduce the amount of work. Personally, I belong to the second type and want to share the most detailed gait on neurons I have acquired lately.

Personally, for a long time I did not try to understand the neural networks, considering it a dark forest, the need for which is zero for me. And then from all sides were filled with stories and reasoning about neurons, their use, evaluations of various services, and I came to realize that I missed something and it needs to be repaired urgently.

I go to the internet and I know I don’t understand anything. How Midjourney works, what is this your ChatGPT, how to deal with it, no general instructions. So, after spending a lot of time, I’ve gathered up an entire stockpile of information to make life easier for all those who are now hard-working in this AI-like world.

Neuroscience – what is it?

In essence, a neural network is artificial intelligence, a program that repeats the model of human neural connections. That is, a program that analyzes, processes information and produces a result based on this analysis.

Why are they all so attached to them?

They simplify many processes, both working and household.

Neuroscience does not replace humans. They can help, but you still need to edit and refine what AI has generated.

Neural networks can be used to solve various tasks, such as:

The Generation. Neural networks can create unique designs based on set parameters and requirements: logos, fonts or images that can be used when designing a site.

Neural networks and UX/UI design. A neural network can analyze how users react to various elements of the site interface, how convenient it is for the user to use them. And offer improvements for a more attractive design.

• Optimization of the process. Neural networks can automate certain tasks in the design process, such as image processing or prototyping, allowing designers to focus on more creative tasks.

P.S An example of design tasks.

How to generate a request correctly?

AI can give a variety of tasks, complex and simple, the main thing is to do it correctly. To the best result that will meet your expectations you need to adhere to the following principles:

Write simply. Imagine that you are explaining something to a child. Try to create a query without using complicated words, phrases and expressions.

Write in English. Most neural networks understand English better. Even if there is Russian, sometimes a better result can be obtained if you write in English.
Don’t focus on small details. This can lead to the fact that the neural network will misunderstand and will have to make the request again, which increases the time and may not give the desired result. But do not try to shorten the request as much as possible, you need to stick to the golden middle.
Write without denial. As practice has shown, language models often ignore queries that contain denials, as well as some unions and pretexts.

Top Neural Networks for Designers

ChatGPT - Language model, neuron text to text

A chatbot with artificial intelligence developed by OpenAI and capable of working in dialog mode, supporting queries in natural languages (this is how the Wikipedia bot described it).

In fact, the bot can be used for anything, I wrote an article with it, made a resume, asked to write a prompt, in general, everything you want. He explains books, articles and translations very well. For example, I asked him to talk about the differences between Human Interface Guidelines (HIG) and Material Design respectively:

Want more details? Ask for more details. Want a shorter? will reduce.

Here you can explore the best text commands for chat.

Midjourney - Image Generator in Discord

Artificial intelligence software that creates images based on text descriptions.

Better to study it here:

Article 1: The introductory article.

Lesson 2: Using AI to Create Midjourney Illustrations

Lesson 3: How to scale items separately in MidJourney

Lesson 4: How to Use Referrals in MidJourney

Lesson 5: How to Create Cartoon Style Characters in MidJourney

Lesson 6: How to Use Remix in Midjourney

Lesson 7: How to Create Patterns in MidJourney

Lesson 8: MidJourney Prompt Resources

Lesson 9: How to Create a Comic/Midjourney

to read:

“Quick Start Guide”, official documentation,

The Art of Prompts for Artificial Intelligence, Kevin Ashley

Midjourneys: Images and Their Prompts With Midjourney by Antonis Tsagaris.

DALLE - Image Generator

Unlike MidJourney, here you can see how many credits on the account - by standard they are 50 and exactly a month later they are updated to 15. That is, no matter how many loans you have – they will be 15 and more each month will be too.

There are 3 operations available, each costing 1 loan:

1. Generating 4 images on a single text request

2. Generate 4 variations based on the created image or user-loaded image

3. Generating part of an image (4 variations) on text request (or image extension)

And again here is for you a detailed user guide