How to use neural networks for web studios and whether designers will die out. Free neural network online and how to use Midjourney

More recently, Internet protests and almost funerals of artists were organized in connection with the advent of the Midjourney neural network. Now there are quite a lot of AIs that generate images. Sber and Yandex launched their neural networks.

How will this affect web designers and is it time for them to look for a new profession? Or should we approach it from the other side, harness artificial intelligence and use it for our own purposes? Let’s figure it out!

The development path of a web designer :) — The development path of a web designer 🙂

Hello everyone, I’m Sasha Kombarov, I run a web studio. Today I’ll tell you how we use artificial intelligence in the work of designers and whether designers should stress about this 🙂

In the material we will consider two methods of use – free and paid, for those who can pay for foreign services. Let’s start with the paid option – Midjorney.

How to use the Midjourney neural network

Enough manuals have already been written: how to register, how to set the settings correctly, and how to form a request to a neural network. This is not what the material is about, so I’ll tell you how to get started with Midjourney, and you can find all the other gadgets on the Internet.

How to access the neural network:

Install Discord, then launch the application or open the web version of the service in a browser. I use the application, it feels faster than the web version.
Register or log in using your username and password if you have previously used Discord. To register you need a phone number and email. The service operates in Russia, so save your dancing with tambourines, or rather, save your foreign card for payment 🙂
As soon as you log into Discord, you can go to Midjourney using a special invitation link.
Afterwards you need to pay the tariff that suits you. Payment is via stripe and you must use only a foreign card.

Clicking on the link will take you to a public chat where dozens of Midjourney messages will be displayed based on user requests.

How to set up Midjourney to use at work

There is a Discord channel called Newcomer Rooms with newbies chats. This is where the picture is created. Select any number of chats, click and enter /settings. Then Enter.

Select the following settings:

First line. Engine version. Select MJ version; MJ Test has two experimental images instead of four; MJ Test Photo – with a focus on photorealism.
Second line. Image quality. Select Basic quality to reduce latency.
Third line. Configuration level. Set to Medium, Style med, to prevent the neural network from getting too fancy. High or too high results may be far from the requirements.
Fourth line. Algorithm for improving image quality.Select Regular upscale – standard quality increase; Light upscale – increases image resolution, but with less detail; Beta – Can reach twice the size of Regular, but is unstable.
Fifth line. Access mode. Install Public mode, since private mode is paid.
Fifth line. Editing mode. Again, only one is available for free, Fast mode – instantaneous, with a production counter. Relax mode without a counter is available for a fee. And immediately turn on the Remix mode so that you can combine pictures later.

Then type /imagine, press Enter, enter any word in the promt value and confirm sending. A window with the rules will appear, click “Accept”.

You are awesome! Now you are ready to get started.

How to use free neural networks: Masterpiece, Fusion Brain and others

Everything is much simpler here, but there are fewer opportunities. Follow the link, log in if necessary, and make a request.

For example, Fusion Brain generates images according to the description, draws on the images, and you can also use the eraser to change some of the images. There are many interesting styles, I liked the “Soviet cartoon” style. No SMS or registration.

How we use it in our web studio

Searching for references, drawing up a moodboard, brainstorming, developing a concept – all this is well known to those who develop websites, applications, services and the like. Especially designers. Endless scrolling of dribbles, bihances and other advances in search of cool references.

What if you make a request based on your vision, and the neural network gives you options? Not bad, we thought 🙂

And they started feeding our requests to AI. For now we use it on internal projects and for content marketing, we approach this process carefully, but if it finally proves itself, we will implement it on client projects.

For obvious reasons, I cannot demonstrate the services that are currently in development, but I will show the basic principle.

1. Formation of blocks upon request

You send a request to the neural network, and it creates block options for this request. For example, try sending “promo block for the site of a company selling trout” and post in the comments what you came up with.

The user sent a request to create a banner for the law office landing page

Sometimes it turns out to be wild game, but overall it shows good options 🙂

Or this is what came up in response to a request from a site with an Egyptian theme

Life hack: add the keywords “figma landing page design” and the result will be cooler!

If the proposed options do not suit you, then you can send the same request again and you will receive completely different pictures 🙂

2. Formation of options according to the example

Sometimes we ask for options based on a specific example. That is, we send a link to images of the original version that we like, and write to the neural network what we need in this style. For example, you can drop a link to a screenshot of the Apple website into the chat and request that a website for Nizhny Novgorod felt boots be made in this style 🙂

The user uploaded the first image as an example, asked a request, and received options in this style for his request

More examples of resulting options for similar queries:

3. Formation of alternative concepts

Have you created a concept or some kind of block, you like it, but would it be interesting to see it in a slightly different design? Please! Upload the concept with a link, create a request and Midjourney will give you alternative options.

Request. Add csv awards and count the reward in your pocket 🙂

The original version suited me better. But getting alternative options in a matter of seconds would have seemed fantastic a few years ago.

4. Creating Images for Content Marketing

We prepare images for marketing, but are often limited to a standard set of icons and elements. And this is not enough information to illustrate.

In order to create unique images, they began to use AI. We enter a request, select the most optimal option, and then process it.

For example, in Masterpiece they introduced promt: “cartoon caveman smiling, full-length, with a palette and brush in hand, 3D model, octane, realistic, 4k, high resolution, high detail, bright lighting, uniform background.” We got a cool caveman for our article about Lottie animations, and then added this Neanderthal to the card for the Telegram channel.

Results

If you haven’t figured it out yet, the answer is that web designers will not die out. Even if the neural network is upgraded to an unknown level, who will recolor the buttons? 🙂

As we can see, a neural network is more of an assistant to a designer than a replacement for him. Even simple promotional sites are a stretch. There is no talk yet about services, CRM, ERP, etc. Although I would like to see how the neural network copes with the level project Tarkett.

I recommend that everyone who is looking for an idea use a neural network: designers, managers, customers. Not only for workers, but also for creative tasks.

Do you use neural networks in processes? Perhaps it has already been fully implemented into client projects somewhere? Share in the comments!

Good luck and thanks for your attention 🙂