ECORN Research: Mass AI Image Generation (Part 2)

In the previous article, we start to meet you with our streamlining graphic design workflows through generative AI. Let's continue our journey!

‍

Learning About Generative Model Policies

‍

Ethical Considerations in using generative AI

‍

When working with neural networks, some ethical considerations must be taken into account. For example, when generating images, you must avoid creating content that may be offensive or inappropriate for the user. In turn, various services using neural networks may have their own rules regarding image generation. Thus, when working with neural networks, it is necessary to take into account not only technical aspects but also ethical issues related to content creation.

Disclaimer, since we are already talking about Midjourney here (it was chosen as the most ideal test case), we will only talk about the terms of service, regarding Midjourney.

‍

Terms of Service — Midjourney

‍

Like any company, Midjourney has its terms of use and rights that are given to the user. It is worth considering the most important of them, but anyway, it is impossible to specify all the rules, so we recommend that you familiarize yourself with them: https://docs. midjourney.com/docs/terms-of-service

‍

Commercial Use Rights to Midjourney Images The rights to use images created in Midjourney depend on whether you are a paid subscriber or not.

Midjourney license for free users If you are not a paid subscriber (i.e. using a free trial plan), your usage rights are governed by the international license Creative Commons Attribution-NonCommercial 4.0 (CC BY-NC 4.0).

This means that you can use the images for any non-commercial purposes, as long as you credit Midjourney and link to their website, i.e. credit the neural network as the author.

‍

Note. You can also modify and mix these images as long as you share them under the same license.

Non-Commercial Purposes include many non-revenue use cases.

A few examples of non-profit activities: are personal, educational, research, and artistic activities.

Commercial purposes include the sale or licensing of images or products derived from them. Some popular examples of commercial products include merchandise, print, or NFTs.

Use in advertising or marketing campaigns also falls under commercial use.

Think about things like business logos or branding materials.

‍

Midjourney license for paid membership If you are a paying subscriber, your usage rights are governed by the General Commercial Plans.

This means that you may use your Midjourney graphics for any purpose, including commercial purposes, subject to certain conditions.

‍

Midjourney paid license rules:

You must not claim ownership or authorship of the images or imply any endorsement by Midjourney.
You do not use the assets created in Midjourney in any way that is illegal, defamatory, obscene, harmful, or violates the rights of others.
You are not using the images or the platform in any way that could damage or interfere with Midjourney services, systems, or networks.

Now suppose you have a larger business.

‍

Midjourney says that if you’re a company with over $1,000,000 in gross revenue per year, you must purchase the Pro plan to use the generated images for commercial purposes.

‍

About process automation on Midjourney

‍

Excerpts from the Terms of Service of Midjourney

It is very important to note that Midjourney reserves the right to suspend or deny User access to the Services at any time and for any reason.

Users may not use the Services to conduct competitive research. Users may not reverse engineer Services or Assets.

Users may not use automated tools to access, interact with, or create Assets through the Services.

Only one user can use the Services from one registered account. Each user of the Services can have only one account.

That is, if we move on to practice, then the service itself is prohibited from making images generated automatically or using automatic programs, for this you can get a ban. Therefore, everything described below is taken as an experiment and for educational purposes.

But the future is moving forward, and we think that neural networks will only develop and eventually really come to the fact that this process is done faster, using the same automation, to receive more different generated images.

‍

Disclaimer

‍

We do not guarantee that Midjourney will not change its license policy at the discretion of the company itself, we are not lawyers.

For more information on the constantly updated Midjourney Terms of Service see here: https://docs.midjourney.com/docs/ terms-of-service

‍

Addressing concerns about originality and intellectual property

‍

Midjourney Copyright Overview

In February 2023, the US Copyright Office stated that AI-generated images could not be copyrighted.

This affects images generated by platforms including Midjourney, Dall-E, and Stable Diffusion.

As a user of Midjourney, you will lose any exclusive rights to the works you create on the platform, as well as to the works of others.

For all Midjourney levels below the Pro plan, images are posted to the community website and Discord.

This means that any other Midjourney can use the images you create for their personal or business projects without any repercussions.

‍

About process automation on Midjourney

‍

Excerpts from the Terms of Service of Midjourney

It is very important to note that Midjourney reserves the right to suspend or deny User access to the Services at any time and for any reason.

Users may not use the Services to conduct competitive research. Users may not reverse engineer Services or Assets.

Users may not use automated tools to access, interact with, or create Assets through the Services.

Only one user can use the Services from one registered account. Each user of the Services can have only one account.

‍

Disclaimer

‍

We do not guarantee that Midjourney will not change its license policy at the discretion of the company, we are not lawyers.

For more information on the constantly updated Midjourney Terms of Service see here: https://docs.midjourney.com/docs/ terms-of-service

‍

Ensuring responsible use of AI in graphic design

‍

Creating your images is a very important process that requires maximum responsibility. Midjourney has a policy of not automating its processes, which makes mass production of images impossible. However, we want to draw your attention to the fact that our team is doing research in this area to understand how mass generation works. Our goal is not to encourage the violation of the rights of the service, but to provide users with the opportunity to learn something new and develop their skills in creating their images.

It is important to note that creating your images is a creative process that can be improved and expanded by learning the nuances of creating mass images. Our research can be an excellent opportunity for you to get to know this area and gain valuable experience. Therefore, we strongly recommend that you consider our exploration not only as an opportunity to learn something new but also as a way to develop your skills and become more proficient in creating unique images. Our research is for educational purposes only!

Visual Showcase: Images Generated by the Automated Workflow

‍

In this section, we present you different versions of the generated images with similar prompts that we got using Midjourney, Stable Diffusion, and Dalle-2

‍

Midjourney

‍

Prompt 1 — royal spaniel dog, happy, wearing a royal dress, in a palace, with a crown and a cup of tea

‍

‍

Prompt 1.2 — royal spaniel dog, wearing a fancy dress, with a crown, full face, in a palace, positive atmosphere

‍

Prompt 2 — royal British cat, in a palace, wearing a suit, with a crown

‍

Prompt 2 .1 — Royal British shorthair grey cat, happy, wearing a royal mantle, in a palace, with a crown and a cup of tea

‍

Attempts to test the text — after a constant set of iterations, this is the result:

golden retriever in costume movie poster with text “royal”

‍

movie poster with text “royal club”, gold 3d, a British cat in a costume sitting with a cup of tea

‍

Stable Diffusion

‍

positive prompt: royal spaniel dog, wearing a fancy dress, with a crown, painting, and noble and aristocratic vibe, in a palace

negative prompt: no people

‍

prompt: Royal British shorthair grey cat, happy, wearing a royal mantle, in a palace, with a crown and a cup of tea

‍

Dalle-2

‍

Prompt 1: Royal British humanoid cat, wearing a suit with a crown, digital art, renaissance, 4k

‍

‍

Prompt 2: royal spaniel dog, happy, wearing a royal dress, in a palace, with a crown and a cup of tea

‍

Comparison of images before and after implementing the automated workflow

‍

Let’s now give an example of how the pictures we generated looked before applying the automated workflow and after we did it — using the example of Midjourney.

Before implementing automated workflow

Let’s take one of the prompts that we used at the very beginning when generating images with Midjourney:

royal spaniel dog, happy, wearing a royal dress, in a palace, with a crown and a cup of tea

What do we get?

‍

We have seen that by hand-typing, we get very nice images of a spaniel in royal form, but still, we would like to see more options (let’s say 20 of these spaniels). To do this, we would have to manually enter our prompt 5 times or press V//U to make variations of the picture of the dog we liked or improve the image that we already liked.

As we can see, it takes a long time. What happens if we automate the process?

‍

After implementing an automated workflow

‍

Let’s enter the same prompt, and look again at what we get when we automate the process.

‍

As we can see, using our tool greatly simplifies the image acquisition process. In just 2 minutes, we got 20 high-quality pictures of dogs without much effort. We entered our prompt and simply hit the replay button 5 times, which gave us quite a variety of image options. This allowed us to choose the one that best suits our needs! In addition, while the images were automatically generated by themselves — our team managed to work out other design things :) For example, writing and searching for information for this chapter of our article was created for educational purposes!

‍

Demonstrating the creative potential of generative AI in graphic

‍

Generative artificial intelligence has huge potential for the future, especially in graphic design. We observe how fast the work of neural networks is developing. If just a few months ago we could not even imagine that images could be generated with a single request, now and in the future the creative potential of neural networks will only develop and move forward! We are confident that in the future, many companies will find it easier to generate their ideas thanks to artificial intelligence, which simplifies many design processes and helps designers to apply AI-generated images well.

In addition, generative AI can be used to create unique and high-quality design elements such as logos, illustrations, and backgrounds. This not only saves time but also allows designers to focus on more complex and creative tasks. We look forward to seeing how generative AI will be used to create effective and original design solutions in the future. And we are also sure that soon everyone will be able to apply it to their work!

‍

Conclusion

‍

Artificial intelligence (AI) will be a useful assistant for agencies in the future, ordering or providing design services. AI-based technologies do not threaten the work of designers, but help them solve routine tasks and focus on more creative aspects. AI also keeps projects up to date with the latest trends, does the hard work, streamlines workflow, and helps increase user engagement and conversions.

‍

Recap of the research process and findings

‍

Returning to a summary of the study process and its results, we would like to reiterate that this study was created for educational purposes only, as we are against violations of the terms of use of Midjourney. Our team conducted a study to show how the method we found, can simplify the familiarity and use of neural networks by automating the work with image generation. We found out that there is a great way that can be done with Midjourney, Google extension Effortless Midjourney and process automation software Integrately. With the help of this combination, we got what we announced! We’ve found a way to mass-generate images! In addition, we also examined other neural networks in detail and talked with the community and users who can be considered some kind of professionals in using AI, since they use them very actively. A lot of cool and interesting insights were found!

During our research, we found that this feature is not as widespread as we thought, and some services (for example, Midjourney) do not even allow you to use their artificial intelligence for bulk image generation and process automation. We also found that most of the services that provide this feature require payment to use it.

However, we believe that in the future, developers will understand how important this feature is and how it can greatly simplify the work with neural networks. Our research confirms that the generation of images using neural networks is a very useful feature that requires further development and implementation.

We hope that the results of our research will help developers better understand how they can use this feature to create more efficient and user-friendly applications and services.

‍

Interestingly, our team also discovered that there are a small number of services that provide this feature for free (for example, the service with Hugging Face and code from GitHub):

However, such services usually have limited capabilities and cannot compete with paid counterparts.

‍

Overall, our research has shown that generating images using neural networks is a very useful feature that can make it easier to work with large amounts of data, as well as help create more efficient and convenient applications and services. We hope that our results will be useful for anyone interested in application development and the use of neural networks.

‍

As a bonus, I can also show you another interesting find of mine — the Runway service! Yes, it does not generate perfect images like the same Midjourney, rather it is more similar in image quality to Stable Diffusion and Dalle-2, but it also can generate images in bulk! Just click on the Batch button and you can generate up to 1000 images for an additional fee! If you are considering using the images for commercial purposes, I highly recommend that you read their terms of use.

‍

In addition, this service can generate video! This is very cool, and in the future, this feature will only improve! So before we got used to the fact that you can generate photos, now you can generate videos! We also leave a link: https://research.runwayml.com/gen2

‍

The Future of AI-driven graphic design in Agencies

‍

The future of AI-powered graphic design is at hand. We are sure that shortly, AI will become a useful assistant for agencies that regularly order design services. AI-based technologies do not pose a threat to the work of designers. On the contrary, AI will help them solve routine, monotonous, and template tasks. It is important to note the term “generative design”, which allows the program to create options by the given parameters, and the designer only has to choose the most suitable option.

‍

Generative AI has the potential to revolutionize the creative industry by providing artists, musicians, and writers with new and innovative ways to create unique and original works of art, music, and literature. Artists can use generative AI to create new and unique digital artworks by training a neural network on a large set of image data and then using the network to create new images. Similarly, musicians can use generative AI to create new and unique musical compositions by training a neural network on a large set of musical data and then using the network to create new musical pieces. Writers can also use generative AI to create new and unique literary works by training a neural network on a large set of text data and then using the network to create new text pieces.

‍

Some argue that the use of generative AI in creative industries could lead to job losses for people. For example, if generative AI can create unique pieces of music or art, there may be less demand for human artists or musicians. This could lead to job losses and a change in the way the creative industries operate.

‍

Another problem is that generative AI can lead to the homogenization of the creative output. If a large number of artists, musicians, and writers use the same generative AI algorithms to create their work, then the output may be lacking variety. This can result in all works of art, music, or literature created using generative AI looking and sounding the same.

‍

Despite these concerns, generative AI remains a valuable tool for the creative industry. It can help artists, musicians, and writers overcome creative barriers and find new inspiration. It can also be used to create exciting new works of art, music, and literature that would not be possible without this technology. The job of a designer will only change, so don’t worry about losing your job. On the contrary, the sphere will only develop and the designer can transform into an art director who will observe the work of AI and set the concept for his future creations. In addition, the work of a designer can also become the work of a prompt engineer, who will have to be able to select the right prompts to generate exactly what the customer needs.

‍

In conclusion, we want to say that the use of generative AI in the creative industry is a matter of debate. While there are concerns about job losses and a lack of diversity in creativity, generative AI also has the potential to change how we create and consume art, music, and literature. As the technology continues to evolve, it will be interesting to see how it is used and what impact it has on the creative industry as a whole.

About us

‍

We hope you found our article useful, like and share it with those who may find it useful. If you want to read more about this topic you can read the full version of the article. And of course, stay tuned for the latest in AI, we follow the Creators’ AI mailing list, which comes out every week in https://thecreatorsai.com/

The article was prepared by: Ecorn Agency

We pride ourselves on our ability to create beautiful and effective web designs for Shopify. Our agency builds projects of any complexity, using our technical experience and creativity. Our team consists of experts with many years of experience in e-commerce, design, marketing, and development. We offer a full range of services to create a unique user experience, including design, development, optimization, and support. We not only create beautiful websites but also make sure they are efficient and functional for our clients and help your retail. Ready to work with you to create something amazing and unique. Let’s start working on your idea today!