Why I Sometimes Use AI-Generated Images in My Warhammer 40,000 Lore Videos

From time to time, I receive comments questioning my use of AI-generated images in some of my Warhammer 40,000 lore videos. The two most common criticisms are that using AI is "lazy" or that it somehow takes work away from artists.

I wanted to explain my thought process and how AI actually fits into my content creation process.

https://www.youtube.com/watch?v=4JZRWFp7JWg

Telling the Story Visually

When creating a lore video, I try to ensure that what appears on screen matches what I'm talking about at that moment. If I'm discussing a famous battle, a well-known character, or a major faction, there is often plenty of official artwork, cinematics, and game footage available to help tell that story. For example, a video on the Blood Ravens has plenty of options available in the form of the Dawn of War cinematics.

However, Warhammer 40,000 is an enormous setting. Once you move beyond the most popular topics, the available visual material becomes much more limited.

For example, if I'm discussing an obscure event involving a lesser-known Space Marine Chapter, there may be very little official artwork that accurately represents what I'm describing. While there is often fan art available online, I avoid using it (to my knowledge) or only use these images if I'm discussing and providing evidence of community engagement with the faction. I don't think it's fair to take an independent artist's work and use it in a monetised video simply because it happens to fit the topic and is an easy stand in for a video slide.

Official Games Workshop artwork is different. I'm discussing their setting, characters, and lore, so using official material as part of commentary and analysis falls within what I consider a reasonable and respectful use of source material.

Most Images Still Require Significant Editing

Even when official artwork exists, I often spend hours modifying it to better fit the story I'm telling.

That might involve altering colours, chapter markings, insignia, backgrounds, or other details to represent a particular event or faction. For example, I might edit artwork of an Imperial Fist to better represent a Lamenter or as in the example below, create an image of Lamenters fighting the Orks.

Whether my Photoshop skills are any good is certainly open to debate, but the point is that a significant amount of work goes into creating the visuals you see on screen.

Where AI Fits In

Sometimes there simply isn't any suitable artwork available at all.

To use a specific example, imagine I'm describing a situation where the Lamenters are trying to evacuate millions of civilians from a planet before an Ork invasion. As far as I'm aware, no official artwork exists depicting that exact scenario.

That's where AI can be useful.

However, AI is not producing finished images that I simply drop into a video. Far from it.

The generated images are frequently inaccurate. Chapter symbols are wrong. Armour trim is incorrect. Eye colours don't match. Banners, heraldry, weapons, and countless small details often need fixing. In many cases, the AI image is simply a starting point that still requires substantial editing before it resembles what I actually want to show.

The end result is usually a combination of AI generation, manual editing, Photoshop work, and lore knowledge.

As an example, is there official (or otherwise) artwork depicting the Maelstrom wardens? No, there's not. So I got AI to create a slide for me, but it still needed huge amounts of editing for it to be even close to what I wanted;

Edits like this can take a good few hours, tweaking and refining.

Is It Lazy?

One criticism I often hear is that using AI is lazy.

I understand why some people feel that way, but it doesn't reflect the reality of how I create my videos.

I'm not an illustrator. If I attempted to hand-draw every image required for a typical 15–20 minute lore video, that single video could easily take months or even years to complete. The reality is that these videos often receive only a few thousand views and generate very modest revenue.

For a small creator, that's simply not practical.

To me, AI is a tool. It's no different in principle from a calculator helping an accountant perform calculations or a crane helping a builder lift heavy materials. It doesn't replace the creative vision, research, scripting, editing, narration, or decision-making involved in creating the final product. It simply allows one person to accomplish more than would otherwise be possible and in my case, it helps me as a creative miniature painter be more creative than my immediate skill-set allows for.

As an example, even a 15-20 minute video (below) could represent 40+ hours of editing. Not to mention the research, script writing and replaying video games to get the exact scene I had in mind.

Am I Taking Work Away From Artists?

The second argument is that using AI takes money away from artists.

In my particular situation, I don't believe that's true.

I'm a very small YouTube creator who is only just reaching the point where the channel might become financially worthwhile. I am not choosing AI instead of commissioning artists. The reality is that commissioning artists simply isn't financially possible right now.

So the choice isn't between hiring an artist and using AI.

The choice is between using AI-assisted imagery or not having those images at all. Or using irrelevant images during those scenes or simply relying on Space Marine 2 game play footage, something I think I rely on too much anyway.

If the channel continues to grow and eventually generates enough income to support commissioned artwork, I would genuinely love to work with artists. In fact, one of the most exciting possibilities of a successful channel would be being able to commission custom pieces that perfectly capture the stories I want to tell.

But until that becomes financially realistic, AI helps me get close to that eventuality and bridge a gap that would otherwise be impossible for a small independent creator to fill.

Is it theft?

Another criticism of generative AI is the claim that it is inherently "theft" because it learns from existing images, text, music and videos. The argument is understandable, but it raises an important question: if learning from existing work is theft, where exactly do we draw the line?

When an AI model is asked to generate an image of a Space Marine, it will draw upon patterns learned from countless examples it has encountered during training. Some of those examples may be official artwork, photographs of miniatures, fan art, or other related material. Critics argue that because the AI has learned from these sources, any resulting image is a form of theft.

Yet human artists work in a remarkably similar way. An artist tasked with drawing a Space Marine will almost certainly study existing references. They may look at Games Workshop artwork, painted miniatures, fan illustrations, or screenshots from video games. They will observe proportions, armour details, colour schemes and stylistic conventions before producing their own interpretation. The finished piece is informed by previous work, even if it is not a direct copy of any single image.

In reality, almost all creative work builds upon what came before it. Artists learn by studying other artists. Writers learn by reading books. Musicians learn by listening to music. Even entirely original creations are often inspired by existing ideas, genres and influences. Creativity has always been a process of absorbing, interpreting and recombining existing knowledge.

This does not mean that anything goes. There is a clear distinction between learning from a work and reproducing it. If an AI system or a human artist creates something that is effectively a copy of a specific copyrighted work, then legitimate legal and ethical questions arise. However, learning from existing examples in order to create something new is not a principle unique to AI; it is fundamental to how humans create as well.

The debate becomes even more complicated when intellectual property is involved. Games Workshop's crackdown on independent Warhammer animations before the launch of its Warhammer TV service demonstrated the tension between fan creativity and IP ownership. Many of those animators were producing original works, but they were doing so using characters, designs and settings owned by Games Workshop. Whether one agreed with the company's decision or not, the legal basis was clear: the Space Marine is Games Workshop's intellectual property.

The same principle applies regardless of whether the creator is a person or an AI. The core issue is not simply whether existing material was referenced or learned from. The question is whether the resulting work unlawfully reproduces protected intellectual property. If every act of learning from existing creative works were considered theft, then much of human creativity would have to be judged by the same standard.

The challenge for society is therefore not deciding whether AI learns from existing works—it clearly does. The challenge is determining where learning ends and infringement begins, a question that has existed long before AI and will continue to exist long after the current debate has passed.

Final Thoughts

I completely understand that not everyone will agree with the use of AI-generated imagery. Reasonable people can disagree on the topic but I also don't think the knee jerk reaction "AI all bad" is helpful or even true.

All I can do is be transparent about how I use it.

For me, AI isn't replacing artists. It isn't replacing creativity. It isn't removing work from a commissioning budget that doesn't exist.

It's simply one tool among many that helps a small creator tell stories that would otherwise be impossible to visualise.

The research, writing, editing, narration, and final creative decisions remain entirely human. AI just helps me put those stories on the screen.

As an aside, if there is any of the AI generated slides people want a copy of, I'd me more than happy to index them somewhere if there's the interest.