Stable Diffusion Prompts for Realistic Photos of Video Game Characters
Photo of rare Pokémon by National Geographic wildlife photographer
AI Generated Images
In ancient times artists created images by drawing or painting. Then the analog camera was invented and you could create a picture by pressing a button. Later the digital camera was invented. An image could be saved as an array of tiny dots called pixels.
Now you don't need to draw, paint or take a picture with a camera. Enter a text prompt, click the Generate Image button and the Stable Diffusion AI can generate a digital image. Create new pictures based on a text description.
Recently I started making digital watercolor paintings. Come up with a good art prompt and you get an amazing drawing or painting. Text to AI image generators are mainly used to generate art but you can create images that look like a photograph.
The pictures on this page are AI generated images based on text prompts I entered while I was using Stable Diffusion. They look like they could be screen grabs from a movie with CGI characters.
Weird, Scary Pictures
AI generated images are often weird looking. The picture is distorted, there are extra limbs or it is not centered correctly. If you see an AI image online then it was probably not the first attempt. It could take 10 tries or more to get a result that looks normal
That means the AI art generators are good for making weird pictures. Ask for an image of a person or an animal and it may look like a scary monster. You may want to delete them but monsters, mutants and zombies are interesting.
As a result of using text to image generators I have hundreds of frightening images on my computer. Scary pictures is a popular search term. It gets a lot of search traffic according to Google's fill in the blank and the keyword tool I use.
So Bad It's Good
Some of the bad images are good because they are not normal. It is stranger than reality. I asked for a picture of an endangered animal and ended up with something that could be from a horror fantasy movie.
An image can be terrible and amazing. Stable Diffusion pictures that went wrong may be more interesting than the ones that followed your expectations. If you end up with a low quality photo, remember that small low quality GIFs are extremely popular. It just needs to be interesting or funny.
You are more likely to accidentally generate weird images while using photo prompts. It is frustrating but once in a while I am rewarded with an unexpected masterpiece. The photo below is not what I was going for. It is not centered and the quality is not very good but I think it is astonishing.
Weird scary picture. The prompt was "Photo of an endangered animal".
Michael H
CGI Video Game Characters
Video game characters are designed to be interesting to look at. They come from a person's imagination and they don't need to look realistic. So most of the creatures are strange and some are terrifying. At least they would be if you met one in real life.
They tend to be pixelated or cartoonish until someone makes a movie about them. Films based on video games may create a more lifelike version by using costumes, props and CGI. You have seen CGI(computer generated imagery) in movies and TV.
It usually takes a lot of time, money and equipment to make CGI effects. With AI image generators you can do something similar for free. Make a computer generated picture of a monster from your favorite video game.
AI Generated images of human video game characters may look like Cosplay. The result appears to be a person in a costume. Generating a photograph of a person wearing a store bought custom is not very impressive. So I mainly focused on video game creatures and monsters.
Photography Prompts
An example of a simple Stable Diffusion photo prompt is "photograph of a deer by National Geographic wildlife photographer". You tell the AI that you want a photo and it creates a photo-like image. Add photography terms like "Canon DSLR, lens 300mm" to improve the quality.
Text to image generators can be used to make fake pictures of almost anything. The subject can be real or imaginary. It can be from the present, past or future. Instead of a deer you could ask for a dragon or a pirate.
Most art is inaccurate and unrealistic. People don't expect it to be an incredibly accurate representation of real life. Quick sketches and abstract art that does not really look like anything can be worth millions of dollars. Photographs are supposed to be a nearly perfect 2D copy of a real life scene.
If a computer generated photo does not look incredibly accurate then it looks bad. So you may need to use a lot extra keywords and phrases in the prompt to improve the quality.
Negative Prompts
Tell the program what you don't want by adding keywords and phrases to a negative prompt. If you want a high quality photograph then you don't want an unrealistic result. So you would add words like unrealistic, distorted and extra fingers.
I used negative prompts to improve the quality of the photo-like images I was making. Then I used them to make my paintings look more realistic. If you generate an ugly abstract painting you may not need to switch the style or the artist. Try adding some negative keywords.
Starting with an unrealistic prompt and increasing the realism with a negative prompt can give you some interesting results.
This is supposed to be a photo of a video game boss.
Michael H
Useful Keywords
unrealistic | blurry | retro |
low quality | out of focus | contemporary |
inaccurate | out of frame | cartoon, cartoonish |
disfigured | colorless | CGI |
ugly | gloomy | computer graphic, graphic |
low contrast | close-up | art, digital art |
dull, dark | signature, watermark | abstract |
low-resolution | analog | visible lines |
unreal | digital | visible strokes |
Creating a Monster
After accidentally making a realistic image of a video game monster I wanted to figure out how I did it. Using photo of a video game monster as a prompt gave me pictures that looked like video game art from the 1990s. Some of them looked pretty good but they were not realistic.
Adding by National Geographic wildlife photographer improved the realism. It also increased the chance that the monster looked like a mutated animal. Many of the images were too messed up to look real. So I added words like unrealistic, inaccurate and low quality to a negative prompt.
The next step to improve the quality of the image was to include more photography terms and keywords. I added studio quality, ambient lighting, 30mm, background woods to the prompt. You could use different terms.
To show a step by step comparison I made a before and after picture with 4 monsters. Each monster is a different step. Adding to the prompt made a big difference. The monster evolved into something that is more realistic.
What happens when you use a photograph instead of animation for a video game character?
If you are using an image generator then you end up with a lot of weird pictures that get your attention. They are interesting to look at and they stand out. There are all kinds of video game characters. So it was easy for me to generate unique images.
Include keywords and phrases for a high quality photograph in a video game character prompt and you end up with the live action version. It becomes more realistic than the original. The character looks as real as a CGI creature from a sci-fi movie.
Monster Evolution
Photo of a video game monster. Added quality and realism keywords to it 3 times.
Stable Diffusion Prompts List
- Classic video game monster
- CGI monster based on video game, stunning environment
- Hyperrealistic CGI monster based on video game, stunning background, lifelike, accurate
- Video game boss, hyper-realistic graphics, stunning background, movie quality, fantasy
- Movie monster based on popular video game, HD, 8k, stunning background
- Movie villain from a video game, HD, 8k, spooky environment, CGI, photorealism, face, medium shot
- 8K, extremely high quality photograph of video game creature by professional photographer, stunning background, trending on National Geographic
- Award winning analog photograph of rare Pokémon creature by National Geographic photographer, Unreal Engine
- A hyperrealistic portrait photograph of a video game character, high quality, 8k, hyperrealism, highly-detailed, background night sky
- Hyperrealistic, accurate, extremely high quality photograph of video game boss monster by professional photographer, stunning background, trending on National Geographic, 8k
- Photograph of video game creature from 20 feet away, medium shot, stunning background, picture of the day, highly-detailed, smooth, sharp focus, 8k, 4k, post-processing, ambient lighting
- Accurate, National Geographic photo of the day, high quality, hyperrealistic, 8K, HD digital photograph of video game character
- Award winning professional photo of video game boss monster by National Geographic wildlife photographer
- Movieland Wax Museum sculptures, accurate, high quality, hyperrealistic video game monster, highly-detailed
Tips to Improve Quality and Realism
Stable Diffusion will show the generated image at a reduced size. I generate a 768x768 picture and it looks great at 336x336. So I save it to my laptop. Then I view it at 88% of its original size and it looks bad. It is a common problem I have.
Stable Diffusion's maximum width and height is 1024x1024 if you use their DreamStudio. An easy way to improve the quality and realism is to reduce the size of the picture by 50%. Start with an extra large image. Then make it smaller.
Sometimes you can make video game creatures more realistic by using real places. I used Banff National Park, Amazon Rainforest and the Mojave Desert as backgrounds. It places the video game character in an environment that looks real and it increases the chance that the character will look lifelike.
Reduce the randomness by using a specific game like Diablo, Doom, Mario Brothers or Pokémon. If it is a well know character then don't say it is from a video game. Just mention the character's name. You could try portrait photograph of Laura Croft.
It is easier to generate a photo-like image when you start with something that is analog or non digital. Things that would not appear on a screen as a computer game graphic. They include people, people wearing elaborate costumes, wax figures and animals.
Phone Screens
Devices that have small screens display pictures at a reduced size. All the low quality images I saved to my laptop looked great on my phone screen. If people are going to look at it on a small mobile device then it does not need to be a high quality image.
The downside is that the pictures that look good on your phone may look bad on a laptop, desktop or tablet. Using text to image generators on your phone is very convenient but the picture may only look good because it is being displayed at a reduced size. That is why I stopped using Stable Diffusion on my phone.
Realistic Photos of Video Game Characters
If you are going to generate a realistic picture of a wolf then make it a giant wolf. Use Stable Diffusion to generate extraordinary images. Photos of video game creatures and monsters are interesting and usual.
When using AI image generators you want the pictures to be incredible. Combing computer graphics or art with photography is a way to do that. Try to come up with ideas for generating images that don't look like an ordinary common camera photo.
The prompts I used were random because the description was short. It gave me a wide variety of pictures. I could not say who or what the subject of the photo is but they get your attention. Generating a large number of interesting images was easy. The hard part was only picking out a few examples to display on this page.
With art you reduce the realism to turn the ordinary into the exceptional. The idea here was to start with something that is unrealistic and increase the realism. Photos of video game characters is just one approach.
For me the hardest part of using Stable Diffusion to make photo-like images was coming up with good ideas. Generate a photo of an ordinary subject like a person and you get an ordinary picture. Thinking of great art ideas was easier because I was not trying to make realistic images.
The thing I liked most about using video game character prompts was the variety of eye-catching, memorable pictures. Generate a photo of a monster or a video game character and you usually end up with an extraordinary picture. I didn't even need to come up with a specific idea.
Try my prompts as they are. Then change them or get inspired to create a completely different prompt. You should always try to come up with your own ideas. Experiment with text to image prompts and hopefully you will generate some incredible images.
Useful Links
- Writing Great Prompts for Stable Diffusion
I started using text to image apps when I ran out of photos, paintings and drawings to post online. It is an easy way to create pictures, but you need to come up with a good prompt. Use the right prompt and the results could be amazing. - Stable Diffusion AI Art Prompt Ideas
Using text prompts to make AI art is easy but you need to start with a great prompt. I experimented to see what I could create. Then I made lists of the best keywords and phrases to make it easier for me to generate stunning paintings and drawings. - How to Make AI-Generated Art With Style Transfer
Create non-photorealistic images by combining the subject of the first picture with the style of the second. Making AI-generated art is fun, and the results are impressive and interesting to look at.
Text to AI image. Interesting photo of a video game monster.
Michael H
This content is accurate and true to the best of the author’s knowledge and is not meant to substitute for formal and individualized advice from a qualified professional.
© 2023 Michael H
