Microsoft Bing is entering the AI image generation fray.
Bing's image generator, simply called Image Creator is free to use. All you have to do is create a Microsoft account.
I was quite disappointed when I first signed up, because of the slow generation speed. I'm guessing this is the result of surge in users since their launch: there's just way too much demand.
Since it is Microsoft, they emailed me when my generations were complete.
Let's get into the prompts:
Graphic Design
With logos, it's important to state exactly what you're looking for, including the background, shapes (if any), colors, and letters.
If you're looking for a logo with geometric shapes, below is a prompt that'll suit you best:
colorful geometric shapes on white, graphic designHere's another example with circles and a blue background. The idea is the same, but slightly modified.
A simple logo with 4 circles and a blue backgroundRenaissance Paintings
Now, moving to 15th century renaissance paintings. Microsoft's image generator seems to do them pretty well.
You can opt for anything normal or as odd as humanly possible - like an elephant wearing a tuxedo.
To produce a renaissance painting, specify the term 'renaissance' in your prompt.
Below is an example:
A renaissance painting of an elephant in a tuxedoBelow is another prompt following the same style, but instead, it's a monkey steering a boat. He's wearing a captain's uniform for extra effect.
A monkey steering a boat while wearing a captain's uniform, renaissance paintingDigital Art
It's clear most people would intend on using Image Creator for digital art - for social media or blog purposes.
To generate digital art, simply state 'digital art' at the end of your prompt.
I started with a guitar made from flowers. The image contains exactly what the prompt says on the tin; the shape of a musical instrument constructed with mostly flowers.
To get a guitar made from flowers, copy the prompt below.
a guitar made of flowers, light green background, digital artThe second prompt features a cute fuzzy hamster holding an orb of light. The detail of the animal is good, but it still looks a little like a painting.
The background has other orbs, implying there's other hamsters in the image. Perhaps, this hamster is holding the universe - the other orbs being parallel universes.
Cute fuzzy hamster holding orb of light, digital artThe third prompt involves robots building other robots - a concerning reality we may be facing in the near future. After all, we already have AI building its own AI models.
This prompt is inspired by Terminator Salvation. If you have't seen it, it's worth a watch.
A robot building other robots, digital artThe fourth prompt was inspired by a David Holz interview, where he stated people were coming up with prompts like "A space dog with lasers". Of course, this isn't Midjourney, but it's interesting to see other models generating Midjourney prompts.
The image includes a dog in a space suit, shooting lasers at a target. In the background is the infinite void of space, along with a galactic cluster.
A space dog with lasers, digital art3D Digital Art
On the topic of 3D digital art, I created 2 3D Funko Pop prompts. One of them is MF Doom, and the other is of David Holz.
3D Funko Pop of MF Doom, Digital ArtThe David Holz funko pop does show a resemblance to him.
3D Funko Pop of David Holz, digital art Ancient Art
Recreating ancient art with modern technology sends shivers down my spine. It's truly an exciting prompting style. One of my personal favorites is Ancient Egyptian tablets.
The first prompt includes an Ancient tablet of people holding, looking at, and using modern smartphones.
An ancient tablet of people using smart phones in the style of Giza in Ancient EgyptThe second prompt has a Roman sculpture of Julias Caesar wearing a VR headset. Although this could easily be done at the museum where sculptures of him are present, this image is still pretty special.
A Roman sculpture of Julias Caesar wearing a VR headsetCartoons
Creating cartoon images can easily be done with AI. No matter how tame or atrocious your prompt is, Image Creator is likely to come up with something accurate - if it complies with their policy.
The first cartoon prompt is of a boy getting abducted by a UFO. He's clearly in his teenage years, as he has a beard and mustache. Alternatively, he's younger and has made one from chocolate milk.
A boy getting abducted by a UFO in his garden, cartoon styleThe second prompt features humanoid dogs wearing suits in a courthouse. It's inspired by 'Dogs Playing Poker' by Cassius Marcellus Coolidge.
Humanoid dogs in a courthouse, cartoon styleTV
Using the styles of TV networks is a great way to pass the time.
The first prompt features a lady holding the number 2. It's in the style of a BBC cooking show, and looks exactly like what you'd see on BBC 2 in the UK.
the number 2 with vegetables in the background, the style of BBC cookery showsThe second prompt is something closer to home. It includes a CNN news reporter standing and holding a newspaper - waiting to read out the latest headlines.
A news broadcast in the style of CNN newsAnime Characters
A lot of people are interested in generating their own anime characters.
The first prompt includes an anime character with a glamorous outfit and detailed background. It's in form of digital art, which does help with the intent of the prompt.
Anime character with glamorous outfit and detailed background, digital art
The second prompt doesn't feature digital art styling, but instead, is fully anime. It has a blurred background, while the character is wearing all things blue. Her hair and eyes are also blue.
This is the style you get, unless you don't instruct the bot with an additional style.
Anime character with blue outfit in shopping mallMaking Cartoon Characters Real
Turning cartoon characters into real people, or something close to that, is very fun.
First, I tried making Homer Simpson look real. The result looks more like a clay motion or porcelain model.
Homer Simpson in real life, realistic photo visualThe second prompt is an attempt at making Rick and Morty into real people. In a way, I kind of succeeded, but only with Rick. Alternatively, the person on the right is Morty in the future, and Rick has time traveled to see him.
Rick and Morty in real life, realistic photo visualTurning Real People Into Cartoon Characters
Turning real people into cartoon characters can also be fun.
I started with HM King Charles III in the style of The Simpsons. Although, the image doesn't resemble him in any way. Instead, it looks more like King Charles II who reigned over Scotland (1630-1685).
A portrait of King Charles III, Simpsons StyleFor the second prompt, I went with Wednesday Adams in the style of Family Guy. This result is far more accurate, and looks just like the character I tried to recreate.
She has a suspicious look about her, which adds to the accuracy of this image.
A portrait of Wednesday Adams, Family Guy StylePhotovisuals
Image Creator's photo visuals aren't as realistic as Midjourney or Stable Diffusion, but they are still excellent pieces of work.
Prompt 1 features a flying car. The art follows the prompt to the letter, and has included a background blur to imply speed - or to make the image look more professional.
Realistic photo visual of a flying carWith prompt 2, I wanted to see how well the tool generates people. It looks a little cartoon-ey, but the level of detail is good. It also features a background blur to make the image appear more professional.
realistic photo visual of man selling ice cream in IndiaInk Drawings
The ink drawings Image Creator produces are highly detailed, and tailored to the prompts word-by-word.
The first prompt includes a robot sheltering from the rain. There's a lot to say about this image, as it shows a time where physical robots are the norm and are possibly conscious. However, they still aren't waterproof.
robot holding an umbrella in the rain, ink drawingWith the second prompt, I decided to go for New York at night. The end result is a thorough ink drawing of the city after sunset.
As the reflection of the light pollution illuminates the puddles, the distant buildings add depth to the image.
New York at night, ink drawingPixel Art
For pixel art, I opted for a dog floating through space. This came from asking the bot to generate something at random.
Pixel art of a Jack Russel Terrier floating in space with moon and starsAs per the second pixel art prompt, I came up with the US president sitting in the white house with Bowser looking through the window.
Pixel art of the US president sitting in the White House with giant Bowser looking through the windowNote: I could not specify which US president the art could contain, as it goes against their policy - and was blocked as a result.
Fantasy
Creating fantasy artwork on Image Creator is easy and effective. First, I started by writing a simple prompt of 4 words: mountains, fantasy, digital art.
Below is the result, which is a masterpiece on its own accord.
mountains, fantasy, digital artFor the second prompt, I went a little bit more experimental. Also, I made things more complicated for the model.
Castle Coch is a real castle, situated in Cardiff, Wales, UK. For this prompt, I thought it would be a cool to create a fantasy image based on a actual historic structure.
In essence, it's possible to generate fantasy art based on any building. Although, using a building from an older period makes it more believable.
Castle Coch (Wales) in a fantasy settingPrompting Tips
Image Creator prompts are simple one-liners with art styles added to the end of them.
It's default style are both photographic and digital graphics. It's what gets generated if the prompter doesn't receive specific styling instructions.
If you want to generate something different photographic or digital graphics, you'll need to include the styles you want - separated by a comma or part of the overall sentence.
For example:
Castle Coch (Wales), fantasy paintingA painting of Castle Coch (Wales) in a fantasy styleThese two prompts would generate the same thing, more or less.
A lot of prompts get blocked if they go against Image Creator's policy. So, this is something you need to be aware of.
Conclusion
While the quality of Image Creator isn't as good as Midjourney, it's certainly nothing to dismiss. It's also free, while Midjourney starts at $10/month.
The AI produces images of adequate quality, similar to those of Stable Diffusion.
Image Creator does come with 18 fast generations, but after that, you'll have to wait for 45 seconds for each prompt. Its' actually around 2 minutes wait-time.
