Free AI image generation websites

For all the people who have company budgets to pay for computing power and other fees, good for  you. But others enjoy free AI tools on the web. 

These initial images are crucial for seeding a series of images, or a video. More on that another time.

Here's a snapshot of free tools right now - January 2024 - and a few tips and tricks to move you forward.

 

 

The answer is that many sites can get you started if you use good prompts. Oh, and here's the thing about prompts. I think of them like a community resource - take 1, share 1 - and we can all enjoy.

Part 1: tips I've picked up from various sites. For example, let's say you want to see...

• Two future cyber-women kiss

Include where:

• A vast indoor tech operation edge of space, iron city

Style

• Photo-realism, fantasy, futuristic
  

Colors

• Iridescent gold, colorful background

Lighting

• Cinematic lighting

Website where similar high-quality art is featured

• artStation

Quality

• Sharp focus, highly detailed faces, intricate sharp details, realistic bodies, 8k

Negative prompts

• Grainy,  Bad art, Draft, Blurry, Distorted face, Amateur, Beginner, Overexposed, Underexposed, Low contrast, Cut off, Signature, Watermark, Bad anatomy, Deformed, Disfigured, Extra limbs, Tiling, Ugly
   

One last tip: take inspiration from everywhere and use it to seed a variety of options. I'll show you the results of this prompt on various sites and show what a tool like Wombo, one that uses image seeds, can do.

 

Part 2: free sites.

aiimagegenerator: not a bad start, but didn't get the point

bing: good colors and reflections, strong composition

Craiyon: example of a good basic characters, but poor clarity. That's often the case with Wall-E.

 Now drop this into Wombo and you'll see how it completes the image with no fussing

deepai: missed the initial action, the rest is interesting, similar to Craiyon

seeded in Wombo

dezgo: good job

 freeimagegenerator: very interesting response

with Wombo:

huggingface: pretty cool unexpected response to the prompt

What if we drop that into deepai? It's a bit clearer

Here's how Wombo interpreted it

imageeditor: potential, very different response

Perchance: good

Wombo changed it a bit

 

tinywow: missed the point, but interesting

 

Wombo could initially only take it a bit further

Time for:

Part 3: a major trick here: if you're not getting the character response or action you're looking for, try changing the style with your best seed. You might find that you get closer to what you're looking for.

anime style in Wombo

and then back to photo-realism