If you want to generate images with artificial intelligence but don't have access to a graphics card or a local installation, there's no need to worry. The online offering is much more abundant and interesting than we imagine, and the best part is that we can get a large number of images without spending a cent. Today we take a new look at the main free services, both on computers and mobile devices.
The local option is superior, but…
… we know well that it doesn't apply to everyone. A typical installation of Auto1111 or ComfyUI (the most popular Stable Diffusion interfaces) requires some compatible graphics card, for the simple reason that nobody is going to waste half an hour waiting for an image that will probably be discarded immediately. The trial and error cycle is inevitable in image generation, and time is something we shouldn't lose here.
That's the big advantage of online services. Unless there's a technical issue in the background, results appear in seconds, and if we don't like what we see, a simple click triggers the next iteration. Today we will review some of the best AI image generators that maintain a free profile, and that don't drive us (too) crazy with their restrictions.
Bing Image Creator
Bing's image generation uses DALL-E 3 technology, and without doubt it is a very effective route to access it. Bing Image Creator works with 'boosts', an equivalent to traditional tokens that are consumed at a rate of one boost per four images. Boosts renew weekly, the base resolution is 1024x1024 in JPEG format, and all results include a tiny watermark.
Stable Diffusion XL via Google on HuggingFace
HuggingFace is one of the most interesting platforms for evaluating different models and projects. However, not everyone knows that Google has an official profile there, and it maintains a generator based on Stable Diffusion XL. Its interface is relatively limited, but with ten presets available, guidance control, and support for negative prompts, the truth is we don't need anything else. I recommend taking advantage of it while it lasts, because it's being bombarded with requests…
Ideogram AI
As we mentioned before, the strong point of Ideogram AI is text generation. Conventional engines do nothing but present illegible scribbles, but with the help of a simple prompt, Ideogram AI responds wonderfully. If the plan is to create a logo or something similar, text control and its tendency to generate images with Midjourney's aesthetic place Ideogram AI a step above the rest. That said, I recommend not wasting time with photorealism here. Its priority is something else.
Leonardo.ai
Leonardo.ai is probably the strictest when it comes to tokens and free image generation, but also one of the most complete. It includes proprietary resources such as PhotoReal, Alchemy, and Prompt Magic that improve prompts, modify generation parameters, and optimize images. Of course, this increases token consumption (in fact, it skyrockets under certain combinations), and all these functions are actually trials, but Leonardo.ai enables 150 daily tokens, more than enough to get started.
Mage.space
With the disappearance of the free mode in Clipdrop, many users turned to the Web looking for an alternative, and fortunately they found it in Mage.space. Three models available (including Stable Diffusion XL), control of steps, guidance, seed, refiner, negative prompt, image format... Mage.space includes several hardcore options that generally only appear in local installations. If the copy of Stable Diffusion XL that Google offers doesn't work or isn't appropriate, Mage.space will do a great job instead.
GenZArt
Unfortunately, the people at monAI Art decided to limit their beta even further, and personally I think that reduces our options for generating images on Android to two. The first is GenZArt, with a very generous free mode (no cap or ads), three different formats, image upload as reference, and additional artistic styles. Add to that layer-based editing to replace objects, and upscaling. I insist: there is no cap on the number of images generated in GenZArt, and it's growing quickly.
Dream by Wombo
The second option for generating images on Android is none other than Dream by Wombo, the mobile version of one of the first generative models we've seen here at NeoTeo. Its operation is very similar: first the prompt, then the style, and finally a reference image (optional). Images are downloaded as if they were trading cards, or take on the role of wallpaper on the mobile device, but Dream by Wombo also incorporates video in its premium mode.
In summary
Yes, the number of AI image generators that abandon or reduce their free modes keeps growing, but there are still many free platforms to explore that deliver extraordinary results. Our selection only points in the right direction, and we hope it evolves over time. Good luck!
