What is the most realistic free AI for generating images? We know of many generators and Midjourney surely takes the cake, but as it is paid, it is not within the reach of someone who wants to use it occasionally or who simply does not want to spend a euro.
This is why we agreed to five free image generators for iPhone and which we can access, either through the web or through an application. To do this, we asked for two different images (with identical prompts in each). We've already told you that there isn't a huge difference between them, although there is one that, in our opinion, stands out above all.
What prompts did we use
So that there is no trap or cardboard, I chose to use same prompt in all AI. I actually tried it in both Spanish and English. The majority gives similar results in both cases, but in some specific cases they tend to understand English better and that is why they also used it in Shakespeare's language as a wildcard.
My idea was to analyze a little about natural elements and also humans, given that it is the latter who perhaps continue to resist generative AI the most. For the first one I chose a landscape and I used this prompt:
“A mountainous landscape at sunset, with an orange and pink sky. In the center of the image is the waterfall of a river that winds and has golden highlights. On the sides are rocks with moss and many deciduous trees. The environment must convey tranquility and majesty.
For him environment with people I chose to order a market in which several characters appeared:
“An open-air market on a sunny morning and several people are interacting with each other. There is a woman in a floral dress bargaining with a fruit seller. A little boy with a straw hat is holding a red balloon in his hands while curiously looking at a street musician who plays the guitar. The others stroll through the stalls. The atmosphere is lively and very colorful.
DALL-E 3 with ChatGPT and with Bing
DALL-E 3 is OpenAI's image generation technology and is present not only in ChatGPT itself, but also in Bing. In both cases, it can be used both from the browser and in the iPhone (or Android) application. The results are similar.
In the ChatGPT image we find a landscape quite faithful to what we asked for and which could pass for real. Now, like the market, we see how it fails in certain elements like not understanding that it is the child who is holding the ball and, incidentally, draws faces very badlyespecially in the background characters.
Access ChatGPT via the web | Access ChatGPT via the iOS app
Zero problems also in the landscape image generated by DALL-E 3 on Bing, although there are some in the market one. It omits elements like the ball and doesn't end up generating too colorful an atmosphere. On the other hand, it must be said that in general it seems to be a very realistic picture. Remove the face of the lady walking in the background, firstly you might think it's a real photo. With an editing filter, but real nonetheless.
Access Bing Image Creator via the web
Grok
X's AI, called Grok and available for free, has what Elon Musk wanted: no limits. Many of us have seen creations of popular characters in abnormal situations that we find funny and hit home. However, since other models are censored due to copyright and publicity rights, we did not wish to make a distinction and request identical prompts.
With Grok we find ourselves in a strange position: These are surely the most realistic images, but less precise. Although in the landscape it conformed to what was asked, in the market it ignored many elements. For example, it generates the ball in the background without the child carrying it and there is no trace of a musician playing the guitar. Additionally, he draws strange hands as if these people have some sort of deformity. And in all the generated images the same thing happened, with this one ultimately being the most accurate.
Access Grok via the Web | Access Grok via the iOS app
Ideograms
Ideogram is accessible for free via the web and, like Grok, also offers the possibility of generating realistic images, even of famous people. While following this self-imposed rule, we also chose to use prompts identical to the previous ones.
The landscape, once again, we see that it is completely realized. Perhaps something unreal in the waterfall part, since it starts from a slightly strangely shaped slope. As far as the market goes, it generated up to two musicians, which isn't bad either, but that's not what we're asking for. Likewise, he puts the ball in the hand of a girl and not the boy, in addition to completely omitting the fruit seller with whom the woman in the flowered dress negotiates. However, and despite some imperfections, It also seems very real.
Access the ideogram via the Web
Artgourou
It is not the best known, but Artguru's AI is one of the most interesting because it is free, although like Grok, it adds its own watermark. It is accessible via the web and here are the results.
Both images are pretty, but maybe too much to seem real. The landscape looks more like a very well drawn painting with a water effect which, we insist, is beautiful, if a little unreal combined with too well textured moss. The market image loses almost all of the elements asked for in the prompt and leaves a few unclear faces in the background, but the truth is that it's not bad execution at all.
Access Artguru via the web
And the winner is…
As Mercedes Milá would say in a “Big Brother” from the 2000s, the public is sovereign and it is they who decide the winner. So we showed the results for everyone to judge. Now, if our opinion is useful to you, we think that Grok's AI is above the rest of the free AIs. At least in general since we have been using them all for months.
It's not that there is a gigantic difference between Grok and the others, but the fact that there is practically no censorship makes it very convenient for generating any type of images, although it also has a negative reading. Of course, it remains to be seen how much longer Elon Musk will allow us to use it without having a subscription to X.
As a second option, the truth is that DALL-E 3 on ChatGPT is also very attractive and above all to better understand the request we are making, however complex it may be. It is true that it is only available with GPT-4o and we have a free usage limit, but to generate a few images per day it is more than enough.
In Applesfera | Image Playground: what it is and what it is for, the new Apple application with which you can create all types of images
In Applesfera | Apple Intelligence: what it is, what you can do, compatible devices and when it will arrive in Spanish
Table of Contents
