Who am I?

Hello and welcome to the next article in the AI Knowledge Hub! 🙋‍♂️

My name is Jacob Kowalski - a corporate tax consultant from Poland who is really into AI stuff.

I am creating with AI many really cool things like: chatbots, AI assistants, automations, agents, etc. 🤖

I think we are living in a the biggest revolution of our generation and want to share my thoughts about it. Maybe for some of you they will be interesting or maybe you want to share your take on this too.

If you want to know more about me and what I am doing, check out my page with button below. 👇

“The one to rule them all”

Once I’ve started creating content - like this article, I needed to create some visuals to make everything more user-friendly. I had some options on the table. 🖼️

First - create visuals by myself - completely not possible as my graphics skills are limited to basic image editing on Canva.

Second - take advantage of AI revoulution and use AI Image Generators - this one seems more interesting but was not as easy as I thought.

The main problem is too wide a choice of tools to use - a real curse of abundance. 🎉

I didn't quite know which tool would be best for a given case.

I figured this probably wasn't just my problem, so I decided to collect the tools I used the most - test them for given use cases and ultimately see which tool performed best. 🏆

Introduction

Artificial intelligence is revolutionizing the way we create and perceive digital art.

Today's market offers an overwhelming number of AI image generation tools, each with its own capabilities and limitations.

Choosing the right generator can be daunting, especially when most of us are looking for solutions that offer high quality without requiring significant costs.

That's why I decided to test 5 of the most popular and relatively accessible AI image generators:

To ensure a fair comparison, I used exactly the same prompts for each tool, allowing me to objectively evaluate their strengths and weaknesses.

In the ranking below, you'll find a detailed analysis of each tested generator, complete with examples of generated images and an assessment of their functionality.

Check out the results of my test and discover which AI generator will best meet your creative needs. 👇

Product Visualization

In this case I wanted to create an image to prezent the product (in this case a Juicer) in the most eye-catching way.

Gemini’s image is looking great! I really like the background and the “juice explosion” effect.

When it comes to the negatives I have a hard time to find any, one minor negative is that everything looks so bright that feels little unnatural.

In my opinion almost perfect image: 4,5/5

Deep AI image looks moderately good. The juice explosion is very good, and the fruits also look well.

Unfortunately, in this case, the negatives are visible at first glance.

The upper part of the juicer does not match the whole device, the juicer itself does not look as a very favorable - it is closer to a kettle.

The water glasses completely do not match the graphics, in addition, the fruit next to the apples was probably supposed to be a grapefruit but the model had a problem generating it.

In general I would rather not use this tool to generate product graphics - 2/5

Image from SORA is looking very nice, again, the colorful background blends very well with the fruit and the product.

In the case of SORA, the style of graphics is a bit more cartoonish than actual, but in my opinion this is more of an advantage than a disadvantage.

As far as I'm concerned, SORA generated great graphics with no apparent negatives.

Of course, in the case of graphic style, it is a matter of taste - in my opinion a favorite 5 / 5

In this case, Leonardo AI fell a little short. I'm a fan of the fact that the fruit, the juice and in part the product look very realistic however other than that nothing here fits.

The explosion of juice in places is water, the appearance of the juicer is strange to say the least, I have the impression that everything here is a little mess - 1/5

In this case, it's hard for me to say anything good.

It is possible that the prompt needs to be expanded and clarified, but at this point the graphics are a collection of random items that do not fit together in any way 0/5

Advertising Banner

With this, my goal was to create an eye-catching banner that would inform about Black Friday price cuts.

For me looks perfect!

Everything here matches, the colours, text and products - 5/5

This one looks fine as well. I might have used it if it weren't for the mistake with the text.

Other than that, very decent graphics - 3/5

Again, great graphics, very interesting differentiated backgrounds and detailed products.

Graphics that do not need any improvements - 5/5

As far as I am concerned in terms of graphic concept the best so far.

Leonardo AI apparently has a problem with text generation and in some cases generates strange looking products.

Of course, the graphics can be further modified, but for the above reasons I would give a maximum of 2/5

The concept also looks interesting but beyond that it's hard to find anything positive.

Different fonts, positioning of the text and the products themselves - through all this my rating to 1/5

Blog/Article/Newsletter Header Image

This use case is particularly relevant to me as I myself mainly use image generators for this purpose.

Once again, Gemini is showing its great side when it comes to generating images.

The graphic style as well as the whole image composition is very eye-catching.

The color palette blends very well together - in my opinion 5/5

The graphics from Deep AI look.... rather strange.

The combination of parrot, jaguar and butterfly doesn't look too good, let alone that “something” that is in the upper left corner 1/5

The graphics from SORA just look good.

I like both the style and the text itself.

However, everything is so “ordinary” here I miss a larger color palette like Gemini, nevertheless a very good image 3,5/5

In this case, Leonardo AI has finally shone.

Despite the lack of more animals, the diverse parrots contrast brilliantly with the greenery of the jungle.

Leonardo does a great job of creating views, the jungle looks amazing along with the river flowing in the middle - 4/5

Once again, I get the impression that Runway AI didn't quite know how to go about creating this image.

It all looks like pasting animal graphics into some external background. The graphics are rather unusable - 0/5

Digital Art

The proliferation of image generators has created a new niche in the market - selling digital images (e.g., on ETSY), so I checked how the tools would handle the creation of artistic graphics.

The graphics look great the detail of the Earth as well as the octopus is very high.

The colors in the background are also perfectly chosen.

I would only change the look of the octopus itself - but this is probably a matter of taste - 4,5/5

In the end, Deep AI has shown that it is able to rise to the occasion, it may not be Gemini level but in my opinion a very interesting interpretation.

I really like the blue glow of the planner and the overall color scheme of the graphics.

The only thing I miss is a little more detail - 4/5

The graphics from SORA look completely different from its predecessors, and that's something that really impresses me.

The color scheme, the different perspective and the overall graphics all come together perfectly - 5/5

Leonardo once again confirms that he knows how to create views and distant objects.

The overall graphics look quite good, unfortunately everything is spoiled by this octopus with a somewhat goofy eye - 3/5

This time Runway AI really surprised me.

The concept of the octopus and its color scheme is something that completely appeals to me.

Also the background and the planet look very good, I would gladly frame something like this and hang it in my room - 4.5/5

Inspiration

In this section, I asked the tools to generate an image that best shows their maximum capabilities.

Of course, different tools have different uses, some generate better conceptual produt photos, others advertising graphics, others will be better at generating images from a fantasy world.

Therefore, I decided, to let the tools show their true potential.

Gemini has decided to show off its vision of the city of the future - or something like it.

The whole picture looks quite intriguing, the combination of modern buildings and the branches that wrap around them fits very well.

The flowing river reflects the image very well, and the light has been shown in an excellent way.

Gemini opted for fantasy combined with graphic realism, and in my opinion it passed the test - 4,5/5

Deep AI decided to generate something more down-to-earth.

The road leading to the volcano (?) looks quite interesting but very ordinary.

Little detail here, I like the play with light, while not enough happens in the image to show the high capabilities of the tool 2/5

SORA has quite high artistic aspirations, and likes to play with styles - something I'm a huge fan of.

The generated image is very reminiscent of the previous graphic of the octopus in space - and in this case it hits my tastes.

The graphics don't try to crank up the maximum realism but it shows that AI can also be creative 5/5

After generating this image, I have a feeling that all this time I have not used Leonardo AI in a good way.

The image it created with Leonardo this time beats all the previous ones.

The level of realism and detail in this graphic is simply amazing - the whole composition also hits my taste 5/5

An analogous case, Runway AI created something I never thought it would be able to create.

I really like the colors and style of the whole image - the crystals sticking out in the background look amazing, the reflections in the water also look great.

I have a feeling that the look of the sky could be improved a bit - other than that - brilliant work - 4,5/5

Summary

Considering the quality and flexibility of the Gemini and SORA models, they leave far behind the rest of the stakes.

Whether you want a product photo, an advertisement, a concept graphic or something else entirely, both tools are capable of generating something from nothing.

Deep AI - as a tool quite simple and very friendly to free use allows quite a lot, it has in its resources a list of styles that can be customized for our concepts, nevertheless when it comes to the graphics themselves they are of slightly lower quality than their predecessors.

Leonardo AI - Leonardo is a rather narrow tool and requires at least a few hours of training. As you can see, simple prompts are not quite able to squeeze the maximum possibilities out of it. For the daily generation of graphics for social media - there will be other, simpler and cheaper tools.

Runway AI - in relation to this tool I have mixed feelings once it is able to generate incredibly good graphics, and then something tragic. It seems to me that, as with Leonardo, the applications are a bit different and the tool requires more knowledge and skill to master it.

Sources

The sources I’ve used to create this article are:

• https://sora.chatgpt.com/explore

• https://gemini.google.com/app?hl=pl

• https://deepai.org/machine-learning-model/text2img

• https://runwayml.com

• https://app.leonardo.ai