Which AI Image Generation Tool Should I Use?

While ChatGPT can generate images, it's relatively simplistic and there are a lot more robust options for generating images with AI tools.

In this article, we'll be covering a few different AI tools, in no particular order:

• Midjourney
• Replicate
• Runway
• Dzine
• OpenArt
• ComfyUI
• and notes on other platforms / ai tools.

Midjourney

Is one of the more popular options and is mostly focused on text-to-image (although more img-to-img editing is being released). One of their unique features is being able to get more personalized generations by going through a personalized pick which you prefer (out of 2 images) which the system will then take into account when creating new images.

Overall, I'd say the plan is a bit pricey as it offers around 200 image generations for about $8/mo. And while that might not sound expensive, when you compare it to some of the other options out there, it will definitely seem expensive.

Replicate

Okay, so Replicate is a bit more focused on developers, but there's nothing that says you can't just use the web UI they have provided, use their different models, and get the images you want.

One of the pros of this approach is that you're only billed on what you use, and so some models may cost you less than a penny for generating a few images. You'll usually see the price ranges from around 20 - 150 images generated per dollar, so if you do the math on that, compared to Midjourney, it's extremely cheap, especially since you're only paying for when you actually run the models, so there's no subscription you're paying even when you're not using Replicate.

The con is that you'll have to dig through all the different models and fine-tune them to get the result you're looking for, therefore some knowledge of AI based terminology and how the image generation process works is needed.

Runway

Runway is focused mostly on video generation. They have no free plan for video generation, and at this time I haven't needed video generation, so I don't have much to say about this one, regardless they have a section on their dashboard called "runway watch" which includes films made with their AI video generation process.

Dzine

Dzine takes a bit of a different approach in that it has all the normal text, image, video AI tools, but combines it with a simplified editor like paint which allows you to edit specific portions of the image, remove backgrounds, and add layers from other images you may have generated.

It is my personal favorite, forming a middle-ground between Replicate and Midjourney where I have more control over the images I want to generate, but blending it into a nice interface that doesn't require a lot of work digging through models like in Replicate.

Pricing also feels somewhere in the middle, where I can generate a lot more images and variations of images than Midjourney would provide, but less than if I went with Replicate. This feels like a fair compromise considering the ease of use and additional features that Dzine provides over Replicate.

OpenArt

Initially when I was searching for AI image generation and editing tools, I stumbled upon OpenArt which looked like a good fit.

They have a robust set of styles, models, and consistent character creation features that I was looking for...

However after looking further, I saw their TrustPilot had low reviews and most people complained about their subscription not cancelling, having billing issues, and poor results from generations.

ComfyUI

Lastly, ComfyUI isn't exactly the same as these other platforms since it is a software tool. If you own a computer with a decent GPU, you could run all of these AI tools on your own computer since most of the models are open-source.

ComfyUI takes a workflow based approach where "nodes" that represent an action item like rescale image, run it through a checkout, preview image, etc. that can then be chained with other nodes to produce a final output.

There's lots of tutorials online for this, although your results will be hit-or-miss depending on how old the video is since you'll have to download and find the different models that are being used by the tutorial (which may or may not be outdated by the time you view it).

And that's it!

I'm sure there's lots of other tools online that I haven't included here. A lot of them are just 1-dimensional apps doing one specific type of style transfer or AI generation technique looking to make money, but if there's anything I may have missed on this list that you think should be included, feel free to reach out!

Mindless Blob Summary:

Midjourney:
For anyone not familiar with AI that wants to generate images.

Replicate:
For those that want lots of options for using different models to get what they want done, without a subscription.

Runway:
For video generating and video editing.

Dzine:
My personal favorite —combines AI tools with a simple image editor (paint)-like interface.

OpenArt:
Looks cool, but had bad reviews when it comes to billing and subscriptions.

ComfyUI:
A software tool you can use directly on your computer with a GPU, free, but you'll have to learn how to use the tool with online tutorials.

AI generation is so cool! I'm so glad we have more ways for people to create / dream / envision what they want and bring it into reality.

Till next time!
Mindless Blob