This is a long delayed update: a whole world hapenned since my last post on tools and things in the AIArt world. To make things simple we can say that Midjourney is AIArt today. Its images are photorrealistic, incredible and very far ahead of its main competitors in the space like Dalle or the new AI form Adobe (Firefly).  Tools are also completelly available now. 

AI IS EVERYWHERE

Microsoft integrated a new version of GPT with its search engine and the resukting search with Bing Chat is ratter remarkable. What it does have to do with AiArt? Not only this new GPT can understand images you show it, talk to you like a person to help you with your research and even write flwaless text in many languages, it also offers to generate images when you need. It does so using a version of DALLE-3 that is integrated on Bing Images. Using Edge+Bing is today a much better search experience than Google. And for the first time in more than two decades Alphabet (the Google main company) is playing catch up.

To match Bing, Google accelerated the deployment of its own AI called Bardo. But the move kind of backfired since it is clearly laging behind GPT in capabilities. Stakes are too high to fail so The Big White renewed its efforts and integrated Bardo in the Google search engine. To match the image generation capabilities they partnered with Adobe to integrate Firefly in Google Images and Google Chat which is how they called their new serach interface powered by Bardo. They are catching up and they still have the advantage of inertia: people is so used to Chrome and Google and Microsoft still holds such a bad name from the nineties and early 2000´s that many people are not aware of the incredible new AI capabilities and  not leaving the now clearly inferior tools they are used to.

Both companies promissed to integrate  AI in their office suits , i. e. Office 365 and Google Docs respectivelly.

So everybody have an image generator with advanced text understanding capabilities at the tip of their fingers now, and for free. How this affect AIArt and the AiArt community?  Well... not much.  Firefly and Dalle are rarelly used for anything but first ideas and images for other purposes for atwo reasons: control and Midjourney 5.

• This image was not created with Firefly, DALLE, in fact not even Midjourney. The texta about tools  maybe explains better the reasons why not everybody uses only Midjourney and how is the tool scenario today.

When I first started there was a huge prejudice against using AI to make images. There still is some, but it changed a bit. Many people are comparing the impact to the arrival of photography. As an analogy some sugested a new name for the art form synthography (we dont call photography "camera art"), other names like generative art, prompt art, AI photography and simply digital art are also used in some cases, although AIArt is still the most common widelly used moniker. With the arrival of Midjourney 5 also a clear trend, a style school if you may,  installed in the community: insane photorrealism.  It is this style tthat is more commonly called AI photography. Many people of the photo art community were fast to assimilate the new tools, there are some expositions and galleries allowing synthography pieces and I believe the barrier of prejudice was a bit worn out as more people realized hwo the tools work and their limits.

It helped also that some AI powered tools (like object erasing or background removal) were already widely used by the digital art community and photographers in general.  Some prompt enhanced versions of these tools are now available and incorporated on the most recent versions of the Photoshop family of tools. So digital photography was the first "traditional" visual art to embrace synthography. Digital and 3D concept art are begrudgingley following suit. The  main reason is that digital editing is still crucial to obtain really astonishing results in synthography. You can be a prompt jockey but, without composing and image prompting, at least for now, there are  clear limitations for what you can achieve.

With more and more people realizing that there is technique to achieve amazing results, more people find out that there is an art to AIArt after all.

There is no doubt that, right now, Midjourney is the state of the art in AI image generation. Its images are perfect and it achieved a degree of photorrealism no other tool alone was able to.  Also, unlike DALLE and the openly available Firefly at google, MJ allow important features for image creation and refinement like impaiting, outpainting, style transfer, pose transfer, object and background removal, object description, etc... all this coupled with a interface that allows more fluid like prompts, closer to DALLE and Firefly ones. There really is no way of getting the same results of MJ without resorting to many, many tools. Even Adobe is playing catch in this aspect.

So why use anything else? Well, most people dont. Synthography is perhaps more than eighty percent made om Midjourney right now, and that is the main reason to use other tools: it is a black box. It also so user friendly it created a whole separate community that is not used to make things in other tools. The experience does not translate. So it is a fenced playground. Pretty much like iOS and Mac were in the past. If you are experimenting with video or 3D or anything like it, you will probably have to rely on other tools and MJ is not gonna be a huge help besides the definition key frames.

Midjourney is also insanelly prude. Making a piece with it is like trying to make art with a "Sister Josephina" over your shoulder constantly checking if the skirts are bellow the knees hight. Those are reasons some users add other tools to their workflow while still remaining mainly MJ users.

Those who dont use Midjourney usually follow two "paths":

• If they can make a local PC with strong graphical capabilities and dont mind working always in the same machine, most people install the lates version of Stable Difussion, with a interface (usually Automatic 111) and add models and other tools to the interface in order to make it fit its needs. The are also a couple Colab notebooks online with this setup and a few more used models and pipelines (for instance Controlnet, outpainting, inpainting and many LORA´s. HuggingFace has tons of models and tools you can add and personalize your setup. This is the more powerful alternative to MJ right now. But is very hard to make it work properly (although there are many good video tutorials).

• They can use a digital editing tool and many different online AI tools to try to get the same functionalities. It is never as powerfull and it is slower than both the previous alternative. Still recquires a smaller level of commitment and may be a good starting point. I can not say that this is the best alternative, but it is mainly what I do. So I will present some of the tools for AI generation I use here. This will be an update (quite an update) on my last text about the same topic. Since this is a large topic there will be a post about it.