I've seen a lot of chatter in the Archviz community regarding A.I.. Forums, Facebooks Groups, and Discords are full of archviz artists/3d modelers that are terrified A.I. is here to take their jobs. In general for what we see in Archviz, A.I. engines mostly work by taking an image of a model the user uploads, and then generates a batch of renderings based on the prompt, or lists of words related to the hopeful solution that the user inputs.
So having played around with a couple of these A.I. engines, I wanted to quickly put this out there: YOUR JOBS ARE SAFE...for now. Here are a couple of reasons why:
- Accuracy - As it currently stands, I haven't seen an A.I. platform that can accurately depict what the user would need to illustrate; something real in the built environment. Sure, these engines can pump out some wildly creative conceptual shapes and images, but they aren't depicting anything real. At least not anything in the way it would be constructed in the field. Not by a long shot. Archviz is a communication tool. Communication that, even at the conceptual level, depicts more than what an image and prompt can currently illustrate.
- A Complete Project - One limitation that our friend, Adam Muir-Hails, pointed out is the inability of these engines to capture a cohesive set of images across an entire project. Currently, these engines are just editing an image based on user input; not affixing plants, entourage, etc. to a fixed point in the model. Also, as we already discussed, these engines lack accuracy, so it would be virtually impossible to change a camera angle or view and get 2 images that looked like they are the same project. With the entry level into some of the popular render engines being pretty low, the world of archviz is more accessible now than ever, allowing users to build out projects quickly and accurately, while still having the ability to get cohesive renders from multiple views. So unless you are only producing one image for a project. A.I. rendering just doesn't make the cut.
- Last but not least, the Human Touch - a lot of what draws us to art is the communication of emotion and experience. This is no different when it comes to archviz. Some artist are better than others at capturing these feelings into an image, sure. But while A.I. can create vivid conceptual images, it lacks the ability to tell a story quite like any of us could. You could load the image down with prompts in hopes to paint the scene in your head but it will never be the visual storyteller you are. We can relate to our clients as visual storytellers because we are all part of the human experience. A.I. can't be expected to output a human attribute it's never experienced and could "relate" to. If and when it can, there will be bigger problems than it taking over Archviz :)
None of this is to say that A.I. doesn't or can't have a place in Archviz. Even now, from generating light settings, raytracing, smart selection tools, and identifying materials, A.I. is already hard at work making our workflows more efficient, effective, and user intuitive. I'm excited to see how these tools evolve and how creatives will push the envelope of working with and embracing AI, instead of being fearful. Hopefully this poorly written article will help ease your mind a bit. Keep working at your craft and we will all be fine. The robots haven't won...yet.
