In the wake of Nana Banana Proband and all the coming even better AI models:
Shouldn’t we have a dedicated thread especially for discussing AI related issues?
what’s the use of AI enhancer, when one can easily do that with Gemini 3 (NB pro) ?
why render at all, when you can genérate a picture out of the preview window with a realism nobody will reach in Corona or other engines?
.. so Corona Team - we could have a Special thread to talk about all that!
Or?
I earn my money with ArchViz and i am threatened and enthusiastic at the same time.
I think this is a very good and timely opening post. The topic of AI, especially in combination with tools like Nano Banana, is becoming increasingly relevant for architectural visualization. It’s worth seriously considering that traditional render engines may play a much smaller role in the future, and that even classic CPU/GPU-based rendering might need to be questioned. The more realistic scenario, at least in the near term, seems to be a close interaction between AI and existing workflows rather than a simple replacement. Talking about this now is important, because it gives us the chance to understand and prepare for these changes early instead of reacting once the shift has already happened.
I think the future needs to offer systems that inherently understand polygonal geometry from the inside out, rather than treating it as a static surface. Based on references, such systems should be able to add elements, materials, textures, and lighting in a context-aware way.
In that scenario, proxies would no longer just be lightweight stand-ins for geometry, but placeholders for AI-generated inlays that are resolved dynamically. This would represent a fundamental shift in how scenes are constructed and visualized.
I admit that Nano Banana Pro got me to be freaking out for a few days.
The whole situation is hard to comprehend. I’m still not sure if ai will be just a tool added to the workflow or a complate change of paradigm in Archviz in the next 5 years.
It’s going to be a ride.
For now, it saves me time by creating textures and replacing skies. Not much more than that if I want to retain the control of the scene and changes from a client.
May be for a last minute change of anything on an image…
Regarding ai tools offered by Chaos, I don’t use any.
I think the only way I will exit the Photoshop workflow is if Chaos or anyone else creates some sort of ComfyUI specifically made for compositing and editing, that allows for selections and to use models like Nano Banana pro.
The role of AI in 3D visualization exists in a space of simultaneous inevitability and ambiguity, where innovation feels less like a decision and more like a background condition. While AI is often praised for accelerating workflows and expanding creative possibility, this acceleration raises questions about whether speed itself has become a substitute for intent. When outputs arrive instantly, the boundary between exploration and automation grows increasingly abstract.
Supporters claim AI enhances creativity, though this assumes creativity is a measurable quantity rather than a situational outcome. At the same time, concerns about visual homogenization suggest a stability in aesthetic language that may never have existed. AI does not so much flatten originality as redistribute it across an opaque network of prior references, making authorship feel both collective and oddly absent.
Ultimately, AI in 3D visualization is neither a solution nor a threat, but a persistent presence. It invites engagement without clarity and progress without definition, encouraging a posture of informed uncertainty that feels productive precisely because it resists conclusion.
Your post made me think several things: First, I’m not sure if you wrote that or was it ai. And second, “Ultimately, AI in 3D visualization is neither a solution nor a threat, but a persistent presence.” this, I agree with, which may hint at a complete change in how visualization is done, forever.
The first point makes me think that if I don’t believe what I see on facebook, instagram, youtube anymore, why would someone believe a rendering anymore. Architects will need to ask the client to please trust that it is accurate, sort of.
Which takes me to the second point: buyers will expect more than renderings in order to buy a concept.
In the end, we create images to communicate. But comunication is changing. The way I comunicate with my daughter’s teachers on emails feels strange now. I use chatGPT and they use chatGPT. So we’re not really talking to each other. It’s weird and cold, but with a strange calm.
Before we get into a philosophical discussion about why humans create artefacts and what role the development of tools plays in this (and where exactly the creative aspect lies) -
a short recap:
Around 40 years ago, I had my first lectures in Descriptive Geometry at university, and nearly all tools have changed since those days when my hand rendered the perspectives.
C4D came along in ~95 (yes, before 3DS) and we became renderers!
30 years ago.
We hated the original C4D render engine so much that Vray was a godsend.
20 years ago.
With Corona, we thought for the first time, ‘They understand how architects think and want to handle the rendering process.’
10 years ago.
So.
In those 30 years, the artificial distribution of light in all those programmes got better – but it still doesn’t look real.
We have spent endless time tricking every engine to behave like a normal camera would do.
Now,
with Gemini 3, you have that camera!
So the question for me is:
How should Corona integrate AI? What tools do we need? Which other engines guide us?
D5 Render – they at least try without fear – light presets, style presets, etc.
Integrated node-based system with free model selection?
First, it ain’t solely Google’s “Nano Banana”.
Then, consider privacy and IP (Local VS SaaS).
Lastly, “the medium is the message” – in-formation – the psyche, the mental picture, the society of the spectacle…
In this world of CGI, an essential difference between machine simulations (predictive rendering, visual prototyping, …) VS hallucinations (Illustrations, AI visuals) is now becoming clearer.
If interested in theoretical aspect of visuals, build on: “An Essay Towards a New Theory of Vision (1709)” by George Berkeley.
In regards, it helps to know Aristotle’s & M. Kant’s texts on art & aesthetics (visual judgement).
