Hi Group , I started using Sketchup to present my Exhibition Design projects and I would like to use some AI to render them. I also saw some walkthrough videos that I really liked.
I prefer easy-to-use AI renders as I am just started, and if it is possible, free.
What would be your recommendation ?
thanks all
ChatGPT is likely one of the best options available for free use.
Chat GPT is able to “read” my lighting design , for example ? I understand that ig the AI tool is connected with sketchup the app (or extension) will “read” the components and create a more accurate renders..
Am I mistaken ?
Tks again
well, if only, but no
AI render, the official tool, makes a screenshot of your model and runs it through their current model, at a cost.
These days they use Nano Banana by Gemini, so it’s better than it used to, but it’s still not a real rendering tool that will read the structure of your model, lighting and all.
You can check both Nano banana (gemini) and chat gpt for free, even a free account will allow you to transform your images. just upload the image, tell it you want it to be transformed into a render, and add all your positive but also (very important) negative prompts, “don’t change the geometry, don’t add elements / plants, don’t change materials…” you’ll see, they’ll try to grab every opportunity to do what they want.
in chat gpt, if you have a pro account, you can access ready-made GPTs, one of them is named sketchup renderer. it’s pre trained on this task.
overall, it’s quicker and cheaper. or it’ll look like it. you’ll have to accept a certain degree of variation, it’ll never be exactly like your model, you still have to do all the 3d work (a more detailed model gives less freedom to AI).
re-runs, ie. changing your model, and re-asking the same exact prompt will not give you a 100% accurate new image. also, if you prepare 4-5 different views, there are no guarantees that you’ll have 100% consistency between the various images.
there are some paid tools out that are built specifically for that, I know of SUapp, Veras (now owned by Chaos → enscape/vray).
My recommendation ? well, the more detailed your image is prior to using AI, the less freedom ai will take with it. if you want some very specific lighting in your Exhibition, might be worth making a “pre-render”. send your model to a rendering tool, even a free one like twinmotion, and use it to set your lights / main materials. that way you can tweak the important things in your render (lights, shadows, reflections) and only use AI as an enhancer.
and here is what chat gpt would have done solo without a pre-render in twinmotion to set the lights / mood.
off course, it added plants because I didn’t tell it not to. and the lighting is so… uniform and bland. and gpt changed the perspective, the point of view, and the size / position of the mirrors. too much freedom there. oh, and the nightlights, despite having the lightbulbs sticking out of the shades, it decided the actual light was coming from inside.
what’s the point of modelling stuff they way you want it to be if gpt then just changes it, right ? 
I’ve recently started using ChatGPT to generate renderings from just a sketch or a CAD image, and the results are often very interesting. It isn’t always perfect—and the feature seems to still be under development—but it’s genuinely useful. You can access it for free on an occasional basis, though there is also an $8/month subscription.
Here is what you can get from a simple drawing—without color, texture, or lighting effects… the AI ​​interprets it and adds all those effects :
