AI Renders in SketchUp Are Here! 

Hi everyone! Let me know what you think about AI in SketchUp, but also just in general! Are you using it? How do you feel about it? Let's have a (civil) conversation in the comments! :)

I think this could be useful especially when you are still doing conceptualization and experimentation with the right materials to use.

We are creating a lot of renderings of our work for our clients. But they need to see what they get, not some stylized AI mood board. However I like the idea of clients being able to create their own ideas and then have it handed over to us to see if we can actually build it. It would take a lot work off of our shoulders creating images for them, as we figure out what they want. So I see AI positive in this regard.

If we are using a free trial, I adjust the sliders of prompt importance or difference from the original every time I click on the render button. Does it count from the 30 numbers of renders every time I tweak the slider?

i think the best use of AI in skp 3d would be a texture / material bases or as proxy generator for a render engine like trees, cars and furniture, but if it gets good enough details generation in models with freedom to control selection.

6:30 what plugin did u use for that original texture function? Thank you!

I, actually, researched this as well last week. I like that Arko hooks directly into SKP and, i believe, is the only one to currently do so directly but its output quality isnt up to par to its competitors which outputs much more realistic renderings. These are lower quality more SKP style output. The category is useful for ideation now but will certainly be more useful in the future. These kind of things will displace people and be useful as tools to those that wield it. There is a youtube clip of 3 actual architects going up against an AI for rendering, written paragraph and (i forget 3rd category) and randomly placed entries on a forum (without saying which was which) to pick and the architects lost on all 3 categories (not to mention the time difference of the 2 sides). Its notable that the AI was, and is, not currently able to really process actual architectural requirements or logistics or really assist in a worflow beyond basic ideation imagery but that all can easily change in short order and it shouldnt be too difficult to have it learn rules, regulations and requirements plus integrate design principles and art principles and logistic principles and be able to siphon through all this data as any database can and output it into usable functional ways far faster then any human. Computers have already become far superior to humans on all sorts of specific things and now they are becoming smarter in more aggregate and all encompassing ways. This will continue.

What other rendering software would you suggest, I don’t want to spend ages adjusting lighting etc. I juts want a he render of the SketchUp models I create. I used to use twilight as it was simple to use.

Interesting Video. Hard to see its implementation commercially at this point in time. A more useable adaption of this technology down the line might let you highlight/mask certain areas of the image where you want to see variations or brainstorm ideas.

Yeah, I pretty much on the same boat as you, wether you like it or not, technology will keep advancing. Also, for this specific ai application, in my opinion it's good for brainstorming kinda thing, but I'll be using other rendering application for my realistic render.

after i download arko ai i cant find it anywhere even when i press install extension its no where

You made some very good points. It may work well for conceptual drawings, but it is basically an averaging of the information it has.

I agree, as a rendering tool, right now, this tool doesn't look like something I'd use - I'd want something that automated the stuff like lighting etc, as you said. I did find the option of changing the style completely (in the original pic) it was interesting. This will get better as it learns, and maybe there won't be as many weird iterations. AI in general, I find, can be fun to play with - you can put in your prompts and see what comes of it. The AI may come up with an option or iteration that I hadn't thought of, which I can then incorporate into my design. It can be fun to throw in something completely different to see how the AI will interpret your prompt (I dunno, maybe Edwardian manor with 80s revival on Saturn) and see where it goes from there. As a brainstorming / imagineering tool, I think it's amazeballs. For "serious" research, pinterest / industry mags are probably the way to go. When they (AIs) learn how to do the rote stuff like tweaking lighting and focus etc in renders, then they can be used for that. I think they'll be a fun addition to the design world, and the world in general. Maybe they can add some serious benefits, maybe they can help us to lighten up and try different things. Either way, they're here to stay, we may as well teach them to be what we want them to be!! 😁🦄

downloaded and install but nothing show up in my folder and also sketchup extension manager couldnt find anything.

What are the correct settings to achieve the right scale in vray for Oculus quest 2 renders. I'm having an issue regarding the scale of model. i.e. the near objects apear nearest and far objects are being displayed to far away in the same image , say object A and object B located 5m and 20m respectively. (Actual distance of object A- 5m displaying 2.5 m from eye in Oculus quest 2. Actual distance of object B - 20m displaying at - 35m to 40m away from eye in Oculus quest 2.) I'm using SketchUp pro , vray 6 - stereosperical panorama.

The best architectural AI renderings I have seen so far have been from Midjourney using just a simple image or sketch of a building to get initial stylistic renderings. This plugin with such a detailed model doesn't seem helpful. I would love to see an AI render plugin where you assign prompts to individual materials. So for the dozen or so materials you have in a model, you could have prompts for the type of stone, or type of wood, etc. then give prompts for landscape. Then a separate prompt for lighting or setting. Essentially accelerating the process that you would usually use in a typical rendering plugin and saving you all the computing process that realistic renderings. Also, if you were able to re-render the scene through the plugin while only changing one material and the rest of the renderings stayed consistent would be great.

can you do more videos of sketchup to unreal engine i wanna know more abt how to use it for interior design

Instead of knowing that AI will produce weird from the input image we give , there should be AI where it can render a low res mildy detailed render to high res qualiy render that way it's benefitting like actually what AI is supposed to deliver instead of weird diffused pixels. Also with that technique one doesn't needlessly spend hours detailing and post processing and get paid less instead he can go to AI and give a low res render and AI gives a high res and we too can charge more for that and save time and fire 3-D Artists

Vray rendering stuck at compling geometry....what to do.?

It seems to me to be a simple way to see it rendered and I like the option of not having to set up a model as that has always been the "time" consuming part of design, and fast. Not as realistic as you can make from other programs. Still a great option. Rendering for me is at the beginning of a design project and most of my clients have really wanted them to be as realistic as possible and is always changing with each render, so I spend a lot of time "setting" up and changing to meet those needs. I really enjoy that part. This will be a good addition to our options to render. Great job Justin on the video! Have a great day!.

not run with Sketchup 2019?

The best way an AI based render engine could work for architects would be if we could apply low resolution material / texture and the output could look a lot like PBR material based output and a separate controls for lighting. This gives architects / archiviz more control over the renders, and the AI engine is more focused on producing high quality renders based on specific defined parameters / prompts , rather than provide design suggestions, that could be another AI based service in itself.

Hi and thanks for the vidoe , ive try it and its really will not give you what you render as it shoud be , but you can you can use it to give you an idea , andits hard very hard to control, for example if you hit the render time after time itwill change the render image evreytime and ithink something like this its not like the renders we know , sorry for my bad English

Please make a video showing Sketchup colobration with Open AI midjourney. What propmts can be used? How to gain a better control?

This would be a better tool for landscape design.

It gets a little crazy. Not sure if it is ready for prime time!

Interesting. I wonder when some AI will appear that does the legwork on hillside multiple developments and the driveways, garden beds and edging and contours

In the time i spent find out the right prompts and render it endless to find the image i am searching for… in this time i am faster with enscape or other tools like that.

I wonder if Sketchup uses a similar model to that of diffusion models like Bluewillow and others. I can say that the AI made ease of use level up.

Now if they can animated my model, automatically created fully dimensioned floor plans, elevations, sections etc from that model, and let's just through in a perfect rendering of the model....then we are going somewhere. But this my friend would be a huge time waster just tinkering around with it, and not getting exactly what we need. Yes- in the future I am sure it will improve. Until then....its a big fat NO for me

this looks terrible , this is miles away. Why would they possible have this as a paid extension. They should have this free and let the community contribute to its data base.

what your gpu ma man?

None of the links work for the Mac free prior

Hey SKETCHUP ESSENTIALS, I have did tons of research on this stuff, the best way is not use arko ai, it doesnt have any flexbility. I downloaded the ai model which is diffusion stable model directly on my harddrive and let my gpu run the model with pre trained models to produce much better result.

The main problem I see with this instance of AI rendering is that it's somewhat random and until get a result that suits me I will have made possibly no progress. It's not like the more time I spend the closer I get to the final result. This uncertainty I believe is a deal breaker on professional work.

I bet Picasso wanted those for his interior design.

You can just do this yourself with dall E

I don't think AI can replace human creativity 100 %, it can help, but replace

Ai seems to be future....seems like will be upgrade fully in upcoming days...hoping for better future 😊

Whats the point of having a "render" app or engine that changes the outcome to what the AI thinks is best. To me this defeats the reason for having a final render of "my" work turned into the "AI's" work....seems pointless.

can you export it to Unreal Engine 5 and make it better?

this is going to be great but it needs more trials overall i guess to get very clear just like mid-journey, so at this early stage to ask for $30-$40 is not fair. They should give the ai more trials from the public to learn faster

Well... It's useful if you want to generate ideas for your spaces or exteriors. Especially if one is having a hard time deciding what exactly to do with space or so... But personally, I can't depend on this for my final renders. It terrible for final renders. + It's not realistic.

It’s a similar problem that AI is not good at drawing hands, it can get the texture right, but not always the right number of fingers.. If this tool randomly changes the spatial qualities of the model it is not very helpful in architecture (yet).

Thanks for the updates but I have to say that this is too much guess work for my liking, we go to 3D for practical specifications that has to be built in real life, i do yet see the constancy for practical applications.

still in early stage . not for serious works or for my clients for sure . .

I think it would be better to take an image of the scene one sets up and use it in AI for now....

This add-on is really bad. I totally agree with you Justin in that automating light setups and maybe some AI generated background scenes and environmental filters would be helpful, but otherwise this just promotes laziness.

I would rate this a 5 out of 10. It has a lot of potential but the output is too rough and is full of errors as you pointed out.

I'm not even sure if I woul call this AI. The results are somewhat clunky, random and disorganized. More like it's just getting a simple approximation of what you told it to seek, and incorporating it into your scene without much consideration or analysis. It's an automatized rendering process... But I'm not sure if I would call it "AI". They probably did so because ChatGPT is super hot right now Maybe, like you said, if you're trying to get ideas for some concepts, this may help you out, but honestly, it's a lot more inspiring to see professionally done works in a site like pinterest, and it gives you a lot more to work with than just the fact that it's the exact same scene Over all, there's room for improvement, but as is... I hardly see this tool being useful at all. Would like to see how it deals with a lot of control of an empty environment though. Does it leave it empty or tries to populate it? That would work better for concepts

Like it or not , in the end we have to learn AI, inevitable.

this 'plugin ' is of no use to me at this stage. As Justin says , I don't want any change to structural element. I don't believe this is really a 3d plugin if it just manipulating a 2d exported view, and pays no attention to layer or component names within the model . What would be useful for me is - 1) lighting effect , 2) material choice / colour scheme 3) some small furnishing, like placement of object on table etc. 4) some people , character. In short - i think the AI needs very specific parameters to be helpful as renderer. Thanks for another great video though !

U are Sketchup GOD

I find it quite USELESS for the price

A.I Should be A.S for Artificial Stupidity

i will never even 50 cent for this software

why are they charging for this. It dosnt give good results

Yikes! Not sure there's much practical use for that, maybe if it was free of charge then people might use it to create very loose concepts

AI is great, what concerning is, are the people that will use the AI to replace a man's job.

So bacicaly MidJourney AI for aketchup.

not worth it!

How is this even useful if the render does not resemble the model? Hours wasted modeling? I don’t understand. Plus what’s with the MC Escher results? $39 mo.! No gracias.

It's a messy tool

Eu pensava que o I era de inteligência...

The premise that it plugs into a natively 3d program and instantly takes a 2d screenshot to work from is absurd! Seems a gimmicky assemblage of existing free AI tools and you lose all 3d function. The quality doesn't compare to any relevant paid versions like stable diffusion so pricing seems out of range.

People are paying real money for this!? 😂

Hummm... not for me. This type of AI is quite far off from the OK level so far.

Not so good AI can not took up all the things

F-subscriptions… blender is the way to go moving forward.

Expensive for not accurate render

This trend is too exaggerated

To me it is a bad script. at least for what i need. I need a render nott materials and crazy ideas.

waw, not worth.

It’s so badddddd😂

Seems finicky

I hate the idea of AI, it's taking away the skills we have as archviz artists. I just don't see how anyone working in the industry would think this is a good thing.

Stable Diffusion is far ahead of these guys and is free. I can't use the output this software generates and certainly will not pay for it. Too little, too late.

*AI will either (1) put people out of work or (2) diminish their ability to make higher wages - - simply bc the AI is DOING the work.* *Moreover - - I used to think AI could NOT be CREATIVE - - but after experimenting w chat-gpt + seeing this video I’m not so sure I still believe that. “Creative” based on a vast storehouse of known parameters ... yes. REVOLUTIONARY or STROKE-OF-GENIUS CREATIVE - - no. Thanks for the great video Justin - cheers!*

like a child who doesn't listen to words ... she should grow up a little more

I generally hate the current trend of implementing Ai in every form of art. It's a form of a "random generator", lacking the emotion and reasoning a true artist (architect, musician, painter...) pours into his work.

seems terrible

True rendering is artistry. This is generic crap.

This tool looks very messy

They're so confident to ask money for this terrible tool 😂

Yikes - This plug-in needs a lot of work.

Think they have a lot of work to do on this before they start charging for it. Looks like a hot mess at the moment. 👎

is these a paid plugin?its a scam

This extension is useless. It’s still conceptual. Not worth it’s price.

Terrible for a payed tool. The makers are selfish and greedy and are looking for a quick buck from AI. they need to do more work and make it free till they can fix or make it work better. 25$ my as*

This absolutely sucks. Stable Diffusion and Midjourney are way better. I’d never use this.

These are far too cartoonish. Stable Diffusion does a lot (A LOT) better and it's free.

pretty useless

Waste of time money

lol everyone posting videos on this joke of a tool. AI tech is trickling down to Skeychup slowwwwwww

In the time i spent find out the right prompts and render it endless to find the image i am searching for… in this time i am faster with enscape or other tools like that.