What I’d love to see in the (hopefully _near_ future is an AI extension for Sketchup where you can take your crude model, some reference photos of the actual furniture, cabinets, and appliances you plan on using, and have an image generator match them in a 1:1 scaled rendering. Without having to go to different sites, having this all in one program would be such a workflow powerhouse.
Hi, yes you can create super high-quality images with fine details, but the only downside is the accuracy at the moment. Even though it is possible to get close results to your original input, it is pretty hard compared to Stable Diffusion. Thanks for your feedback!
If one day, an addon is developed that can convert the drawings produced by Stable Diffusion into 3D models, it would truly be a game-changer. I've been thinking about this a lot lately because, for example, Stable Diffusion can create amazing chair models. Being able to export them directly as 3D models would significantly speed up architectural modeling.
Hi Ali, thanks for your nice comment! Totally agree, there are already lots of development happening in text-to-3D or image-to-3D. There are a couple of available tools, and I will upload a new video about them soon. They are not perfect yet, but we can see the development process.
That is far fetched and would probably take a long time to develop. Stable diffusion and the likes do not work in 3D. They simply use gaussian techniques (literal paintinq) to come up with images. It is like expecting a painter to sculpt exactly what they paint. Another issue is that generating models from images requires many images of different angles for them to be decent. Models from text might actually be feasible. Usable and editable models from AI are far fetched.
"Amazing video! Could you create a tutorial video on how to turn SketchUp 3D models into realistic renders using AI (not just rendering within SketchUp)? I would love to see more videos like this. Can you also suggest some keywords related to 3D rendering and hyper-realistic images? It would be really helpful. Thank you!"
I was really inspired by your video! It reminded me of another AI-based design tool I came across, known as Stylar. This tool has a feel similar to Photoshop, offering features like interior design styles, and it's more user-friendly. However, it's still quite basic and seems to be in a beta phase.
This is very helpful! Can you try now with the Midjourney 5.1 and explain us the prompt configurations (style, raw, etc) for interior and exterior renders. Or do two videos. One with all the prompts in MJ 5.1 to render images like this one from sketchup for interior and exterior. Trial and Error. Another one comparing the new 5.1 with stable diffusion. Can you get a realistic render for a building with this instead of using Corona for instance ? Great work!
Very interesting and insightful videos, helped me get started with my first experiments using AI image generation for my projects. Keep it up man! Also wanted to ask what your PC specs were, particularly VRAM, and what subscription package you used in Midjourney
Hey, thanks a lot for your supportive comment! So happy to hear that it was helpful for you :) I am using a laptop with RTX3060 6GB VRAM and 12th Gen Intel(R) Core(TM) i7-12700H CPU, and I can use it without any problem. I have the mid-tier subscription for Midjourney; I prefer that because you have unlimited generations.
Thank you so much this was really helpful! Midjourney seems the best! Can you actually do rendered images and do editing in Ai? Maybe edit the lighting make it realistic and add objects?
Hi Merve, thank you for your lovely comment, you are very welcome! Of course, it is possible to edit them with AI and specially add some objects like, people, plants or basically anything you can imagine. I think Photoshop Generative Fill would work best for this case, feel free to check this video to learn more about it :) th-cam.com/video/Duf7SfLuoWk/w-d-xo.html
Thank you so much for this video. Useful and straight to point😎 How do you use a stable diffusion? Subscribed? I tried it on the iPad but it doesn’t work. Interesting it’s because subscription or because ipad😅
Hi, you can check this for sketch-to-render of product design: th-cam.com/video/zSkJRNLJxTs/w-d-xo.html&lc=UgwYuWn2ZmFyrunUs1F4AaABAg I hope it helps :)
@@designinput That sounds very promising. And what about change different camera angles? I sure this will be possible in the future and I can ditch my render programs haha
@@keshams3665 It can be a bit challenging at the moment, but as you said, I am sure it will be possible to try it soon :) Haha, exactly, it is coming sooner than we imagined.
If I just want to make a 3D render more realistic but keep the elements as they are, which way would you suggest..? Maybe photoshop..? Have no idea, thank you :)
Hey, I would try it with Stable Diffusion as image-to-image with ControlNet extension and keep the denoising strength low, so the image won't change a lot but maybe it will generate it as more realistic.
this is a great video! but I tried a couple of times to insert an image from my model to midjourney, but every time he totally messed up the geometry and made it something completely different- any tips on how to "force" it into the geometry I input?
Hey, yes, unfortunately, it's not so straightforward. Try to start your generation with the /blend option and add the same image two times (don't use the exact same images, use very similar versions of them) and continue adding some prompts with "remix mode." I hope it makes sense! Thanks for your comment!
Hey, thanks for your comment :) Sure, you can access it from here: stylesourcebook.com.au And all the other tools used in the video are available here under the resources section: designinputstudio.com/turn-your-mood-boards-sketchup-models-into-realistic-renders/
Well at this example, midjurney makes the best images, but i would miss the control given from Controlnet. So i imagine, stable diffusion with a pretrained Model, for your special Architectural style, and with some prompt tweaking could do an as great image as midjurney, but perfectly folowing your sketch? I imagine this in an interface like Invoke ai and the model running totaly on your own hardware.. My company wouldnt like to make all the stuff in the net, as long as we dont know who and how our data are secured there.. But its develloping so fast... in a few month.. installig something like this on your own computer will be no more big deal
Hey, thanks for your great comment! You are right; ControlNet changes the game pretty fast; they just announced a new "reference" mode, so I feel like it is getting similar to the Midjourney.
Hi, it is not very straightforward to keep your image the same but just change the quality/style. It is possible but not as easy as the ControlNet + Stable Diffusion. I will share a video about it soon. You can try to use the image weight prompt (--iw) to give more priority to your reference image.
Hey, thanks for you comment, I really like Midjourney and Stable Diffusion because of the ControlNet. You can find out more about all the other platforms and comparison of them here: th-cam.com/video/Yfh7yb5xcVE/w-d-xo.html
You can find all the recourses for this video here: designinputstudio.com/turn-yo...
What I’d love to see in the (hopefully _near_ future is an AI extension for Sketchup where you can take your crude model, some reference photos of the actual furniture, cabinets, and appliances you plan on using, and have an image generator match them in a 1:1 scaled rendering.
Without having to go to different sites, having this all in one program would be such a workflow powerhouse.
Hmm. Mid journey really appreciate the finer details,the over mood and lighting is really good .
Hi, yes you can create super high-quality images with fine details, but the only downside is the accuracy at the moment. Even though it is possible to get close results to your original input, it is pretty hard compared to Stable Diffusion.
Thanks for your feedback!
That was fun to compare them all! Midjourney is the winner for me, i really like the images that it generates from that basic skethup model
That one was my favorite too! Thank you for your comment
Same but you need to know what your typing in???
If one day, an addon is developed that can convert the drawings produced by Stable Diffusion into 3D models, it would truly be a game-changer. I've been thinking about this a lot lately because, for example, Stable Diffusion can create amazing chair models. Being able to export them directly as 3D models would significantly speed up architectural modeling.
Hi Ali, thanks for your nice comment! Totally agree, there are already lots of development happening in text-to-3D or image-to-3D. There are a couple of available tools, and I will upload a new video about them soon. They are not perfect yet, but we can see the development process.
That is far fetched and would probably take a long time to develop. Stable diffusion and the likes do not work in 3D. They simply use gaussian techniques (literal paintinq) to come up with images. It is like expecting a painter to sculpt exactly what they paint. Another issue is that generating models from images requires many images of different angles for them to be decent. Models from text might actually be feasible. Usable and editable models from AI are far fetched.
"Amazing video! Could you create a tutorial video on how to turn SketchUp 3D models into realistic renders using AI (not just rendering within SketchUp)? I would love to see more videos like this. Can you also suggest some keywords related to 3D rendering and hyper-realistic images? It would be really helpful. Thank you!"
Hey, thanks a lot
I was really inspired by your video! It reminded me of another AI-based design tool I came across, known as Stylar. This tool has a feel similar to Photoshop, offering features like interior design styles, and it's more user-friendly. However, it's still quite basic and seems to be in a beta phase.
cool video will be using a few of these sites !
Hey, thanks a lot for your lovely feedback :) Happy to hear that!
This is very helpful!
Can you try now with the Midjourney 5.1 and explain us the prompt configurations (style, raw, etc) for interior and exterior renders.
Or do two videos.
One with all the prompts in MJ 5.1 to render images like this one from sketchup for interior and exterior. Trial and Error.
Another one comparing the new 5.1 with stable diffusion.
Can you get a realistic render for a building with this instead of using Corona for instance ?
Great work!
Hey, thanks a lot for the beautiful suggestions and comments!
I will definitely test them out and share my experiences :)
It is really amazing to see all these intersection websites. And specially there comparisons. Hope your channel get more famous 😊 Amazing content 👌
Oh, thank you so much for your lovely comment and wishes. Really appreciated it ❤
This whole video was very helpful and it has made a lot of things much easier for me , thankyou so much .
Hi, thanks a lot for your lovely comment! Glad to hear that you liked it! ❤
Can you share tips for prompt especially for architecture/interior?
Hi, sure, I will upload a video about creating better prompts soon. Thanks for your comment!
@@designinput do you have discord server?
@@itsdaffxter I don't have one right now. Should I create one? Would you be interested in it?
@@designinput yes
This my favorite Ai tutorial thanks for sharing !
My pleasure!
Very interesting and insightful videos, helped me get started with my first experiments using AI image generation for my projects. Keep it up man!
Also wanted to ask what your PC specs were, particularly VRAM, and what subscription package you used in Midjourney
Hey, thanks a lot for your supportive comment! So happy to hear that it was helpful for you :)
I am using a laptop with RTX3060 6GB VRAM and 12th Gen Intel(R) Core(TM) i7-12700H CPU, and I can use it without any problem.
I have the mid-tier subscription for Midjourney; I prefer that because you have unlimited generations.
Thank you so much this was really helpful! Midjourney seems the best! Can you actually do rendered images and do editing in Ai? Maybe edit the lighting make it realistic and add objects?
Hi Merve, thank you for your lovely comment, you are very welcome! Of course, it is possible to edit them with AI and specially add some objects like, people, plants or basically anything you can imagine.
I think Photoshop Generative Fill would work best for this case, feel free to check this video to learn more about it :) th-cam.com/video/Duf7SfLuoWk/w-d-xo.html
@@designinput thank you so much! I will check it out!
Great video, thank you! How were you able to maintain the exact layout of your picture after it was rendered with Midjourney?
Thank you! It is not always possible to keep the layout the same in Midjourney. You can try to use --iw parameter for that. Maybe that can help.
Thank you so much for this video. Useful and straight to point😎
How do you use a stable diffusion? Subscribed? I tried it on the iPad but it doesn’t work. Interesting it’s because subscription or because ipad😅
Hey, thanks a lot for your feedback and comment
Really amazing video!
Hi Jose, thanks a lot for your amazing feedback. I really appreciate it ❤
Do you have any suggestions for product design? Or, like sketch like images rather than renderings? Thanks
Hi, you can check this for sketch-to-render of product design: th-cam.com/video/zSkJRNLJxTs/w-d-xo.html&lc=UgwYuWn2ZmFyrunUs1F4AaABAg
I hope it helps :)
Hey thank you for this amazing video! Have you tried uploading existing renders and making variations?
Hi, thanks for your lovely feedback
@@designinput That sounds very promising. And what about change different camera angles? I sure this will be possible in the future and I can ditch my render programs haha
@@keshams3665 It can be a bit challenging at the moment, but as you said, I am sure it will be possible to try it soon :)
Haha, exactly, it is coming sooner than we imagined.
If I just want to make a 3D render more realistic but keep the elements as they are, which way would you suggest..? Maybe photoshop..? Have no idea, thank you :)
Hey, I would try it with Stable Diffusion as image-to-image with ControlNet extension and keep the denoising strength low, so the image won't change a lot but maybe it will generate it as more realistic.
@@designinput Thank you very much :) going to give it a shot ..!
amazing results with MJ
Hi Robin, thanks for your nice feedback and comment!
thank you so much I will try Iiked your video and I subscribed
Hey Umut, Thanks a lot for your comment
Stable diffusion so little time you allocated!
wwooooow I really love it thank you
Hey, thanks for your nice comment :)
Great video. Very helpful
Thank you for your comment and lovely feedback; really appreciate it ❤
this is a great video!
but I tried a couple of times to insert an image from my model to midjourney, but every time he totally messed up the geometry and made it something completely different- any tips on how to "force" it into the geometry I input?
Hey, yes, unfortunately, it's not so straightforward. Try to start your generation with the /blend option and add the same image two times (don't use the exact same images, use very similar versions of them) and continue adding some prompts with "remix mode." I hope it makes sense!
Thanks for your comment!
very interesting, you make some very useful content
Hi, thanks a lot for your lovely feedback ❤️
another cool video, thank a lot!
Hey, thank you so much for your lovely feedback ❤️
5:29 how do you get to the photo link on the section Lora ? The pink Icon below "generate" is not active on my Sd
wich of them is fully free ?
Hey, except Midjourney all of the other tools/websites I used in this video are totally free. And you can use Midjourney starting from 10$ per month.
thank you for Thais video can i know name website use to mood board
Hey, thanks for your comment :)
Sure, you can access it from here: stylesourcebook.com.au
And all the other tools used in the video are available here under the resources section: designinputstudio.com/turn-your-mood-boards-sketchup-models-into-realistic-renders/
@@designinput thank you
Well at this example, midjurney makes the best images, but i would miss the control given from Controlnet.
So i imagine, stable diffusion with a pretrained Model, for your special Architectural style, and with some prompt tweaking could do an as great image as midjurney, but perfectly folowing your sketch?
I imagine this in an interface like Invoke ai and the model running totaly on your own hardware.. My company wouldnt like to make all the stuff in the net, as long as we dont know who and how our data are secured there..
But its develloping so fast... in a few month.. installig something like this on your own computer will be no more big deal
Hey, thanks for your great comment! You are right; ControlNet changes the game pretty fast; they just announced a new "reference" mode, so I feel like it is getting similar to the Midjourney.
I have a kitchen. Can you help with design and drawing?
Hey i need your help i haven't any laptop so could you plz one image rendering for my college project
Hey, thanks for the comments! How can I help? Let me know please, thanks :)
Can I send you one picture on ur Whatsapp number
actually I am confuse what kind of landscape is suitable and what colour is suitable.....for house
does anyone have a tip or command so that midjopurney keeps the proportions and doesn't completely change the picture
midjourney change my pictures every time
Hi, it is not very straightforward to keep your image the same but just change the quality/style. It is possible but not as easy as the ControlNet + Stable Diffusion. I will share a video about it soon. You can try to use the image weight prompt (--iw) to give more priority to your reference image.
The link in Stable diffusion doesnt work
Can you update ?
Hey, it is working for me. Which link do you mean?
@@designinput how do you access the running tab on stable diffusion?
how long does it take you to generate imgtoimg photos?
nice
Hey, thanks for the feedback!
Stable diffusion is not working
Hey, what is the problem?
@@designinput
Thanks for reply 🙏🏼
It's now ok and working perfectly 👍🏼
these are great but the only free one is stable diffusion :(
Hi, yes that's true... But most of them have trial versions so at least you can test it out for free. Thanks for your comment!
MJ is the goat
Hey, thanks for the feedback :)
😳
Which one is your favourite AI?
Hey, thanks for you comment, I really like Midjourney and Stable Diffusion because of the ControlNet. You can find out more about all the other platforms and comparison of them here: th-cam.com/video/Yfh7yb5xcVE/w-d-xo.html
impresive, but only as a tool, in my opinion
Hi, thanks for your nice comment! Totally agree, we should only use it as a tool to help us nothing more.
Ur Exemples are ugly but thanks for tools suggesting😅
Wow this is a good video 👌🏼thanks for sharing the process. I’ve not heard or seen this process before. Possibly going to try out these guidelines ❤
Hey, thanks for your comment :) Happy to hear that!