Hey great to see this I can't wait until offline models of SDXL are more stable! I actually just today finished the last of my write ups on using a custom Stable Diffusion set up as a step in a generative architecture pipeline. My implementation was trained using models specifically for architecture and focused on pulling hard edges and depth out of simple renders of 3D models. If you're interested sometime let me know and I will spin it up on a public instance for you to play around with. AI might generate a lot of low effort garbage but as a tool for designers we are just starting to scratch the surface of what we can accomplish with focused application
Great video! thanks so much!!! However do you aso have this problem? Dreamstudio's said that the image failed and it will not use any coins, but it still used coins : ( It did this a few times already.
Sorry to hear about your bad experience. Yes, all these platforms are super new so they are undoubtetly buggy sometimes. I had a similar issue with lookx.ai recently
Yes, it is. As mentioned, all their models are open source and available on github, so anyone can run them on their local Video card. However, if on the cloud, requiring servers to run, then that server rendering process is typically paid, same as in render farms. Also, the SDXL model is currently in Beta, so Stability AI haven't released it yet as open source. They will once it's out of beta.
What are your though on the current state of architectural concept design & visualization. From what i can tell 99% of people in the industry are on the way out. There is no reason to not do what you are doing here, especially with other people already hand made images.
Everyone in concept architect and visualisation will still be in the same field, however they will have a new type of tool to assist with the design. Typical design process is never as straight forward as a prompt and getting the right response. In fact I saw a quote floating around social media that was saying something along the lines that if clients knew exactly what they wanted so they can put it in a precise prompt, then architecture and design jobs may be in danger. But that's not the case. And frankly, all the AI media scare talk is nonsense.
@@UHStudio >Everyone in concept architect and visualisation will still be in the same field, however they will have a new type of tool to assist with the design.< But it's not a just tool. It's a complete process that can provide you with a completed concept render. >Typical design process is never as straight forward as a prompt and getting the right response.< Yes. But doesn't what you doing here and all the other deisgner like you doing to the ai generatior in this types of workflow (& video) are hugely similar to what a client would do in a process of interaction with a human designer? They may require some touch-up afterwards, but it wouldn't need someone as qualified. >In fact I saw a quote floating around social media that was saying something along the lines that if clients knew exactly what they wanted so they can put it in a precise prompt, then architecture and design jobs may be in danger. But that's not the case. And frankly, all the AI media scare talk is nonsense.< I find it strange that in a video where you showcase SD using mostly a picture with only a slight steering from a promt to generate concepts that you would still be the one claiming that the prompting is somehow important and hard to do part. As a vastly more experienced designer, you can correct me if i'm wrong. But aren't most of the clients operate at a very basic level of understanding and often resorting to asking something similar to other people's design or to improve on their crude sketch, or all of this in different proportion. And the process you are using in this can provide at a rather satisfactory levels. Maybe not for a client by themself, but for a contractor with a vastly lower qualification than like someone like yourself. And if you want to claim that a client or this contractor won't have image to put them in AI of the similar level as you do, than you missed the most important part of "AI scare". The important part of "AI scare" is that if you don't have a means to protect you copyright & authorship, than you have none. It was rather hard to do photobash, remixes and fit other people work onto reference, but ai generation made it mead it order of magnitudes times easier and while yourand many other people in the same line of work use their own images & sketch to put in ai generator. The vast majority of ai generation users aren't even using their own prompt, let alone images.
![We tried to compete with AI... [AI vs. ARCHITECT]](http://i.ytimg.com/vi/N709ZrxoIP0/mqdefault.jpg)
![We tried to compete with AI... [AI vs. ARCHITECT]](/img/tr.png)
Hey great to see this I can't wait until offline models of SDXL are more stable! I actually just today finished the last of my write ups on using a custom Stable Diffusion set up as a step in a generative architecture pipeline. My implementation was trained using models specifically for architecture and focused on pulling hard edges and depth out of simple renders of 3D models. If you're interested sometime let me know and I will spin it up on a public instance for you to play around with.
AI might generate a lot of low effort garbage but as a tool for designers we are just starting to scratch the surface of what we can accomplish with focused application
Outstanding 👍👌
Thanks!
Great job!
Thank you!
Hey ,,i need your help could you plz one image rendering for my college project, coz i haven't laptop
the moment dreamstudio will also offer controlnet it will be wild
Suits it offer it now?
@@UHStudio Just logged into it. I don't see ControlNet. Not for SDXL or SD 1.6.
Great video! thanks so much!!!
However do you aso have this problem? Dreamstudio's said that the image failed and it will not use any coins, but it still used coins : (
It did this a few times already.
Sorry to hear about your bad experience. Yes, all these platforms are super new so they are undoubtetly buggy sometimes. I had a similar issue with lookx.ai recently
excellent 😍
Thank you!
how do you start to understand the shape to be able to model it?
From experience, you have to sort of mentally "unwrap" it into its simplest and most essential few moves to achieve the form.
@@UHStudio will you be modelling this one on youtube? Pretty please - your vids are the best thing on here.
Convinient Workflow but i am afraid of a "Credit system"... Wasnt the Idea of stable Diffusion to make it for everyone, for free?
Yes, it is. As mentioned, all their models are open source and available on github, so anyone can run them on their local Video card. However, if on the cloud, requiring servers to run, then that server rendering process is typically paid, same as in render farms. Also, the SDXL model is currently in Beta, so Stability AI haven't released it yet as open source. They will once it's out of beta.
Dreamstudio is free or not
You need to buy credits to use it, as it uses cloud GPUs whose usage tends to be on expensive end of things as far as cloud architecture goes.
@@UHStudio do you know a workaround for non-creditcard holders? is there like a prepaid temporary credit card service?
What are your though on the current state of architectural concept design & visualization. From what i can tell 99% of people in the industry are on the way out. There is no reason to not do what you are doing here, especially with other people already hand made images.
Everyone in concept architect and visualisation will still be in the same field, however they will have a new type of tool to assist with the design. Typical design process is never as straight forward as a prompt and getting the right response. In fact I saw a quote floating around social media that was saying something along the lines that if clients knew exactly what they wanted so they can put it in a precise prompt, then architecture and design jobs may be in danger. But that's not the case. And frankly, all the AI media scare talk is nonsense.
@@UHStudio >Everyone in concept architect and visualisation will still be in the same field, however they will have a new type of tool to assist with the design.<
But it's not a just tool. It's a complete process that can provide you with a completed concept render.
>Typical design process is never as straight forward as a prompt and getting the right response.<
Yes. But doesn't what you doing here and all the other deisgner like you doing to the ai generatior in this types of workflow (& video) are hugely similar to what a client would do in a process of interaction with a human designer? They may require some touch-up afterwards, but it wouldn't need someone as qualified.
>In fact I saw a quote floating around social media that was saying something along the lines that if clients knew exactly what they wanted so they can put it in a precise prompt, then architecture and design jobs may be in danger. But that's not the case. And frankly, all the AI media scare talk is nonsense.<
I find it strange that in a video where you showcase SD using mostly a picture with only a slight steering from a promt to generate concepts that you would still be the one claiming that the prompting is somehow important and hard to do part.
As a vastly more experienced designer, you can correct me if i'm wrong. But aren't most of the clients operate at a very basic level of understanding and often resorting to asking something similar to other people's design or to improve on their crude sketch, or all of this in different proportion.
And the process you are using in this can provide at a rather satisfactory levels. Maybe not for a client by themself, but for a contractor with a vastly lower qualification than like someone like yourself.
And if you want to claim that a client or this contractor won't have image to put them in AI of the similar level as you do, than you missed the most important part of "AI scare".
The important part of "AI scare" is that if you don't have a means to protect you copyright & authorship, than you have none.
It was rather hard to do photobash, remixes and fit other people work onto reference, but ai generation made it mead it order of magnitudes times easier and while yourand many other people in the same line of work use their own images & sketch to put in ai generator. The vast majority of ai generation users aren't even using their own prompt, let alone images.