i used diffusion for the first times yesterday and today. running some of my models through it came up with identical styled renders to yours. however i ran a well made model of a backhoe with the interior photorealistic style and it turned it into a plush pillow version of a backhoe, was actually pretty neat 😅
Excellent overview Carter! This is mind-bending for many. I think a good follow-up video could be a workflow where you bring in the rendering output to Photoshop and iterate specific elements there. This is so promising.
The problem I am facing with diffusion is the consistency of the renders. I am not able to get the same render style in two different angles. I can have only one money shot image. Is there a way to get it?
No other AI tools out there being able to do that so far. Even the paid one. The purpose of those AI tools ( atleast for now ) is just to give you some idea for your project. Imagine if AI can do like you said, maintaining the same design, props, lighting etc even you change the angle of view... you know what, Vray, Corona, Enscape, Lumion, TM will go bankrupt.
It’s certainly not to a visual level or performance of something like midjourney, hopefully they get there eventually but have to remember midjourney has years on this feature
Unfortunately that is one of the problems right now, you can get similar results if you keep the same style and prompt and just change the angle, but still wont be the exact same
Great video I have a question I am new to Sketchup and I wanted to know do you offer private zooms where I can share my screen and you walk me thru the issue I am having with a cityscape render
I do not currently offer that, I still work 40+ hours a week on top of this channel so my time is very limited atm. Hoping I could in the future though
SketchUp Diffusion is based on Stable Diffusion (developed by Stability AI). Stable Diffusion's underlying dataset is LAION-5B. LAION-5B contains links to images of CSAM and private medical photos. Taking that into consideration, along with the copyright infringement lawsuit against Stability AI, Getty Images suing Stability AI, and the fact the dataset itself is filled with [links to] images & art taken off the web without the consent of those who originally posted/own said images/art...I refuse to use any generative-AI software. There are far too many ethical concerns I can't turn a blind eye to. I strongly recommend taking the time to learn how generative AI is currently hurting creative people (and people in general) before using any of it. I'm posting this here because many people out there aren't aware of the damage this stuff is doing.
i used diffusion for the first times yesterday and today. running some of my models through it came up with identical styled renders to yours. however i ran a well made model of a backhoe with the interior photorealistic style and it turned it into a plush pillow version of a backhoe, was actually pretty neat 😅
Thank you for the overview. It certainly adds speed to photorealism. Great tool!
It certainly does, I think it will be a program to watch over the next few months to see what is developed
Excellent overview Carter! This is mind-bending for many. I think a good follow-up video could be a workflow where you bring in the rendering output to Photoshop and iterate specific elements there. This is so promising.
Thanks Henry! Super excited about what’s going to come out in the AI space in the next year or so
The problem I am facing with diffusion is the consistency of the renders. I am not able to get the same render style in two different angles. I can have only one money shot image. Is there a way to get it?
So far I haven't really been able to figure out the consistency aspect of this program.. it's really frustrating
No other AI tools out there being able to do that so far. Even the paid one. The purpose of those AI tools ( atleast for now ) is just to give you some idea for your project.
Imagine if AI can do like you said, maintaining the same design, props, lighting etc even you change the angle of view... you know what, Vray, Corona, Enscape, Lumion, TM will go bankrupt.
Is there a way to make it respect materials instead of just geometry?
I haven’t really figured out a way to get it to respect materials so far. The program is so finicky still
Have not had any really good results at this yet. Pretty basic compared to Mid-journey or Dahli? Hope to get better at it.
It’s certainly not to a visual level or performance of something like midjourney, hopefully they get there eventually but have to remember midjourney has years on this feature
@@DesignItGreen 100% more expensive.
What happens when you have multiple angles/cameras, will it keep same deisgns, or you have 3 renders with 3 different objects in it?
Unfortunately that is one of the problems right now, you can get similar results if you keep the same style and prompt and just change the angle, but still wont be the exact same
Great video I have a question I am new to Sketchup and I wanted to know do you offer private zooms where I can share my screen and you walk me thru the issue I am having with a cityscape render
I do not currently offer that, I still work 40+ hours a week on top of this channel so my time is very limited atm. Hoping I could in the future though
hi! i wanna ask about the plan you use. does the Go package can do it like this too or it have to be the Pro / Studio package? thank you
I unfortunately have the full pro license.. crazy expensive, but I heard the iPad studio version is great
Thanks for sharing...
Thank you!
SketchUp Diffusion is based on Stable Diffusion (developed by Stability AI). Stable Diffusion's underlying dataset is LAION-5B.
LAION-5B contains links to images of CSAM and private medical photos. Taking that into consideration, along with the copyright infringement lawsuit against Stability AI, Getty Images suing Stability AI, and the fact the dataset itself is filled with [links to] images & art taken off the web without the consent of those who originally posted/own said images/art...I refuse to use any generative-AI software. There are far too many ethical concerns I can't turn a blind eye to.
I strongly recommend taking the time to learn how generative AI is currently hurting creative people (and people in general) before using any of it.
I'm posting this here because many people out there aren't aware of the damage this stuff is doing.
AI is essentially a corporative thief, indeed, stealing the work from humans