Put your steps all the way down to minimum, and set the number of images to 3-4 to test prompts to get closer to what you want before ramping up the steps and wasting credits on badly formulated prompts. Even with steps all the way down, you will be able to tell that your prompts are in the right ball park, and only spend .2 credits / image.
@StephenCoorlas Жыл бұрын
Yes, great tip, I realized that would be a good approach after blowing through my free credits. Thanks for sharing 🤘🏼
@Bartetmedia2 жыл бұрын
Awesome!!
@StephenCoorlas2 жыл бұрын
Another great text-to-image option!
@tomschuelke7955 Жыл бұрын
I Agree.. concious Systems would have to be selfreflecting and embeded in a whole bunch of sensors to actively or inactively, get to know all current parameters of their environement. They would see the world in a total diffenrent way as we do. I remember on test, when the Black box of machine learning was opened to understand what makes the algorithm understand what a human in an image makes a human.. and the disturbing answer was " mostly the wringles on the clothes".. so it sees parameters we would ignore.. would categorize things totaly different, even when it outputs the correct answert to us. never the less hypothetically i think conciousness could some times , if out societys won´t crash due to the collaps of our environement be achieved... also feelings could evolve in such systems.. Feelings are our what we have, when we recognize an inner and outer pattern of circumstances, we do usualy benefit from ore not.. so its a pattern recognision system , we evolutionary have a predefined answer to.. Why should nt pattern recognisioin machines be able to do the same.. Never the less i am not sure i want this to happen... it would be an alien directly next to us... maybe with overwhelming intelligence at last and a totaly differing agenda of things it would want .
@StephenCoorlas Жыл бұрын
Great thoughts. Sometimes, I like to imagine all the animals and insects around us as aliens with different agendas. Birds are able to detect magnetic fields, dolphins detect range with echolocation. So maybe AI systems with their unique abilities and agendas wouldn’t be so dissimilar. I agree, it’s not unimaginable for AI to develop sentience and feelings based on interpreting data in a similar way we do. It will be very interesting to see how things develop in this field.
@tomschuelke7955 Жыл бұрын
Yes right.. but one thing in commen with all animals is that we ve been going throug the million years old school of evolution. So even if our inner world understanding apperatus may differ, they still give a glimps of the real real world which finaly no one totaly can understand.. it is and ever will be an abstraction of reality.. But! Even the most primitive animal will still econstruct reality that far correct, that it is good enoug to survive.. so its a good enough approximation.. an ape that has a totaly wrong understanding of the 3 d space in trees will simply fall and die.. Now.. how real ist the reality of AI s. If we are the selecting factor and not the harsh reality of 3 billion years of evolution.. ? And furtheremore, gametheory of interacting individuals teaches us, that friendship, egoism and altruism are a naturaly evolving system for all creatures when interacting dand beeing selected from nature as the most sucssesful ones.. How can we recreate this for Ais? Can we? Do we even have enough knowledge for this? Are assimoves roules possible?
@tombolger9373 Жыл бұрын
Hey Steven, thanks for the video, I'm a visual artist learning to use SD to refine my sketching process. Recently installed a local version on my pc and am working on training models. As an architect, I'm super interested in your approach and thought process because I work with a lot of architectural forms and themes. Was wondering if you are still working with the tool and what your experience has been the last couple of months.
@StephenCoorlas Жыл бұрын
Thanks for your interest and thoughtful question. There’s a massive disconnect between this technology and the field of practicing architecture right now. The practice of Architecture has evolved to be a very rectilinear based discipline, supported by orthogonal geometries produced by hundreds, or probably thousands of different manufacturers. My explorations with this technology attempt to push the field toward the curvilinear, but with a conscious mind of the construction techniques, which in many cases would be 3D printing, cast forms or shaped rammed earth. Each fabrication method has distinct geometric attributes, which I attempt to stay true to. Maybe that’s a long way to say, I try to think of how these forms would be built and then use descriptive words to help shape the architecture accordingly. I’d be curious to see some of your work if you’d like to share🤘🏼
@MONGIE302 жыл бұрын
I use Stable Diffision GRisk Gui O.1 running an my pc. Its free to generate images then. I am managing to run it on a gtx 1070, up to 704 x 512 and then upscale them. Getting some really good results.
@StephenCoorlas2 жыл бұрын
Nice, a saw this was an option but have not tried downloading and running yet due to the size. I’m extremely low on free space right now. For those interested here’s a link with directions on how to run Stable Diffusion directly on your own PC: grisk.itch.io/stable-diffusion-gui?fbclid=IwAR3-0wb3O2cQuujCFQh62cLugh9NQM9QtcfgktTBxXz9Y4vFtHUxW544gaU
@blinkofaneye4451 Жыл бұрын
Super fast if you make the size smaller
@StephenCoorlas Жыл бұрын
Very true I should have showcased that. I've updated the description to include that for others.
@jsusquehanna69512 жыл бұрын
!dream goes in front
@StephenCoorlas2 жыл бұрын
I didn’t notice a need for that. Does it do something specific
@jsusquehanna69512 жыл бұрын
That's how the prompt generator writes it out in the text it's always there
@StephenCoorlas2 жыл бұрын
@@jsusquehanna6951 oh I didn’t see that. I know for the Discord bot something like that was necessary, but in this new web app you can simply just type in the prompt and click the Dream button
@rajandp76532 жыл бұрын
@@StephenCoorlas yeah not necessary here only the promt no additions
@FLEXTORGAMINGERA Жыл бұрын
Dream studio went dead imo..
@StephenCoorlas Жыл бұрын
Definitely not much hype around it after the launch