#🌶|off-topic
1 messages · Page 6 of 1
NFTs are a scam
Will Stable Diffusion bring about a LLM like ChatGPT without moderating it to the point of being just another bot? Coz ChatGPT has been restricted so much in the past few days.
do you speak spanish? te explico
usando el comando / dream en los canales dreambot desde #1047760914008522782 a #1047760279267721236
"/ dream" es todo junto
tambien con un comando pero dejame revisar
algun efecto de cyberpunk pero que se vea 3D... como debo colocor
al final del escrito ?
para que quede con la mejor calidad
yo vengo de midjourney
usando image_init y ahi aparece una opcion para subir la foto
yo usaria "cyberpunk, 3d render, illustration"
todo se veria asi "/ dream prompt:cyberpunk, 3d render, illustration image_init"
al agregar image_init aparece esa ventanita para agregar la foto
siva mi pana gracias dejame ver
este resultado me dio usando la foto de una vaca, no muy bueno porque no describi la imagen, solo puse los adjetivos que te dije
este salio mejor al describir lo que habia en la imagen
me salio muy basica won
trata describiendo la imagen tambien tipo "man wearing a hat, futuristic city background"
lo bueno es que son generaciones gratuitas y puedes intentar varias palabras
no tiene un limite ?
ilimitado hasta nuevo aviso
es una forma de mejorar el modelo en comunidad
hi I'm new
hii welcome
Does anyone know any depository/list/video that has easily summarized and broken down how AI art actuallly works in an ELI5 way, or at least just clears up the misconceptions?
Just wondering, as a community Ive been following has been on the verge of banning it for "ethics" purposes.
@pure estuary 
The vocaloid side of the internet is one I've known for years, but I've never actually looked into it musically. I'm a musician myself and teach production at a local studio, and after having taken a deeper dive into vocaloid production (no doubt inspired by your incoherent rambling about Miku), I am thoroughly intrigued by the software, workflows and artists that are creating and fine-tuning artificial vocals. The Vocaloid plugin & voice packs seem to be the current industry-standard for altering existing voices or generating new ones, so I'll be picking that up to tinker with. I've not seen anyone in the HarmonAI discord attempt to recreate vocals - its all backing tracks and beats over there. My guess is vocals are further away - when language models and music models are combined. This may have already happened - I shall look further. I've yet to do any AI music gen myself, as I've been stuck on anime waifus (for obvious reasons). I'm eager to get proficient in AI music gen & vocaloid production before the two merge so I can participate in that change.
I also stumbled across this - a 17yr old producer making stellar vocaloid music. Figured with your unhealthy fixation on Miku, this might interest you: https://www.youtube.com/watch?v=Oxm99_3b230
🍵
ㅤ
Inst https://drive.google.com/drive/folders/1ld-orYQ_uXjVr9s3X7e4nRkZZfMJwS5f
傘も差せずに No, no,
塞がるEyes
帰れないなら あなたも
Be my guest
よく出来たハリボテの 綺麗な笑顔?
人前じゃ崩せない あなたも Be my guest
約束の時間でも 迎えが来ない?
屋上から出られない あなたも Be my guest
黒猫のように Step'n step'n step
手当たり次第に Tap'n tap'n tap
あなたを連れて Snap'n snap'n sna...
Generated with a prompt taken by ChatGPT.
Hello offtopic but can anyone help me? How can I enable sound on win 7 safe mode
Honestly thats the kind of question i would definitly ask chatgpt nowadays for a clear and fast answer.
I have never heard of safe mode disabling sound so i would expect the normal sound mixer and the volume buttons on your keyboard to still work. If safemode itself doesn't have sound enabled then you will probably have to reboot into a different safemode (similar to safemode with internet, safemode with command prompt) I remember there are settings for safemode in "msconfig"
I am here to find information about this rumored conversation llm stabilityAI is working on but so far the only thing i found is someone asking the same question.
I am really curious if it will have the same code/scripting abilities because since using chatgpt my work has not been the same and i don't wanna go back living in the dark ages now that its nerfed and barely accessible. I really need something local and reliable
did yall heard the prompt music generator based on stable diffusion? shits cool, this + distilled stable diffusion will be so great
Weird observation: I've noticed that the single-word prompt "bump" produces exclusively either wedding-ish photos or interior design photos. Rather than...IDK, say, anything with a bump or anything bumpy.
(No intended usage for this prompt, just tried the word "randomly").
It's at least because services are not running ... services.msc → Multimedia Class Scheduler, Windows Audio, Windows Audio Endpoint Builder... But then "This service cannot be started in safe mode"...
Depending on your usecase, msconfig → diagnostic startup may be an alternative to safe mode that's useful to you. However, changes made by it can't really be reverted.
do u wanna partner with https://www.seratio.com/home DM me ?
lmao
people are so gullible 💀 i saw the comments and those artist were celebrating their "triumph" against ai-art
XD
This is spreading pretty fast, the hate and negativity against AI art. People need to see the positive side of the technology instead of doing this but unfortunately that isn't happening.
ai art is forcing them to improve, i saw some of them peoples art and they were the most amateur pics ive ever seen (today)
they can learn a lot with sd tbh
as a musician, i learned a lot with jukebox ai, it opened my mind to new ideas
and as someone who has naturally a good drawing skill SD its definitely helping me a lot
Yes but only if they understand this. I know that it could be the best source for generating ideas, and if I was an artist with amazing drawing or designing skills, I know how useful AI could be for me and I would make unimaginable art pieces with the help of it.
Currently StableDiffusion is defiantly a tool that artists can still leverage to create even greater and novel works of art. And that should not be trivialized, or banned or censored.
What a lot of artists are upset about is that 2D digital high quality art is now ubiquitous and anybody can create it.
They also think that the AI stole their artwork, but i dont think that is true, i think the AI used their works for reference and studied the. And is recombining those ideas very similar to how any human artist does.
stolen art is really just an excuse, they would have the same problems if it had not trained on their stuff but could still create amazing art.
They are just afraid of losing their jobs at the end of the day, same as every other sector that has been replaced by new innovations throughout human history
i mostly agree. But when i think about it, when somebody uses SD to create art, they are really not in total control, they dont plan how the lighting is going to be or what type of light, or the color harmony, or the composition rules, they just churn out good stuff because the AI does it for them... And in that regard i can see why artist are annoyed, but they need to accept that the game has changed. Anyone can now create good 2D art
But theres nothing stopping artist from leveraging this AI as a tool and making something greater
Im working on something right now that leverages SD and goes further to create somethign that im going to sell on the market places
im just leveraging SD as a tool in my workflow
the best stuff made with ai still requires great artistic talent. Artists that have embraced it make a lot better art then your average user
That is true and i think its mostly down to their artistic vision or lack there of it
artists won't go anywhere, but the skill set might become different from today so it will probably be other people then the ones today. At least for some, and that's what they are really scared of
They should not be scared.. Because with art its not really about the skillset its about the vision, its about the idea that starts in mind, and the artist uses whatever tools avialalbe to reach that vision.
they like using shocking words like "stolen" to gain sympathy, but SD just turned art (or images in general) into noise
and then turning that noise into images
Does anyone know any AI which can generate face rotations of a Generated image. For example I have generated a Face using SD, Now I want different angles of that face.
I tried searching but I found Below links but I don't know how to use them
https://github.com/SonDaoDuy/GAN-for-face-rotation
New implementation of face rotation based on Ganimation and CVPR18 pose-guided papers - GitHub - SonDaoDuy/GAN-for-face-rotation: New implementation of face rotation based on Ganimation and CVPR18 ...
probably the best route would be to apply the image as a texture to an already modeled 3D head in Blender. Then you could do whatever you want with it.
i dont know that software. i wish there was some AI to generate different angles
or the skill set stay the same, they just challenge you touch the pen
that remind me of this Artstation portfolio, this must be the cringest portfolio I ever seen 
Art theft in 2030 gonna be guys trying to hack other guys's PC to find the prompt in their wordpad 
my god, people are uploading artstation portfolios full of MJ or SD images... I can see how that would be a massive headache for employers.
you must see 3 days ago, a dude flood main page of Artstation with 200 pics of MJ images in one single time
yea, I guess that's why we have a protest now 
lol their actual profiles will show software skills " discord, photoshop" and thats it.
damn, the first interview line in 2025 gonna be "have to touched the pen?"
Just wait until an AI can gen a picture in .1s then generate an hours long video of it recreating that image in photoshop or krita or something
you thought it haven't? 
I tried making an ai before that converted pictures to vector graphics and it turned out to be absolute garbage
I’m sure someone else could make it work though
!!!! which voca did you get? im a producer and singer as well so i often use them in my work and it's rly fun! its fun to even make them into some more intricately sound design stuff if you dont wanna box them in to only vocals or smth,,,
lmk if you need or want any help,, maybe I can lend some to you after i've dusted the ol miku shrine of mine. Was on a plane most of yesterday, so just now saw this
Alright, which one of you did this?
S12 Image
That muscle are so flawless
Then the backdrop is S11, which just brighten oveylays
@pure estuary I haven't picked it up yet - will start with the Vocaloid6+Megpoid bundle. Even before that, I want to watch tutorials and familiarize myself with the software so I have a better sense about it. I like the Japanese sound packs more than the English ones, so using them as sound design elements immediately crossed my mind - along with stringing together Japanese syllables that might make-up English sounds. If I could speak and write Japanese this wouldn't be a problem. Might get my sister's help with that 
Once I'm seasoned on the tech, and if I start making any music going the vocaloid route I'll definitely hit you up for advice/collabs

That's true, the curious thing is how true it is for human consciousness in general. There's an assumption that you were free to choose what type of lighting to use in your painting, you chose a front light but you could've chosen a rim light instead, you chose complementary colors but you could've chosen monochromatic instead. But the seeming reality is all of these "choices" were caused by events that you had no control over. Where you were born, when, to what parents, your DNA, you didn't have control over any of that, yet it's the totality of what animates your "art". Anyway it's fucking weird but I guess what I'm trying to say is, nobody is in total control of anything.
i mostly agree but i think theres a different way of thinking about all of that. You listed some things that you say nobody can possibly chose such as parents and place of birth, however those encroach on areas that are considered paranormal or spiritual (not science.. yet) and there are reports of people no only having near death experiences or children totally recalling past lives, but also having chosen their parents before they re-incarnated... And on DNA science is now discovering epigenetics(the study of how behaviors and environment can cause changes that affect the way your genes work). And then there is crisper gene editing...
And when we talk about everyday choices, there is some science which you can consider, and that is quantum physics , the many worlds interpretation and the observer effect where by measuring or observing something causes a collapse of the quantum wave function and out of potential probabilities comes one definite reality. It's complicated by some scientists say that when we focus or think about something, if that reality has a pretty good chance of happening and its not too remote a chance there is a good chance the quantum wave-function will collapse to that event...
sorry to go all crazy deep philosophy and science on you.
you could call the above the law of attraction
Even all of that still fits within my view, but here's a simpler example. The next thing you think and do can only exist based on a totality of prior causes that you had no control over. I'm not arguing about anything on a quantum level, that shits weird but more than that human systems are many times more complex and appear to function in a deterministic fashion. Even if we consider randomness, that still wouldn't give you the free will to choose a different type of lighting in your painting, there is no will in randomness.
I think determinism does not stand up to testing, we have no theory of everything based on determinism, and we know that quantum physics is real, and that implies things like spooky action at a distance, the observer effect.. wave-particle duality and all the other quantum effects. im not a physicist so this is all just me trying to get my head around stuff i dont fully grasp and just my opinnion.
Also i dont believe the universe is random.. its extremely fine tuned by an intelligent designer to allow for carbon 13 to exist, stars to form and planets to form with carbon based life.. the odds of it being the way it is are off the charts totally insane.
Fair enough then, good talk
Roger Penrose calculated the odds at 10^10^123
Thats just the odds of the quarks in carbon 13 meeting 11 stipulations simultaneously allowing it to exist and stars to form
I'm not so invested in examining the nature of the universe, just the idea of total control in analogue art creation is funny to me
so basically the universe is not a random thing, its a carefully fine tuned machine
yea i dont think we are in full control, because we are part of the bigger system.
we are not seperate
You've got that one backwards. We're only around to observe the universe happening to be fine-tuned for life because … it is in fact fine-tuned for life. Given some event, the likelihood of its prerequisite events is 1.
yes the anthropomorphic principle. And it relies on the idea there must be infinite other universes that cant support intelligent life
it i believe that with such odds of 1^10^123 that the universe is infact intelligently designed...
where are those odds coming from?
Sir Roger Penrose
OK, that's not really what I meant
This now tells us how precise the Creator's aim must have been: namely to an accuracy of one part in 10^10^123.
This is an extraordinary figure. One could not possibly even write the number down in full, in the ordinary denary notation: it would be 1' followed by 10^123 successive 0 's! Even if we were to write a `0' on each separate proton and on each separate neutron in the entire universe-and we could throw in all the other particles as well for good measure-we should fall far short of writing down the figure needed.
I don't see how that isn't just blown out of the water by the anthropic principle
it is, if you are comfortable with the idea there are infinite other failed universes.
it's open to interpretation and belief i suppose
there's no concrete evidence for multiple universes, and it really begs the question that other universes would be "failed"
The calculation of 10^10^123 was determining how fine tuned the constraints must be to even allow for the formation of stable stars.
that will ignite
well, stable stars as we know them at least
yup, as we know them.
Just … we're around to observe stars, so the "fine-tuning" is a priori
no way of really telling unless we had the compute power to simulate other univesrse evolutions with different parameters
There are almost certainly others that could sustain forms of life
prove it. 🙂
nah, that's firmly a hypothetical. Positivism demands evidence, and that's unfalsifiable for practical purposes
either way i think we are in one of the very few special universes that have the complexity to be able to observe themselves as it were, if you count humans as part of one entire system
It's just … given an infinity of combinations of parameters, without evidence that most of that infinity is a wash, it's reasonable to suppose that some might match conditions we want
Yeah, that's reasonable, though you have to be careful about "special" because that can lead to magical thinking
yea but alter just one parameter and our universe cant support life
even altering the mass of a proton by a tiny degree or any of the forces
we need magical tinking... in order to progress and gain new science
if you know what i mean. imagination more important than knowledge
Oh, it's a thing https://en.wikipedia.org/wiki/Magical_thinking
Magical thinking, or superstitious thinking, is the belief that unrelated events are causally connected despite the absence of any plausible causal link between them, particularly as a result of supernatural effects. Examples include the idea that personal thoughts can influence the external world without acting on them, or that objects must be ...
does that no conflict with quantum non-locality?
???
we dont actually know how entaglement works
the two particles could be light years apart but once we observe one then the other particle flips into the opposite polarity instnatly
that's not how that works
but there is no actual science to say how they are connected
so we make up ideas like a higher dimension or hyperspace or some magical stuff thinking
Repeat after me: there is zero evidence that information can be transmitted faster than light in a vacuum
yea but we get around that because no information is transmitted
you cant do anything useful with it its not even binary code
quantum entanglement is real
right, just people ascribe a lot of bullshit to "quantum"
(Not a physicist, do know more than average)
but there is definitely some very interesting stuff going on when they design very complex experiemnts based on the double slit.. where by we see the act of measurement or even just looking at the experiment collapses the wave-function? right??
i dont think they ruled out the possibility of the observer effect...
you know theres that whole episode on PBS spacetime
"Observation" happens because of the slits in the first place
but what im getting at is they tried to fool the experiment making ever more compelx ones
double slit is just the starting point.. very basic
I haven't seen it, so I can't really judge, sorry
i seem to recall they did touch on the spooky type stuff like observe effect, and such
mind over matter....
magical thinking
There is room for that kind of stuff. but of course it needs to be a properly tested scientific theory
anywayy ive got to get going
I like quantum so I’ll correct some things.
The slits in the double slit experiment do not cause observations. Otherwise there could be no interference pattern.
Second, quantum entanglement is really simple. It’s just reasoning backwards given a conservation law. Let’s do a pool analogy. Let’s say you hit a cue ball and now it has some momentum. Then you close your eyes, then after it collides with another you open your eyes again and see the new momentum of the cue ball. Even though you didn’t observe the other ball, you know its momentum by observing the difference in the cue’s momentum before and after and applying conservation of momentum
The cat is only screaming until i put it in a box with poison gas and a radioactive chunk of uranium
New Collection released: Themed 2023 Calendars
Currently have Kitten, Puppies, Fantasy, Pagan, and Christina themes.
Set up as monthly so folks can mix and match as liked.
Real Storm chaser Lighting strikes and Lighting strike GIFs created from slow motion video capture.
Lich / undead Trading card sets with various ways to collect them, Numerically, Alphabetically, by gender, by Tomb type, by rarity...
Over 250 different items to choose from all at affordable prices.
Especially with Eth being so low right now. Buy low, sell high; I say...
https://opensea.io/Galwin?tab=created
1930s musicians against recorded music
Holy crap I just thought of something, what if AI art haters go so far that it becomes the 'Cream - David Firth' video XD
most artists i know see this the technology with a welcome eye, what they don't like is greedy corpos trying to appropriate and sell them that technology, which is funny since most online agitators seem to be paid by those companies trying to corner that market for themselves and demonize the open source community, just use the name of artists to fuck artists and close yet another avenue of human creativity, they closed cinema and music already
and it seems drawing and any form of written text is next
oh well

the difference is that the product of todays artists is the dataset of inputs for art or large language models, and our concept of copyright can be considered in conflict with this use of the work. practically speaking though, the challenges will lose, and ai-produced work will be considered derivative work, not plagiarism or license violation. That is unless copyright is re-imagined with stricter limits.
Hey ho! I'm hosting a Christmas challenge for next Friday's AI Art Weekly cover. Would love to have you 🙂 https://twitter.com/dreamingtulpa/status/1603725542355652608
Turns out the convolution and deconvolution used in neural networks aren’t actually true convolutions and deconvolutions. This has been the source of a great deal of headache
I just discovered this thing called NVIDIA Canvas and it is really therapeutic to use. I made this really quickly.
Basically you paint with the materials
And it's super fast
But you need an rtx 4080 minimum
Or something like that
Nvidia canvas is for people with money
And a decent gpu
No you don't
I have a RTX 3070 Laptop version
This is quite fun
New Collection released: Themed 2023 Calendars
Currently have Kitten, Puppies, Fantasy, Pagan, and Christina themes.
Set up as monthly so folks can mix and match as liked.
Real Storm chaser Lighting strikes and Lighting strike GIFs created from slow motion video capture.
Lich / undead Trading card sets with various ways to collect them, Numerically, Alphabetically, by gender, by Tomb type, by rarity...
Over 250 different items to choose from all at affordable prices.
Especially with Eth being so low right now. Buy low, sell high; I say...
https://opensea.io/Galwin?tab=created
My Daughter need a piece of software the do 3d modelling, of hand drawn text that ultimately will be rendered in a Plasticine style.
Please can anyone recommend a system to do it. I would have told her Sketch up ten years ago. But seemingly I am woefully out of date.
She has access to the PS suite but AFAIK that wont cut it.
Also we cant just use like automated tools like the Automatic1111 depth estimation because this is for her art course and she ha to do all the steps herself.
Simple depth estimation on a drawing will be considered "cheating"
Does she have access to Adobe Illustrator? They enhanced their 3D tools this past year. Would something like this be sufficient?
Lesson 1 of a 3 part series. You can watch the videos and complete the 3D typography project with me in the new lessons added to the Level 1 Graphic Design Masterclass: Learn GREAT Design course available on both Skillshare and Udemy. Visit link below...
Ask me anything here! 🔥My full Graphic Design Courses (Best selling design courses in the ...
yes, thank you. Will check that out now ❤️
observing something means taking a measurement. that's the effect. to read something, you bounce something off it. at the quantum level, this has an effect. It's not magical. it's just makes the research very difficult, because simply watching what particles are doing means doing something to them with other particles or fields. no matter how complex experiments that try to observe particles directly are, there's always the problem that whatever you use to measure them affects them.
not that "we create reality through observation" hooey dooey
if we observe the quantum experiment just by watching it then we are co-creators of reality. our influence made the wavefunction collapse into a certain state?
butterfly effect? quantum effects propagating up into the macro scale of reality?
surely if a person has an intention, they can influence reality
at a macro scale, it's not like that at all. conscious observation isn't what affects stuff. it's measureing a particle with a field or another particle, that changes it. This is happening all around us as energy transfers everywhere. It's a giant organized soup of chaos. There's no evidence that conscious thought is what "collapses a wave function". That's just hooey that new ageists use to sell books and dvds
i see your point, but by what you just stated you are implying that there is a solid line between quantum effects and macro scale effects, and that the mind is somehow closed off from the universe and cannot effect it
where is that line between quantum scale and classical scale
no. that "line" is not solid. it's the giant chasm of unified field theory that's undefined still
so a lot of science that is yet to be defined and tested?
how can you rule out conscious thought affecting the collapse of the wave function?
double blind tests
its more like you have to prove that consciuos thought has any affect. It may not be ruled out, you're right. Neither is mice running the world in secret. but to make that claim, it requires a solid rock of evidence to stand on
double blind tests where someone in a room is thinking about the test, while another is conducting the test, should produce some sort of results that indicate an affect
yes but the way you are stating that is to ridicule the hypothesis
maybe they're just not thinking hard enough though right? it doesn't work while cameras are on or something
lets be scientific about it and not silly
if scientists cant take it seriously then we are not going to have any proper testing to rule it out
you just said it has not been ruled out
you're talking about conscious thought. how is my experiment not a serious approach to that research?
what experiemnt?
the one i just defined that you called silly
sorry
oh right. yes but it sounds like something that has been done by a couple of scientists somehwere and then rule out
but heres the problem
my point was, this test has been prepared for decades. It's ready to go. it always goes. it never works. thought just isn't measurable. Usually when people who really want to prove it can't, they make up reasonings like, "cameras were on" or other cheap excuses. the gif i posted was uri geller. a fraud from the 80s and 90s that was exposed by such tests
honestly the tests were probably considered centuries ago
Most people have had this expereience where they think of something, or a person, or have an intention, and then amazing coincidences happen as if that thought influenced their reality... This gives creedence to the idea of thought affecting reality
defining what consciousness is has long been something science wants to do
so that double blind test you mentioned... it needs to be repeated all over the world extremely rigorously and seriously
we always remember those amazing coincidences, but never the 10000th moments aroudn them that were just mundane chaos that didn't go our way
it cant just be dismissed and laughed at
all the times we thought of someone and they didn't show up
thats another cheap way of dismissing it
selection bias is a real thing and scientists have to remove that from any of their experiments
its not every moment that we have coherent thought and intention.. most of our life is meaningless random brain activity
so i cant really buy your explaination
selection bias is a real problem and not just a "cheap way of dismissing it". removing biases is core to the scientific method
and it's the BIGGEST whale in the room when talking about scientific consciousness. Not even an elephant. a gigantic space whale
ok point taken.. but like i said, its not very often that we have such choherent thought or intention
so i dont think there are 10000 moments to dismiss
i would say there probably is a selection bias problem
the mind is funny at convincing itself and self reinforcing. Everytime you think about someone showing up and it doesn't happen, you never dwell on the thought and reinforce it in yoru memory. selection bias in action
but i would not be so quick to dismiss thought influencing reality by collapsing qunatum waveforms or whatever the physics is
I would
well ive had things happen to me that jsut blow your view out of the water
If it was a thing measureable within our models and understandings, it would've been weaponized by now. Humans are really tenacious like that.
I've had some cool coincidences too and yeah. It's selection bias. Life is a miracle but science is serious business
i see events before they happen,, i know when serious events are going to unfold(somtimes) before they happen, i also have specific dreams of disasters,, (several per year) before they happen. AND it runs in my family
but you would dismiss this as a joke
intuition is a good skill. what you're doing is guessing based on past events and how things are going. it's not prophetic visions. you're just good at guessing how a game will play out
you probably get a lot wrong a lot of the time but you never self congratulate yourself in those moments
no it wasnt that.. its like that movie sxith sense
when something flashes into your head
i do see your point though
inner vision is somethign a lot of people have. most i've heard. i don't.. my inner mind is all dialog and auditory
our brains are powerful prediction computers
aphantasia it's called. i'm mind blind
yes i heard of that
yeah. you're the norm. peopel who don't see visions in their mind are the odd ones out
sometimes i think my visualizing skills are dimishing though
especially with all the time spent on a computer
for me, my predictions come through like rambling thoughts . sort of like this guy
🙂
its like GPT 3
but i dont think we can rule out the whole mind accecting reality stuff
because the mind is not isolated from the universe...
i dont have full aphantasia though. i still do see blurs of images and such. now an then. it's definately not the primary mode though. i think of it like left or right handedness
but what about REM sleep dreams?
if you stayed up for days then tried going to sleep, you should see vivid images
from your brains lack of REM sleep that it needs
vivid visual dreams about once every eyar or two. rare. weird. usually when i've ate oysters or something odd like that
i get rem.
i am lucky i guess with having vivid imagination
dreams haven't even been proven to happen during rem. the only way to tell someone is dreaming is to wake them up in that moment, and ask. all stages of sleep have been found to have dreamers
i also automatically see all sound in my head
even speech
my minds eye is like winamp
it really whips the llamas ass?
yep
whupaw
actually with things like speech or lyrics i see the words like 3D text floating in my minds eye
some of the premonitions though, they are completely uncanny.
such as seeing in vivid detail a russian fighter jet crashing off the end of the carrier, and seeing it from like a go pro drones persepctive... then it happening in the news the next day
and all of my premonitions are of transport disasters..
mostly aircraft
i also saw a russian firefighting plane crash into a forest
my mother saw the space shuttle challenger blow up from the perspective of a school kid,, days before the launch,, and she told her friend about the bad dream
then that of course played out.
its like that movie minority report where they have peoples brains plugged into computers to predict crimes
When we get MRI machines that advanced, we can finally start putting things to the test.
There are very good midjourney artwork. Instead of taking artist work it is possible to take those images and train on it. What is issue with it
diffusion radio 24/7 HarmonAI
🎵 Introducing Diffusion Radio - A 24/7 livestream of AI-generated music! 🎵
The Harmonai team has been working hard on diffusion models for music generation. We're happy to bring you this stream to share the progress of our experiments!
We definitely appreciate any feedback you have on the music. These models aren't fully trained and are being...
Introducing our AI-powered image generation app - the ultimate tool for unlocking your creativity and bringing your imagination to life. With our app, you can easily create stunning images with just a few taps on your screen, whether you're an artist looking for inspiration, a business owner needing…
What are y’all’s thoughts on 3D model ai generation?
Surprised it isn't already happening at publishers like EA and Ubisoft that have millions of models in libraries.
Stuff like UV mapping should already be done mostly by AI
Anyone using SD for non AI art use cases? I'm testing to see if it can be used to solve complex business problems.. Would love to hear other people's thoughts on how to apply this technology to other problems..
because midjourney has also taken artists work
besides, i personally dont believe in training AI with AI works
mainly cuz i think that AI generated imperfections will only multiply
like inbreeding
i kinda want to see what happens if we do that tho, i want to see the horrors
how my parents see the future of ai generation
complete apocalypse, nothing will be real anymore, and everything just turns dystopian somehow
Gen Z in 2060:
"Back in my day, we paid $10 just to watch a movie, not generate it from scratch!"
training AI on AI is a way to destroy the model's knowledge. It's fine if your goal is to compress a model or otherwise make a smaller version, but it shouldn't be used for actual main training
s'wat im sayne
🤣
I broke chatGPT
Python Devs beware: https://www.theregister.com/2022/12/21/pypi_malware_sentinelone/
Hi Folks,
Today I'm launching PhotoLeaf.AI after working on it for past 2 months. Planning to charge just $1 for each training since I'm just running my own GPU servers.
Here is my product hunt page -> https://www.producthunt.com/posts/photoleaf-ai
Launch Tweet -> https://twitter.com/therahul_me/status/1605848288711757824
Please check it out and give us any feedback. 🙏
PhotoLeaf.AI creates high-quality photorealistic AI-generated profile pictures that you can use for avatars, dating photos ❤️, and professional 🤵♂️ LinkedIn photos just like yourself in more than 50+ styles
✨ Launching https://t.co/0A4qaQVTTk in @ProductHunt! 😻
📸 Create photo-realistic profile pictures with AI
⚡️ More than 50+ styles and 300+ images
🚀 We train the models longer on our GPUs to produce high quality images
New Collection released: Themed 2023 Calendars
Currently have Kitten, Puppies, Fantasy, Pagan, and Christina themes.
Set up as monthly so folks can mix and match as liked.
Real Storm chaser Lighting strikes and Lighting strike GIFs created from slow motion video capture.
Lich / undead Trading card sets with various ways to collect them, Numerically, Alphabetically, by gender, by Tomb type, by rarity...
Over 250 different items to choose from all at affordable prices.
Especially with Eth being so low right now. Buy low, sell high; I say...
https://opensea.io/Galwin?tab=created
Hi
Does anyone recognize this song, please? 1:30 3:00 https://www.youtube.com/watch?v=0TB79fG8WyQ
https://www.youtube.com/watch?v=W4FUAd6AyYs this maybe
Eze Colombo, Ram-J, Sheism - The Late Rain (Original Mix) - Mirabilis Records
Mirabilis Records MIRA148
Eze Colombo, Ram-J and Sheism deliver the an amazing three-tracker in the form of "DIFFERENT CODE".
The guys certainly do not need a special introduction as their previous releases and appearances on Perfecto, Stripped, Vesta all spea...
Nah, still could't find it tho. I know it's phonk music, but that's all i know..
Does anyone know this song, please? 1:30 - 3:00 https://www.youtube.com/watch?v=0TB79fG8WyQ
Darude: sandstorm
any good graphics apis with windows for c or c++?
I can't find any that aren't ancient
and I have zero clue how to even learn to make a new one
Depends on what you’re doing. I l like dear imgui for quick and small projects, but I probably wouldn’t use it for something that’s actually getting released.
Hi
Merry Christmas @lofty bolt and community: https://ecardai.com/c/q3e2ymLpraXWhXM4. Built a little Santa greetings service with Stable Diffusion in the back. Thanks for open-sourcing such a beautiful piece of technology 🙏❤️
Greetings fellow tech enthusiast! I am seeking a collaborator to launch a platform utilizing a Stable Diffusion, which is capable of producing hyperrealistic images given any text input. Our goal is to surpass lexica.ai in terms of autonomous creativity. If you are interested in discussing this ambitious project further, please do not hesitate to reach out.
Ladies and gentlemen, Presenting you my generated 2minutespaper introduction to stable diffusion for manga artstyle
There's a bunch of people on twitter already trying to claim ai art in general is just art theft. Is it though? Which ai? All of them work differently.
They just panicked, imagine 1 customer comission is equals to 25$ only for headshot or half upper body. And that even for anime-less drawing. I saw many artists claim this is anime art style but for me naah. Even manhua and manhwa still better
That depends on perspective. In general most of the “AI” is latent diffusion, all the same basic tech. Latent diffusion is ultimately statistical; it finds common features across thousands/millions of images and then uses those patterns to “denoise” random noise
No individual art is “stolen”
I don't understand how anime is related (talking to jack). From what I understand, there are people getting confused and thinking every image ai is automatically art theft because it can sample stuff from existing sources. I mean, would that make remixes and rap theft? How about parody? While those things are still created manually and not generated from something that does all the hard work for you, that is not the whole purpose of it. It's a constantly developing thing that gets smarter and learns things. It can be incredibly useful to create assets for yourself to create a bigger thing that you can do manually, which is what I use it for. Not everyone has the materials required to make the art they want. Some artists contribute to ai. nobody does any research.
Can't fight the mindless mob.
Hey, I was curious to know which website the images used to train the AI come from.
Does anyone knows ?
they use a LAION dataset
which is basically a consolidation of images around the internet like an open source google image search
Thaaanks !
https://rom1504.github.io/clip-retrieval/?back=https%3A%2F%2Fknn5.laion.ai&index=laion5B&useMclip=false if yoy want to search it
Clip front
they list the actual dataset on the model card iirc
Quick join link for the game server: (must own Garry's mod on Steam)
https://cli.re/7wpR8j
IP: 45.88.228.104:27015
( add the IP to your favourites )
Discord invite:
https://discord.com/invite/ND8pzmc7rJ
Become a Patron:
https://www.patreon.com/capitalcommunity
Community Forums:
https://capital-community.com/forums/forums/hogwartsrp.66/
Stea...
Just dropped a new track from my upcoming album
https://soundcloud.com/helicopterdown/runaway
I never saw a twitch stream. Was there one?
There was one few days ago, you can still see the video.
Thanks
I went and searched it a bit and love that it goes more back to history.
1913 newspaper advertisement
“A picture is worth a thousand words” originates with the famed Norwegian playwright Henrik Ibsen. He was quoted saying the following: A thousand words leave not the same deep impression as does a single deed. It’s clear that this quote portrays much of what the idiom “a picture is worth a thousand words” does today.
can someone tell me how to use and train gpt-j-6b
I found this on the github for it. https://github.com/kingoflolz/mesh-transformer-jax/blob/master/howto_finetune.md
thank you
Are ai voiceovers good yet? Working on a solo game project
Do you know any open source text to intent AI ?
Basically an AI that can sort texts into categories (insult, compliment, etc) ?
Yeah but the best ones are expensive
https://replicastudios.com/account/signup?referral_code=9b63yVdE here you go though, use my referral for extra 30 minutes of text2speech generation (total 1 hour)
New Collection released: Devine Intervention
Also have 2023 calendars, Real Storm chaser Lighting strikes and Lighting strike GIFs created from slow motion video capture.
Lich / undead Trading card sets with various ways to collect them, Numerically, Alphabetically, by gender, by Tomb type, by rarity...
Over 250 different items to choose from all at affordable prices.
Especially with Eth being so low right now. Buy low, sell high; I say...
https://opensea.io/Galwin?tab=created
fr
not my intention but ill take it
@woeful ivy Hi Sage, I hope you enjoyed the holidays and everything. Anyways, we had a discussion about you on the NAI server and regardless of your workload now, we still love you and you'll always be a NAI dev!
I hope you got a needed break from carrying Stability in the last few months, it was really fun and confusing to see you take on the SD announcements and responsibilities. I am super happy to follow you along with your career progression from the past year, it's been absolutely awesome seeing you go places I never imagined you going, all because you really liked modules.
Anyways, I wish you the best of luck in the New Year's, and hopefully still stay in touch, even though you probably a big cheese now. 🙂
Sincerely,
Potato
that's pretty nice
off topic, but can I link a Christmas song I wrote last year? It's Solstice time so I have an excuse to share it
Maybe you should wait for a mod to respond on this, but it should be okay since it is off-topic section and I think I've seen people sharing links above.
I'll just share.... I don't want anyone to buy, just listen.
Any my little pony fans in here tonight?? (should probably make the intro longer, im so high ive lost all awareness of my surroundings😱😱😱😱😱)
Sorry that was actually meant for the jazz channel But yeah this is definitely a new low for me. I gotta go
I'll leave the song with y'all, ❤️
If I have a set of like, 200 images individually captioned all of people but different people / angles / zooms / positions it's not really gonna overtrain, even if I do a looot of steps at a high LR, right? Because of so many diverse images?
Hello
Do I want to play video games, or watch Anime on my day off?
i just finished doom eternal lmao
am I the only one bothred by the constant attempt of painting women as lesser or as victims? That they've been forgotten and placed a to the sidel when what really happened? Their ideas sucked and venture capitalists didnt want to invest. now they're beeing pressured to annd would rather pull ou
Man 😭
this might be a case of seeing what you're looking for. i don't think theres a constant attempt. there's also quite another edge to the whole "women" topic. Many people focus on the whole "women are useless and that's the real issue" trope that comes out of toxic masculine culture. If anything, i think that's the more constant barrage. But i grew up drowning in misogyny, so maybe i just see what i'm looking for.
Nah, I was more saying that maybe most people suck and there just happened to be more men than women there so fewer women ended up being picked but the same amount sucked
I dunno though. I dislike rushing to anwers before I've comfortably arrived at once and yet it's a failing of mine to do that a lot
I live in a country/region very used to the highest degree of equality in the world though, so maybe I take it ffor grnted
Does this look like a good choice for a desktop system tailored for Stable Diffusion? https://www.nvidia.com/en-us/geforce/news/geforce-rtx-4070-ti/
if you're aiming for stable diffusion, gaming features don't matter. all that matters is memory bank. the 70ti will likely have 10 or 12gb. you'll want more. The most you can get. 24gb if you can manage
looks like 12gb. 16 is better. 24 ideal. more even better
Thanks! So what Video Card would you recommend? Am I right that each card comes with one preset amount of VRAM?
Yeah you can't upgrade a card's memory. you buy a whole new card. i wanted a 3090ti with 24gb. i settled for a 4080 with 16gb. nothing was in stock
i was considering amd with 20gb but CUDA is important
I have a tech guy who was suggesting a 4080. Then I saw this annoumcement today. Do all 4080's come with 16gb?
all are 16gb
I'm currently using a 1660TI so almost anything would be an improvement.
torch.py doesn't fully support the 40 series yet iv'e noticed too. there's some manual adjustments. all part of the learning process. installing newest dlls for cuDNN sort of thing. xformers. different little tweaks because of brand new architecture. That'll all get smoothed out in time
Sheesh, the last thing I want is to get a card that's not quite compatible! So I should be looking at the 30__ series?
i like my 4080. it also games well
it's quite compatible. i just had to tweak a few things to make software catch onto it
xformers isn't enabled by default and a1111's ui has an older version on it's repo
I need something that will run CMDR "Easy Diffusion" right out of the box. I'm not technically savvy.
its always teh way with brand new hardware. lovelace will rise up yet
yeah it should run out of the box easy
little bit heh. its what the diffusion network uses to talk to the gpu. sort of like a machine learning directx
it may be a few layers removed from the gpu actually. its important though
I'll mention it to the tech guy who's going to build a new computer for me.
But he doesn't know anything about SD.
PyTorch is a machine learning framework based on the Torch library, used for applications such as computer vision and natural language processing, originally developed by Meta AI and now part of the Linux Foundation umbrella. It is free and open-source software released under the modified BSD license. Although the Python interface is more polish...
Thanks. I'll forward this thread to him. But if the 3000 series doesn't have any issues with Torch.Py, maybe I should just go that route. I told him I needed a 3060 minimum and he said he could get me a good deal on a 4080.
the issues are more that it just needs to be updated to newer better hardware architectures
that won't take long
for now, 4000s aren't that much better. soon they could be as the hardware iis flexed on better
I got a 4090 and with much struggle managed to get it to work up to 39it/s on SD2.1 single image running on 450W
Do you guys think GPU prices will ever become normal? Maybe 50 series? It just feels like copium at this point
Uh-oh
yes everything will get back to "normal" very soon, just hang tight ...
🤣 2023 is gonna be fucking hysterical
This seems to (partially) happen because of CAA gofundme
No clue wheather there's any actual connection or not
doubt
If I ever run a service, the European version will have to run on the dark web 🇺🇸🇺🇸🇯🇵 🏴
AHAHAHA I read the gofundme by the CAA lmao. WTF is this? Absolutely absurd. Like impossible.
First point seems okay. Second is...uhhh....okay? Third and Fourth are full on off the rails bullshit lmao
Smh, I am just gonna delete Twitter.
Not worth it.
The best part for me was the entire campaign is to hire 1 (one) lobbyist to go to DC where there's 12,000 lobbyists competing for legislator attention already
https://eare.eu/japan-amends-tdm-exception-copyright/
Summary: copyrighted content can be used in validation data sets and training data sets as long as the original copyrights are not violated by distributing the AI outputs
In more off topic news https://japantoday.com/category/national/man-who-mistakenly-received-entire-town’s-covid-19-relief-money-vanishes
Last month we reported on the curious case of one man receiving 46.3 million yen in 100,000-yen COVID-19 relief payments intended for 463 households in the town of Abu, Yamaguchi Prefecture. At the time, the mayor apologized for the clerical error that led to the mistake and was pursuing ways to…
Oh lovely here comes everything bad
The luddites will lose this one muahaha
Oh boy, another politician group making new rules about a topic they know basically nothing about, no way that's gonna end badly
Excuse us for being part of a community that has done more for AI than they can even imagine
That really is so out-of-touch that it's funny lol
i was thinking a lot of these artist may see more business as their name and style are being advertised
Just one more generation - all night, then go to work in the morning. who needs sleep?
It's funny how "estimated time" in literally ANY software dosen't work the way it's supposed to be lol. We just see how much it "jumped" to see how fast it is
captain fabulous
a cloud
These Ai stuff chat or art generation is turning the world upside-down. Are we are gonna see a lot of new clownery rules incoming in different apps or places? (Copyright rule this or that.)
gay rap https://youtu.be/02acbPVxlQU
Quick join link for the game server: (must own Garry's mod on Steam)
https://cli.re/7wpR8j
IP: 45.88.228.104:27015
( add the IP to your favourites )
Discord invite:
https://discord.com/invite/ND8pzmc7rJ
Become a Patron:
https://www.patreon.com/capitalcommunity
Community Forums:
https://capital-community.com/forums/forums/hogwartsrp.66/
Stea...
I just tumbled on this, it's on github but I haven't had time to test it https://devpost.com/software/reblock-xjgkrb
deep learning? i thought sponsor blocks just yanked the subtitle files and nixed any blurbs that like "and if you...X then you should try Y" etc. like spam filters do
sponsorblock works perfectly, although some channels actually do really funny sponsorship skits so worth whitelisting them
society is largely conservative when it comes to the realm of mass technological disruption. even those who want a left government, look at the disruptive power of tech and say "heyyy wait a second!"
I personally say lets disrupt it all. Every industry. Ai through every profession. Lets get this revolution started. But that gets people nervous.
At one point, Digital media was so disruptive to the legal court systems that a compromise was designed. the DMCA. We're still living under that compromise. This new tech has kind of broken it. While i believe copyright still goes to the author, none of this PD on all creations nonsense, i do think that big changes will have to be afoot
And what after all that?
The rich get richer, and the poor get poorer? Or something else?
There will be no singularity
Nor is anyone trying to make it into one
@glacial turtle You didn't think this through did you?
This is automation and while I am not against it, what you said will only bring problems
That's perhaps one of the more "bad" scenarios. But I really can't see anything that it does that makes things better in the long term
Still not against it, ofcourse
you don't think technological advancement is increasing? i think there will be a singularity. a moment where astonishing advances are broken open. It's not a new idea and its one that information technologists have considered for some time. The hooey part of it is what's on the other side
hard to say really
isaac asimov even conceptualizes thhe idea as multivac in the final question
Oh no no I thought you said singularity in AI, not technological singularity lol.
ai is just one tech that will lead to it
Yeah that I can agree. But I worry about the implications it may bring, because the singularity will eventually happen
yeyeh i dont want roku's basilisk to simulate me being tortured for eternity. lets get this singularity rolling
As a Socialist in far left circles, I can tell you that you’re 100% correct here, most other far lefties I know are as reactionary as any conservative is when it comes to tech
I’m a Transhumanist myself, am totally on board with accelerating the singularity, merge me with Helios and Nanotechnologically augment me off biology J.C. Denton style baby. I’m ready.
I dont agree that we'd see some kind of discontinous singularity moment. what we have in the real world is exponential progress but also exponential problems and exponential expectations. tech is still progressing but a bit slower than the original kurzwellian projections
but having said that AI art definitely has a lot of disruptive potential. I'd like to see the way entertainment works evolve more toward the extreme customizability shown in the holodeck (that cliche , still stands as the end goal of the games industry IMO)
modding on steroids
It looks like Apple quietly released their own Text to Speech which sounds pretty close to SOTA. A really interesting detail is that the term "AI" doesn't appear a single time on this page
https://authors.apple.com/support/4519-digital-narration-audiobooks
AI has literally become taboo for a shit ton of people. They aren't taking any chances with the public. Literally the other day I saw an AI program (that had nothing to do with AI art) get released for blender (it had something to do with rigging chars models) and it got hate because the word "AI" was there. And literally yesterday people started an internet riot against Adobe because they "collected data for their use" which literally EVERY software uses to improve it. These guys have become extremely paranoid, and any mention of "AI" will put their radars on high, almost like some form of trauma response (trauma against nothing Bdw, AI hasn't even done anything yet)
On another topic, it's my second day of not using Twitter and I already feel much better
5 days is my next target
Yeah I don't blame them for trying to dodge the heat. It is very obviously AI text to speech though, even if it is a leap better than the Microsoft stock text to speech.
I am happy that TTS tech is moving forward though. I mainly listen to audiobooks instead of reading and I also want to listen to things that can't practically be narrated by a person like blogposts
Surrogates here we commme . lets go!
I actually think we're right on track with kurzweils projections. There's some dips in the progress, but a breakthrough like deep learned networks has rebooted a lot of research initiatives, begun gaining momentum, and acceleration is right on pace again.
I don't think there'll be one singular singularity moment. It'll more likely be a convergence of technologies in someone's basement. probably multiple basements. As soon as all the pieces are ready to fit together, they will just spontaneously become "the singularity" (whatever that rapid advancement that changes EVERYTHING is)
Chiplets have given space for moores law to progress in a new manner and new material substrates are going to allow for even smaller transistors when we transition away from silicon
lithography stalled.. perf per watt and per $$ isn't advancing so fast . GPU prices are a bad sign. chiplets etc aren't giving us "more for less" as we always used to get up until the atomic limits were hit. can't remember the exact threshiold but we've started going vertical ,thats making each step much more expensive;. Anyway it IS still progressing and has certainly reached a level where some remarkable new capabilities are being rolled out
Lithography hasn't stalled. the demand for chips is just so high that they can't keep up. they're producing more chips today than ever before and the market still isn't saturated. What do you mean it has stalled?
progess
Chiplets are first generation still and show a lot of scale potential too
I've seen no such stalls
wasn't the rtx 40 series the largest boost nvidia has ever done to their new models?
the rate of progress was much faster until we hit 14nm or something
mostly brute force and they're also sky high prices.. they're not cost reducing as fast as they used to
that's just moores law reaching the edge of silicon accuracy. Like i said, other materials are being found and new transistor substrates will come into play. A carbon atom is something like 5 times smaller
atomic weight of 6 vs 14
if adding more transistors gets more performance, whats the issue?
Reducing cost isn't really a part of the equation. If every single one they build is sold, advancement is still ongoing
you could always put more computers in a room or make bigger chips. The real progress came from transistors shrinking ,so you were getting more for the same amount of materail. Thats no long the case.. we're having to build vertically, so the rapid price-performance increases aren't happening anymore
performance per price is everything
Shrinking isn't the only advancement. Focusing on it is narrow minded
shrinking is what let the rapid advacnement happen. thats what is slowing now.
the 4080 is by far the biggest gpu i've ever owned. That in no way affects it's performance
the size of the cooler lol
it can slow. we got plenty of scalability in other arenas. Chiplets for example
chiplets are brute force. spin it however you like.. the rate of progress today isn't as fast as it was pre 2005.
The cooler is so big because the die is HUGE and eats watts like they're wheaties
Hello all.
I have added 3 new videos to my Youtube channel in the "Projects" play list section.
- How to make Fermented Pepper powders, Paste and Hot Sauce bases.
- How to make an efficient double walled Terracotta Candle heater for your green house.
- The final section and conclusion to my "Building a Green house with Aquaponics fish tank" series.
More will come over time, to include more gold prospecting how to videos.
All can be found on Youtube channel by searching for "Galwin the Fossilized Prospector" and checking on (Play List).
https://www.youtube.com/@galwinthefossilizedprospec6999
I also have many NFTs for sale on OpenSea(.)iso site and hopefully once the Carrot App is able to handle ETH will have listings there as well.
You can find these at:https://opensea.io/Galwin?tab=created
or by using my provided link in my Discord profile.
yeah and thats the other important factor. perf per watt.
doesn't matter that it's brute force. progress is progress. if bruteforce is the new way to gain performance, that's what is oging to happen
i never said progress had stopped. it's just slower now. There's factors that can kick in like mineral limits . they're reporting that manufacturing costs are increasing. This singularity is by no means guarenteed.
viable fusion energy is now possible thanks to berkley/usa. woohoo. Electrical efficiencies won't matter soon. Heat dissipationg though... 7990 xtx and the 4090 both suffer from extremely hot electrical bottlenecks. But that's a whole different problem and nothing too do with information technology advances
I don't think progress is slower overall. We're right on track with futurist projections. Kurzweil and others
microssoft explained this with the need for the 'xbox series S' - they released this because they anticipate they can't cost reduce the state of the art device as they could in the past
There's so many different areas we can advance tech, moores law is an old paradigm that is loosing it's dominance in the design culture
look at stable diffusion even. none of that is hardware. It's all software advancement
viable fusion energy? you mean that recent announcement where they had EROEI of about 1.2 .. without mentioning the input to the lasers.. fusion is still "20 years away" like it was 20 years ago.
oh you're one of those lol. The paper that was published to nature was peer reviewed already. They mentioned everything. This is real hard science they're doing.
the core ideas in NN's were known decades ago, it's the progress in moores law thats enabled them to do new things
The literal Department of Energy announced it. It's no small event
ideas and implementations are kind of different things
Hi there
software doesn't just spring out of an idea. Real world engineering work must be done. It doesn't just happen because theres a fast chip
the implementations were bounded by compute power. there are tweaks but the bulk of progress with AI lately is down to moore's law allowing bigger nets.
SD can run on a 1050 if you do it right, but we never had it back then
Can't you run it on your CPU to9?
I feel like you're arguing just for the sake of it, and you're digging in. So... now that i feel that way, i can't have an honest conversation anymore
SD on a 1050 is too slow for anyone to care about what it can do, its the higher performance that brings use cases into reach
slowly but yes. I'm sure many people would've loved to run a slow stable diffusion 5 years ago
Can't have a honest conversation anymore because of discord
Well text to image wasn't even that good then
"discord" isn't a human. someone with a bad attitude is. i'll blame bad attitudes instead
It still existed
todays software can run on hardware that old though. So we COULD'VE had it back then if the software advancements were something not to consider
Yeah but did you consider the training time?
but SD is novel and wasn't around 5 years ago. it's an advancement on it's own
peopel only bother with it when its fast enough to be useable
back then people were using other generative aalgorithms which were more practical on the hardware
Be happy that OpenAi revised this idea and sparked literally this whole community
If that didn't happen this wouldn't exist
People use cpu generation for SD a lot. They'll wait 10min in queue on a public service to get their result. 5 years ago if it was available like it is today, just slower it would've blown up then too
did you know that coders used to let their work compile overnight before thye could test? they bothered then. that's why we're here now
Explain jukebox AI that takes more than a day to generate a three minute song, I did that like every day
i only care about it because it runs at a certain speed, otherwise its not much different to finding images online+photoshopping
ahhh. you only care about the speed. now we're at the crux
when i first started 3d modelling, i had to wait a minute just to get a shaded preview
if it takes 10mins to generate you're better off learning how to photoshop things. if it generates in 5 seconds , now thats transformative.
Bru just use like colab or something
This dude can Photoshop in less than ten minutes to get the prompt of whatever I want
Unbelievable
i use my 3080.
I use my 4090
I think you're ignoring a lot of significant advances and hand waiving away a lot, because it's not the most uber convenient jarvis robot butler in the world.
i wanted teh 3090 but gpu retail in western canada is ffffked and i had to settle for the 4080
I want to fine-tune my own model
Anyone have some advice on this?
Like, actually finetuning one, with a few thousand images
I remember it used to take like 20 minutes to generate a single image before SD lol
you've got 24gb so i think stabletuner miight be viable for you
looks like the author uses a 4090 too
Yeah, buuut how many images would I need?
supposedly not so many.. i think people are able to finetune with dozens not thousands
I don't even think you know what I mean
I'm talking about training without dreambooth
finetuning doesn't take many images

finetuning isn't dreambooth. it's native training
if you're fine tuning a model, keep in mind it was trained on laion first
https://twitter.com/Beekzor/status/1608862875862589441 i've got these people in my twitter feed , its some tool for finetunes (someone elses service but I'd imagine there's code around that can do similar things locally as per the earlier reply). I remember someone asking "how many images.." and the author answered to try with low numbers first (contrary to the usual expectation that DL takes huge numbers of images).
training base models typically does take huge numbers. refinement doesn't
right suposedly the original models cost around $150k+ to train, isn't it a billion images or something, but I assumed he isn't wanting to do that
....
I take the trained model and train it more on different images
right thats what I understood "finetuning" to be . I had the number "dozens of images" from the threads I linked above. its not "dreambooth". There's a lot of custom tools around , its all in flux, I dont keep track of all of them
Just finetuning stable diffusion takes more than a few dozens of images
from the above: ' “Fine-tuning” #StableDiffusion is the process for training custom AI models (with specific data/art - usually a set of 10-100 images).'
I dont know exactly what you want to do but from the given information, "finetune".. aparently 10-100 , hence , "a few dozen" , is a valid answer.
That's for dreambooth
they dont mention dreambooth. you haven't specified what you want to do thats different to what they do
It's basically not possible to fine-tune the current v2.1 model on 10 images and expect it to work perfectly
Source?
that it's not possible for finetuning with a small set on 2.1
The tutorial i just watched for stable tuner showed the guy using 200. not a small set but not exactly a large set. 100 with xflip
if you're training a base model, you need a huge set. if you're trying to refine the model into something new, like text 2 pokemon, you need a huge set. you can get good results on subject training for a low number of images still though. 2.1 may even be better at it since the latent space was specifically made to be more smoothed out
digging deeper the link I provided did infact use "dreambooth" behind the scenes, ok. I didn't know that. Perhaps something did actually change with 2.1, or maybe dreambooth was specific to 1.4, and/or maybe "stable tuner" replicates the tweaks done by 'dreambooth' for v2.1
stable tuner has a dreambooth feature, but is primarily just regular training finetuning
perhaps everyone is right.. "yes it doesn't work with 10 images". "yes it does work with 100..". I have also heard that more images in training can still improve the quality when narrowing the scope (so there is still a valid use case for 1000's of images)
(as soon as i get around to getting a new PSU I definitely want to mess with finetuning. I do have a 4090 here aswell)
you probably need to dial it in to get quality on smaller sets still
( https://discord.com/channels/1002292111942635562/1026382406279770152 anyway seems like one could get a definitive answer there )
it's worth experimengint on at least. the tools are all still so new. if someone is telling you for certain it can't be done a certain way, maybe they're just stuck on last week's release... if you understand what i'm saying
( I was going to say originally talk of fine-tuning seems far to on-topic for offtopic lol)
right everything is changing so fast
i basically want to use it for game-art.. finethuning on sci-fi film stills etc could help produce the imagery i'm after.
truly offtopic this time, my 1440p screen from benq, freesync model, has always had ghosting issues for years. If an image stays on the screen too long then the pixels sorta stick towards that image. not burnin since it goes away with some exercise, but it happens. since i got a new 4080, an nvidia card that doesn't do freesync with this screen, the ghosting is so much worse now. like a window in a spot for 5min leaves a ghost image of it
green lines especially
https://www.youtube.com/watch?v=IGWT-t7AS74 actual YEUHHHSSSS
The Second Crisis begins Summer 2023. https://apple.co/_Foundation
Based on the award-winning novels by Isaac Asimov, Foundation chronicles a band of exiles on their monumental journey to save humanity and rebuild civilization amid the fall of the Galactic Empire.
Foundation stars SAG Award winner and Emmy Award nominee Jared Harris as Dr. Ha...
was worried the pandemic killed that one
I made my own GPT model! Let's see what it outputs!
This is a test: fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox. fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox. fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox fox
find the periods
lesson learned: The transformers library is neat and relatively easy to use, but it's super memory inefficient
stands out like turd in a pool
(from finetune: COMEHU — "Today at 14:29i trained myself with 20 images in SD 2.1 and turned out very good, the most important thing is the dataset" ) .. I can imagine its a very hit & miss process .. but here's confirmation that some people can train 2.1 on (ballpark) dozens, for some usecases. maybe the result you're hearing ("can't on 10") was a worst case.
@chilly glacier Did you use dreambooth with this?
yes i used thelastben colab
@ebon haven
Thank you for answering so fast, was not expecting that!
Yeah! I'm trying to train with just the model and no dreambooth
thelastben have a new faster learning rate by default btw, i havent test it yet cause im waiting for distilled sd
Isn't sd v2.1 distilled?
I'm pretty sure it is
it isnt, distilled need like 2-3 steps for hq images
Can you try sd v2.1 with like 4 steps?
or 2
its gonnabe available in dreambot first and then released publicly
results are very blurry
Ouch, I thought 2.1 was the one that solved stuff like that
Apparently it's gonna be 2.2
Buutt, no need to worry, luckily stabilityai is pretty fast with delivering models to the public
when distilled sd is released im gonna leave sd 1.5 in the past
Yeah same
I want it
It should be easier to train on and it's so much better and faster
Do you guys know a good alpha matting ai ? I used to use ARC's Human Matting but doesn't work with large images
- the resolution, my face doesnt look distorted in full body images
Yeah
one i've used for a couple years now is remove.bg but its SAAS. it gives you back a transparent png but that may have the alpha mask you want. It's a bit of a run around but thats what i know of
A Mathematical Limerick (A Joke for Math Nerds)
A dozen, a gross, and a score
Plus three times the square root of four
Divided by seven
Plus five times eleven
Is nine squared and not a bit more.
(and written in python and for you programmer nerds)
from math import sqrt
a = 12 + 144 + 20
b = 3 * sqrt(4)
c = a + b
result = (c / 7) + (5 * 11)
print(result)
🤓
81.0
lol
Integral z-squared dz
From one to the cube root of three
Times the cosine
of three pi over nine
Equals log of the cube root of e
(A Calculus Joke)
from math import cos, e, log, pi, sqrt
def integral(z):
return z**2
a = 1
b = sqrt(3)**(1/3)
result = cos(3*pi/9) * (integral(b) - integral(a))
result = log(sqrt(e)**(1/3))
print(result)
🤓
Extra credit if you show your Math teacher/professor lol
(not important) the answer is: 0.16666666666666674
Just sharing to give someone out there a smile.
[ i & e are made up numbers ]
It worked impressively well, thank you!
For the German users in this discord:
Unser gemeinnütziger Verein braucht Unterstützung für ein Crowdfunding für eine KI Kunstausstellung, für die wir Arbeiten mit Stable Diffusion erstellen und diese dann ausgedruckt ausstellen.
Falls Ihr nach Weihnachten etwas übrig habt und ein bisschen beitragen wollt, ab 10€ Support bekommen wir pro Person 10€ zusätzliche Euro aus einem Fördertopf. Wir sind für jede kleine Gabe dankbar. Hinweis: Physische Prämien werden nur innerhalb von Deutschland verschickt!
https://www.schwarzwald-crowd.de/ki-ausstellung
only works if you pronounce Z like an American (American = "zee", elsewhere = "zed"), though
NBD 🙂
"no big deal"
hi, im looking for a tool (API)? to outpaint video. Created it 1080 x 1920 and i want to have it 1920 x 1080. So i thought I'll give a try with AI.
But I can;t find a tool for video, only for single images.
Can you help?
I'm Canadian and zed has long fallen the wayside up here. Always getting more americanized
I dont think anyone in the world says "Dragonball Zed" either
a
a
a
a
a
a
a
a
a
a
a
a
aa
aaaaa
a
a
a
s
s
s
s
s
s
a
s
s
s
s
s
s
oh
Isn't that a question that you should ask yourself rather than us? Everyone has their own mindset for each thing. For me, If I have a day off where I'm not with the family or have some extra time, I would go for playing some video games as that is what I like the most.
Hey, what's up?
People should be aware about false positives maybe not being false positives in the AI model sharing community. Lets consider attack vectors for a second. One potential attack vector that there is concern about, is that the ckpt files are essentially hacks. Hacks in the name of innovation always happen, but there is a compromise in using a tool for something it wasn't meant for. I don't think this attack vector has ever been exploited in any proof of concept yet, but security researchers are sure theirs a way and they're working on it. Safetensors, if i'm not mistaken, were created specifically towards this concern.
However, as with most network security considerations, there are many other attack vectors when it comes to file sharing. One popular one is appended data blobs. A file could work exactly how it's intended to, like a png or a ckpt. It can also contain a huge chunk of appended data in it. It would load like it should and all that blob would just be useless for the file's purposes.
A slippery trojan enters stage left though. It could be any user space executable. It could be a game crack or cheat. It could be boobs.exe (classic). It could be anything. It would scan fine for viruses and pass all tests. But then a rogue thread would spin up and it would find that saved file, know there was a payload appended to it, and load that payload into system memory in order to launch the actual attack. This likely would be one of many known or unknown privilege escalation attacks and then it progresses from there.
Stay frosty!
hi
got bord, made this in my terminal
using two files i wrote:
this one in my Scripts folder uwu.sh
and uwu-loop.sh in my home folder.
uwu-loop.sh, then executes uwu.sh
(for you non-programmer types)
(just realized i forgot to add BLINK env var incase someone was curious, so It's in the file now)
The comment is wrong for ALERT, it was red, but, cyan looked better, and yeah, i didn't fix the comment, eeehh
comment is wrong about the 1 second as well, that was too long, forgot to correct that too, lol, i suck 🤦
well, it's not AI art just bord nerd art
play them both at the same time 🤓
anyone else just code stuff when you get board?
also, be safe out there everyone (tin.foil.hat.time) a lot of perfectly healthy people are dying out of nowhere, so if you feel out of breath, notice your nodding out and think "why am I falling asleep?" or start slurring your words. as quickly and safely as you can sit or lay down, and close the gap between your head and the ground. everyday more random all of a sudden deaths are happening and no one has an answer (I think we all know wtf it is, but for now, I'll just keep it at that.) I just want everyone to be safe. If athletes and news reporters are falling over dead. Find that Zen inside to keep your blood pressure down. There's more I could say but this aint the place for it. Be safe. Shit's getting weird.
Lol.
I’m competing currently for a contest
If anyone could go and like my post I will love you for life
Ty so much
If you do
people always die for various reasons. qanoners are just signal boosting what fits their story. like 12 incidents are rolling round twitter
common symptoms that everyone have always had for vairous reasons shouldn't really mean one thing suddenly
the CFL guy that went to the hospital took a chest impact. that'll do it
he's doing good today. might be playing again after recoup
Is there a real-time AI voice changer?
Auto start stable diffusion webui with a single file on windows, no more copying the local host address https://youtu.be/DQCURdnS1XA
start stable diffusion using a single file, no need to copy the local host address every time.
@echo off
start "" "C:\Users\x\Desktop\AI\stable-diffusion-webui-master\webui-user - Shortcut.lnk"
timeout /t 28
start "" "C:\Users\x\Desktop\AI\stable-diffusion-webui-master\Webui.url"
nothing remarkable. maybe some filters out there that might be ML based, but i think its mostly pitch bending
i'm about to blow your mind. you ready? sit back || --autolaunch ||
LOL
since its off topic, yall like the discord theme i made
besides AI art, A big passion of mine is theming other software
nice. aww i use to love playing with rainmeter, might give it a play around with now, what with chatgpt for specific dimensions and SD with fun imagery
not sure what rainmeter is but sounds like you like it. 🙂
nice discord theme
Guess Earth had 2 moons for a short period of time: https://www.youtube.com/watch?v=wnqPqV6DdFQ
To Learn More Languages Check Out: https://speakly.app.link/Speakly
PBS Member Stations rely on viewers like you. To support your local station, go to:http://to.pbs.org/DonateSPACE
Sign Up on Patreon to get access to the Space Time Discord!
https://www.patreon.com/pbsspacetime
Einstein once asked whether “the moon exists only when I look at i...
we have a few natural moons right now even
one of them follows behind us at a lagrange point if i'm not mistaken
Guys, how i can earn some money with it ?
Greetings! What's good software to use to animate a picture of someone talking? I've got some audio narration, and a dog image, and I want even a cheezy video of the dog saying the words.
I think this might be more acurate: https://en.wikipedia.org/wiki/Nitroglycerin#Manufacturing
Nitroglycerin (NG), (alternative spelling of nitroglycerine) also known as trinitroglycerin (TNG), nitro, glyceryl trinitrate (GTN), or 1,2,3-trinitroxypropane, is a dense, colorless, oily, explosive liquid most commonly produced by nitrating glycerol with white fuming nitric acid under conditions appropriate to the formation of the nitric acid ...
Go to https://www.squarespace.com/nerdwriter for 10% off your first purchase.
GET MY BOOK HERE: https://amzn.to/3EPDQKt
Support Nerdwriter videos: https://patreon.com/nerdwriter Subscribe: http://bit.ly/SubNerdwriter
Watch the most popular Nerdwriter episodes: https://youtube.com/watch?v=_aFo_BV-UzI&list=PLwg4AG1KkgLx18HrK7lCOjRXZFpmrdkvV&ind...
https://m.youtube.com/watch?v=j7izhmrEwSg
Why download a car when you can hack a license plate?
And the company says they have the problem fixed. So, don't worry!
https://www.lehtoslaw.com
Music and cover generated using ai, give it a listen😬https://youtu.be/qRzc1o0XwJw
guys i wanna share simple one click temp cache cleaner app for free since you're all so swell, i wrote it for working with stable diffusion since sometimes it creates 10k+ temp files: https://ko-fi.com/Post/RFKTRsTempKillerv1-0-I3I5HVIAV
Started scraping all my AI Generated Images from multiple sources and used Zapier in combination with OpenAi to post new content (images and captions) to a new Instagram account to see if I could build a following without any effort, so far working pretty well! → @trickswithpixels
@undone sable Did you manage to get unprompted to work buddy???????
just asking i saw you on github @undone sable
Hi @brittle kiln unfortunately not, and I had to disable it 'cause in that state is useless. It's a pity, but perhaps it works for someone other. Thankfully I did the translation when it was still working, so if somebody from Italy uses it the localization will show up (unless the developer changed it).
damnit 😢
Just finishing to translate the latest stuff recently added to SD Web UI. Yes, but eventually you can play with Dynamic Prompts, or Style Pyle, which are good ones, if they match with your expectation or typical usage.
well i was told unprompted is good for creating bodys from faces that are masked @undone sable
is that true?
i was recommended txt2mask
i was to create generated images from faces that i mask
Not really sure it was specifically meant for that scope. txt2mask, that's another option for masking, or you can try with Detection Detailer extenstion, but it seems to be too much demanding in terms of stuff to download; I didn't activated nor translated into Italian.
I am not really super expert of all these tools so perhaps someone else with the same needs you have can help you better. Sorry. It's too much stuff to learn to use and I am more focused on the UI localization, even if this means that I have to read and learn how these tools works to properly (hopefully) translate them. Sometimes, I had to read inside the Python scripts the comments to the code to grasp the meaning of some cryptic acronyms, but that's the fun part of the localization job.
omg yeah this stuff is so confusing to me and theres just constantly so much stuff and new stuff being released, its very overwhelming to be honest @undone sable
You don't have to learn everything. It's like Blender 3D, you can't learn it all. Learn to use well only the Scripts and Extension that you really need, the rest only if and when you will need it. When I finish the translations I usually deactivate all the Extensions that I am not using, like Dreambooth, Animator, booru2prompt, Deforum, Embedding Inspector, Additional Networks, riffuzion, Dataset Tag Editor etc. I keep only those that I need like Checkpoint Merger, Prompt Generator, Image Browser and a few others.
SD Web UI is really like a Swiss knife. Otherwise, there is InvokeAI which is simpler but powerful enough to play with. And don't forget to use Breadboard to manage all your AI generated images, it's a great companion tool.
thank you so much! i really appreciate it! i just added you if you could kindly add me back ? @undone sable
😄
AI art, where the only thing cheesier than the subject matter is the artist (hint: it's a machine). But don't let that fool you, because these digital masterpieces will have you looking at corny Ai Art saying "I can't believe it's not human or butter!"
https://aiartinformer.com/
Ai Art Informer is an news and community hub for all subjects related to Ai art.
thats pretty cheesy lol @limber knoll
I know, right! LOL
u made it? @limber knoll
Sorry, made what?
that site @limber knoll
Yes, it's my site!
very nice did you use chatgpt @limber knoll
No, I'm a coder Dev. But I have adopted Chat GPT in my work flow.
better discord?
linux, steam, windows & amd meme
Enjoy
amazing chill (no singing) guitar instrumental, it's so smooth and her skill is fucking dope
https://youtu.be/SUt4wwvOCWE?list=RDMM
not into guitar's? here's a japanese super model that also plays drums to metal https://youtu.be/mVE_074PQro?list=RDMM
... ikr 😏
never judge a book by its cover
last 1, more cool shit you prob would have never heard
✌️
i'll get to adding those images i produced today to my gallery
(nylon acoustics are also my fav, great acoustic guitar for metal)
My guest is Rick Rubin, one of the most renowned music producers of all time, known for his work with a wide range of artists, including Run DMC, Public Enemy, Beastie Boys, Red Hot Chili Peppers, JayZ, Adele, Johnny Cash, LL Cool J, Slayer, Neil Young, Ye (formerly Kayne West), Tom Petty, and many more. He is also the author of a new book, "The...
do we have a name for the people or their prompts that just end up in the wrong channels? they make me giggle. you could be having a deep conversation on the nature of reality and another person will chime in with "fearsome horse dislikes apples"
unless they're spies and im supposed to respond with the correct coded message
totally off topic, it's music so, its still art so i'm not 100% off-topic so i think it passes
The Japanese super model that drums to metal just uploaded like 14hrs ago
#Metallica #LuxÆterna
Metallica is back !!!
▪ Instagram ___ https://www.instagram.com/ayeon3131
▪ Facebook ___ https://www.facebook.com/ayeon3131
▪ Contact ___ hmient01@gmail.com
▪ Endorsed by ___
Roland / Pearl Drums / Paiste Cymbals / Techra Drumsticks
I do like a girl drummer, never make it on a live set, even old warhorses croak from heart attacks but its cool
always the drummer goes first
female anything in metal is kinda awesome. i mean other than sexy voice metal stuff, gutteral screamy shit is amazing. there was an idea that when lemmy popped his clogs skin from skunk ananse could take over, i'd have loved that, she has such goddamn pipes
@supple shuttle true, but have you ever witness cute metal?
metal meets hello kitty
= Baby Metal
shows featured - 10:
LEGEND 1999
LEGEND 1997
BUDOKAN RED NIGHT
BUDOKAN BLACK NIGHT
MAKUHARI MESSE 2014
YOKOHAMA 2015
MAKUHARI MESSE 2015
SSA 2015
WEMBLEY
TOKYO DOME RED NIGHT
audio:
WEMBLEY
TOKYO DOME RED NIGHT
it's good, and yet still cute. weird combo, cute + metal. I don't get it. But. It works. ¯_(ツ)_/¯
The band I'm into atm is an Australian metal band, Once Human https://youtu.be/TtSjFoyfjEY 😍 I love this woman
I didn't make it playable because it's metal af
saw em live at download festival, honestly they went hard as hell
That's cool
other acts phoned it in and did their bits, they went HARD, like not my thing but better than the screamos that were playing on other tents that year.
everything beats emo
https://www.youtube.com/watch?v=M8-vje-bq9c their bizzaro world rival 😄
2015.7.29 ON SALES. LADYBABY 1st Single「ニッポン饅頭 / Nippon Manju」
(Sound) DL/Streaming:https://linkco.re/hGZc4BPy?lang=ja
(Video) DL/Streaming:https://linkco.re/91NfYMq0
Title:ニッポン饅頭 "Nippon manju"
Words:只野菜摘 [ Natsumi Tadano ]
Music/Arrange:浅野尚志 [ Takashi Asano ]
Guitar:ROLLY
Keyboard & Shout:HAJIMETAL
Dog-Shout:マルクス [ Marx ]
Sound Producer:成田大致...
what's the best grass to have?
Emo grass, it cuts itself
ok fine i'll make it playable, and yeah, i know of them ☝️
https://youtu.be/TtSjFoyfjEY
Order here: https://www.ear-music.net/oncehuman
Formats: ltd. colored vinyl transparent curacao, ltd. colored vinyl crystal clear, black vinyl, digipak-CD & digital
SCAR WEAVER - THEIR BIGGEST AND MOST LIFE-AFFIRMING ASSAULT ON THE WORLD’S SENSES
Four years on from their last album, Once Human are primed and ready to launch a fresh assault o...
no no, SCREAMO, the worst kind of emo. literally like death metal screaming but emo, and whining. actually the worst genre
my fav atm ☝️
Frantic amber is close second
Please subscribe:
https://www.youtube.com/user/franticamberofficial
Artwork by Ganbat Badamkhand
Lyric video by YOD multimedia
Single "Joshitai" from upcoming album "Bellatrix"
Album release dates
Digital August 9, 2019
CD and vinyl August 23, 2019
GMR MUSIC GROUP
www.gmrmusic.se
The Orchard Enterprises
Band: Frantic Amber
https://www.face...
A song about protecting your mother and sister from murder n rape.
pov you are the samurai
so if you like metal, you'll like these two
sounds a bit edgelord ill pass, i prefer absurd
wut
just because of my take on the song, and because you sound dumb af, you a person into metal, is gonna skip it?
That is the most re tarded thing i'v heard when it comes to metal
They way you made it sound upsets me, and i wont listen to it. --- wtf ever
later dude
not intimating anything about you by not listening, just a fucking strange opening gambit. some things are just not cool to infer or suggest
its metal, and you are acting like a pussy
stop it
besides, metal, is edgy, that's like the thing about metal
what would you like metal to sing about? happy sunshine mornings or something
dumb dude
anyways, i'm gonna end this stupid spat
-- over
(btw Frantic Amber is an all female metal band, so them singing a song about a samuria protecting his mother and sister from murder and rape, would be a females nightmare. so ya. u r dumb)
Just created a youtube channel of AI generated stories. https://www.youtube.com/@StorytimeByAI if you are interested to check it out 🙂
Welcome to our YouTube channel where we showcase captivating stories crafted by AI. Each video features a unique, thought-provoking narrative that will take you on a journey through the realm of artificial intelligence and its limitless possibilities. Whether you're a tech enthusiast or simply looking for a new form of entertainment, our channel...
without clicking any, explain please
Stories generated by AI, Simple
Hey folks, I have a question, is Etsy allowing the selling of AI art? I see they are closing down shops which display their work as Handmade... but there is no option to state that a listing is AI.... so this is confusing. Are there other printing services that are available to AI artists?
If you search for it there are a lot of Ai art on Etsy https://www.etsy.com/market/ai_art?ref=pagination is there an option for digital instead of handmade? or even digital art that gets printed is tagged as handmade?
metal has taken some cool roots in japan. here's babymetal drummer https://youtu.be/TfML1WEWfwk
he's a mascot for an apple farming town. he's an apple that's been posessed by the spirit of a dead cat 🍎🐱
Her name sounds Korean, are you sure she is Japanese?
ガチンコザホルモン~コッテリの継承者たち~
マキシマムザホルモン
過去の配信はこちらからチェック!↓
https://www.youtube.com/playlist?list=PL2110GL-5A_HzpNpZIpWBXJosL_Y27SZs
NYANGOSTAR HPこちら↓
http://www.nyangostar.jp/
yeah nyangostar has gone a lot wider than just babymetal. does all sorts of stuff now a days. rocks so hard
by babymetal you mean the babymetal genre or the band?
i thoguht that concert i linked was a band event but maybe its a genre. i am not knowledgeable
that apple cat rocks hard i know that though
I mistook it, the genre is Kawaii metal (J-Pop + Metal) which is what babymetal identify themselves as, but nyango star is not part of their band, tho nyango might be someone famous lol
nyango star is too good to be some random person wearing a costume, way too underrated.
whoever it is, they own all the rights to nyangostar and keep it secret
I like how nyango star changed the costume to make a drum cover of this babymetal song lol https://youtu.be/kFr4StoaYus
BABYMETAL
album「BABYMETAL」
PS.NYANGOSTAR
Hello、NEKO METAL*DEATH!
NYANGOSTAR HP↓
http://www.nyangostar.jp/
its funny but amazing at the same time
i'm not 100% but I thought i looked her up or one of her other vids and seen she was Japanese
#🏞|general-with-images message help pls
not necessarily off topic, but can anyone help me with this issue? nothing that won't let me generate still, but it is very annoying to see these hundreds of errors appear each time I click a button, each time a generation step happens, or literally anything at all happens. Anyone know why this happens; or better yet a fix?
File "B:\AI Generation Stuff\stable-diffusion-webui\venv\lib\site-packages\fastapi\encoders.py", line 160, in jsonable_encoder
raise ValueError(errors)
ValueError: [TypeError("'coroutine' object is not iterable"), TypeError('vars() argument must have __dict__ attribute')]```
are you trying to use robots to sell me a timeshare?
Hi there! Made a Hasbulla Dreambooth for a Hackaton 🤗 Leave a like on the model in HuggingFace if you love little Hasbi and want him to win the Hackaton! ♥️ https://huggingface.co/carlosabadia/hasbulla
"The emergence of AGI (Artificial General Intelligence) will lead to rapid developments in other technologies, such as fusion power, which will greatly decrease energy costs." -- if thats true everything gets cheaper, including commodities, land (because with fusion power you can pump water, mine more etc etc). But I dont personally think AGI changes so much compared to existing trends. it's just the general trend of increasing computing power that we're seeing here. we've already got 3d printers/CNC changing construction etc.
besides that "invest in land and commodities" is an existing theme which is consistent with a world of increasing scarcity due to fossil fuel depletion
21st century Hobo music (that doesn't suck)
https://youtu.be/Stf7bFlFqQs
ART
To stream or download the "Annoying" documentary film go to
https://projektor.com/u/xking/annoying846
or
https://exilekings.gumroad.com/l/AnnoyingDocumentary
To find CHAD FONTAINE T-SHIRTS OR HOODIES - go to
https://www.redbubble.com/shop/ap/102295153
FOR CHAD FONTAINE'S MUSIC GO TO
https://exilekings.gumroad.com/l/Chad_Fontaine
https://www.ex...
@pale bolt
yes, quit rare
some of those 11 years were a lot like this
I achieved the highest achievement, never once caught a drug charge or felony those 11 years
dances a jig
shit... u know
I believe that skill set is genetic.
lol
evasion 150, max limit 100
Is there limit for generations if yes whats the limit
Am new here
@rich mortar
idk what you mean, this server doesn't generate anything, unless you mean the bots, but I don't use them
I have Stable Diffusion running locally on my PC
I mean that isn't the server same as midjourney where we
Text to image through imagine prompt
- I'm not staff for this server, just a guy like you
this server, showcases what people make using that software
I hath been summoned.
and that's the same for all ai art discord servers
So far as I know, you can dreambot to your heart's content.
Can u plz mention the channel name for doing so
Scroll down until you see the dreambot channels. You should be able to see them.

lol, scroll down, and you'll see many "dreambots"
jk
Okay thanks appreciate it
if you want the real deal, you'll have to install stable diffusion on your PC @fast oak
Yeah
Is there any limitations for using
/ dream prompt
No NSFW. That means nudity, sexual acts, extreme violence or gore. There are some gray areas but just use your best judgment in the understanding that this is a family friendly server.
Don't try to be sneaky and skirt the naughty detectors. No "dress made of air" or "invisible bikini" or nonsense like that. If it's not something you'd feel comfortable showing to a 15yo, don't prompt it.
So far as I know, that's the gist. I believe there are time limits to prevent spamming. You may have to wait a minute or so between prompts.
I suggest familiarizing yourself with the rules just to be safe.
this is true for this server but if you mean for home use, depending on which web-ui or whatever there options to not block nsfw
I'm currently creating a Private discord server for the 18+ crowd. Just waiting to hear what incompetence excuse the web-host provider is going to tell me why my web-page/site to the discord server still isn't up.
Prompt Studios SD 18
stupid simple landing page to the discord server invite link
I bet you if I really wanted to, and setup a real AI SD gen site for adults. I would probably make millions.
for sure I would, but I don't have that kind of money to setup high end vps with high end gpu's
that must cost a fortune to get going
but sites like nightcafe are making hand over fist $$ right now
Good luck
@fast oak thanks, I have the discord server ready, the logo, and built the site, but for whatever reason, the webhost provider still hasn't got the site up
😒
Hosting name I mean provider name?
idk, some random one, i knew it was a bad idea but, for $20 i can talk shit to them directly
I've been creating several sites for years and like haven't faced these delay problems yet
Fir making it live there's no issue btw.
Until*
I was pretty stond and playing around with there logo creation tool, and I like the logo, so I thought i would be nice and NOT pirate what I just made
Name it if u can rn
and here I am, .... waiting
lol
like I said, i was pretty stond, like,... really really stond
Thats a very elite name domain
Logo.com is like onspot ngl
but the incompetence level is through the roof
U already know that these names aren't available easily
i know how to server a site without a host, but it's whatever, i was being nice, and didn't pirate what I made.
Competition is High af
For making it live first it cool
Then u can shift the site in future
i also have scripts to use virtual port forwarding with full ssl, and the time i get hi, and try to be nice and pay for what I just made, lol, it bites me in the ass 😅
oh well, it was only $40 but ..... I'll get my $40 worth in talking shit to there IT team

Logo.com i just visited
The website is cool ngl
its an alright site
but it should not take more than 24hrs to setup dns records and get an ssl
it's not like the 1 fkng page takes up resources
it's just 1 fkng page, lololol
funny thing is, i been a penetration tester the last like 5 years
not saying i would hack them and cause damage, no sir, i could hack them and fix there shit lol 😂
Ohkk
what i been saying...
I asked here and u just names i thought u own it lmao
named*
here's my main site https://www.cursedcode.com
What's the site name u have been working now
Which hasn't went live yet?
basedlabs isn't rendering correctly though, idk y, when i build the jekyll site and self server it works
Bruh u made a fucking mini dark web right there lmao.
when I do it through gitlab, it was working and then idk wtf happend
YES!
that was the feeling i was going for
validation, feels great, TY
U got the vibe check fr💀
😁
👍
😎 👍
What's the motive of this site
URS Portfolio or something else
cursedcode? nothing, i just use it for practice
Did u code the whole thing or template shit?
a bit of both
Cool
basedlabs page i made but its not rendering correctly
Hosting provider?
"follow the white rabbit" leads to another site i made
multiple
zoneedit, google, github, gitlab, tumblr
and I feed all of them into 1 sites
Charlie's video embedded I saw that
yeah, i could have just made sub domains but where's the fun in that
