#Stable Diffusion Prompt Generator -Updated!
102 messages · Page 1 of 1 (latest)
😂 that picture.
.

((masterpiece)), (best quality), (detailed), ball, Green, cat, fire
Ok
How to do ??
wdym
wdym ??
I speak french
what do you want to do

test
Does this prompt result work perfectly with Midjourney/image prompt as well?
This is not bad. Could it be expanded somehow, so that it also created a negative prompt?
just use your regular negative prompts
idk, i have no idea what midjourney prompt looks like bc i never used it, but if its like "word, word, word, etc.", then it can be used.
Beautiful 🥰
loool 😂 . you are like really into this stuff. 🤣 🤣 🤣 👍
what do the double brackets imply? ((masterpiece)) whats this about?
I have to admit, those pics do look really good.
it means the modifier is very important, more the brackets, more important it will be.
Can you try it without this "cat" thing? Those ears are a little strange,lol. I don't have a Stable Diffusion account yet. (but i guess i should get one... lol)
nope you dont need an account, you can run it locally in your pc, search "stable diffusion ui" in google for the one click install script, or you can use novelAI to run stable diffusion in their data center if you dont have a good computer(4GB+ vram), but its paid.
Where does this anime style come from? I don't see anything in the prompt. 🧐
oh wow.
It has a tail, too. 😳
Another example for today
:
This looks really interesting wow
Whats going on here
This is cool, how can you do it within Stable Diffusion?
@karmic cove we got a problem on this
When using a small line of text the Ai goes nuts lol
Start making HUGE list of stuff without adding comma between words
yeah it probably doesn't know what do to.
one solution is, you can use chatgpt to expand your small line of text to a more detailed one, its really good at doing this.
there is like nothing even lewd to it
In a world where people are not desensitized.
in a world where everyone is amish
i agree
I mean I can see his point, but it's a projection and should be self-realized more than condemned against the picture itself. Sitting on a surface with pretty clothes on is not "soft sexual" it's not ANY sexual; it's a person sitting down and being classy about it. I can see how some people may be attracted to this, because she isn't hideous or covered in a tent, but there is nothing bad about the image itself. Some people are attracted to people on a toilet who look 105 (thankfully for less, lol); but that has nothing to do with the picture itself.
This prompt system is brilliant. Really working awesome, thanks for sharing it
.
Sometimes the results are too verbose, but it's still producing nice images, lol
lol, you can change the generated prompts by removing some verbose modifiers, i found that this will make the images better
So far so good 🙂 actually good for stirring the imagination anyways
A good simple negative prompt with this I'm using is:
*signature, watermark, lowres, text, cropped, worst quality, low quality, normal quality, bad-hands-5, jpeg artifacts, username, blurry, artist name
download the bad-hands-5 embedding and place it in your stable-diffusion\embeddings folder or take that word out of it. (it's still hit and miss though)
Try having these open at work infront of everyone 🙂
Attention: For ChatGPT Plus users, please use the "Legacy" model for this prompt, it will generate higher quality results!
Need to activate PLUS to change modes correct?
yes
Hallo, I generated the ((masterpiece)) tags. Where do I put them to get da image
https://github.com/cmdr2/stable-diffusion-ui#installation
I just installed it and everything needed to make it work, its really cool.
Thank
@karmic cove did you use midjourney for the images?
I Use for my Midjourney Prompt Generation this ChatGPT Prompt:
Create a very rich and detailed image description in 450 characters for a deep learning text-to-image generator like Stable Diffusion. The image shows... [Describe here what the image, what should be generated]. Then separate the individual aspects of the image with a mark in the form of two colons one after the other.
ps: I am still optimizing the prompt. But it works already. The result is probably too long for Midjourney, so that he does not follow it completely.
no, i used Stable Diffusion
Hosted locally?
yes
Hobby
If you don't mind can you send a link to the one you're using, there are a lot of stable diffusion Uí on google
For Midjourney my prompt is well suited. It directly creates a working multi prompt for Midjourney.
Even if fine tuning is still necessary.
I also work with a local installation called Automatic1111. Builds on the official UI.
@quasi isle
Image generation is not available in this server.
To start generating images, click here.
I'm.. just so confused how whatever you used generated an Anime image without any mention of "anime"
its in the first image
a cardinal at the bird feeder
What’s a negative prompt? Things to avoid? What would you suggest putting in that?
Depends on what you want to show. For example, if you want a prompt from someone standing, you could get:
positive: standing, ...
negative: sitting,...
It is a dumb example, just to show the point of it
Dont take it too seriously
I'm completely new here. When I use this prompt, ChatGPT just gives me a detailed description of what I want the pic to look like. Am I doing something wrong?
its probably because they recently nerfed the free version's IQ, so the free accounts' model cant understand this prompt anymore, i will find a solution to this somehow.
sorry about that
Is that what they’ve done? I wondered why all my prompts had sudden gone haywire
That's a bummer. Hope they didn't nerf it too badly... 😬
What are the verbose modifiers? Is this in gpt?
probably, they even announced it in #announcements , its kinda vague tho
i meant in the prompt for stable diffusion
it forgot how to use overridden operators in cpp, cant understand this prompt now, also its less natural in terms of nlp.
this is all ive found for now.
Is this the AI model you downloaded from civitia? Sorry if I’m being thick I know how annoying that is. Love your art, you’ve got a real talent at this.
Ah ok gotcha. Thank you very much 👍
i will add "-Updated" in the title if i ever updated this to work with the braindead turbo model
Stable Diffusion Prompt Generator -Updated!
Updated, now it should work in free accounts!
Link to the top: #1072982848979611659 message
No problem! The update worked! Thak you!
I think openAI is messing with me, the old one worked for a bit today, then it stopped, just tried this new prompt, same results as I’m getting with the old one now, I’m on the free version atm so maybe that’s it I dunno. It was inevitable I suppose, stuff like this is going to increasingly more expensive I think and there will be an AI class divides, especially got the best AI which will only exacerbate the wealth gap even further.
well you can try again, sometimes it just doesn't work.
try opening a new chat, then paste the prompt and try again.
Hi
man openai nerfed free accounts was really a terrible move, the newcomers will consider chatgpt stupid and useless now, they will know nothing about the legacy model and just quit chatgpt
i will somehow make a tutorial on how to modify complex prompts to make them compatible with the nerf
how did they nerfed the free accounts?
Reduce token probably
yeah I compared the results of the three models and the one that gives the most detail is the legacy while the free version is awful right now
I think they changed it from the davinci gpt3 model (which chatgpt is running off of) to a lower down model (which also run faster hence the turbo name)
||My diffusion is not stable||