#🏞|general-with-images
1 messages · Page 141 of 1
great stuff as always
looks like something out of berserk
underwater cityscape lit by ancient crystals
ask the fortune cat a question...
aircompressor
New James Bond Enemy ^^
*hero
To dramatic 😄
In a world of trash he has found a treasure ...
Spoiler? 😄
yes panda without pants
? 😉
I'd spoiler Panda with pants ... cruel for the animal ^^
ok thank you
No need to spoiler everything 🙂
he spy
ok
I'd spoiler nudity or violence ... if it's not to hard to post here ...
nice
Has anyone had any luck generating high tech transformers like what is in the newest movies? Every time I try to do an autobot or transformer it always gives me old generic type transformers I can never get the super high detailed ones like in the newest movies. (Looking for this type of result in image provided)
this is what I end up with I have no idea what prompt to use to get the results I'm looking for been trying for two days with no luck
what prompt are you using currently?
These are generated with the extreme creative prompt: "detailed image of a transformer autobot"
yeah, most importantly you just need a model or lora that has the modern transformer images
I had not even thought about looking for a lora yet that helps with it I'm just using basic stuff but adding things like extra detailed armor, etc and nothing is working
that looks pretty awesome. I'm currently using automatic1111 since I'm super new to AI so still stuck on sd 2.0 not sure if that matters
this one is sdxl which is better than sd 2.0. also works on automatic
do you have a guide to install and use it? I've read about sdxl but I have not used any yet
just download that, put in your models or checkpoints folder for automatic and select it after restarting automatic (or hit the recycle button next to the checkpoints list in the upper left. Set resolution to 1024x1024 and put prompt and hit go.
should work
lol I tried that prompt and this what it gave me.
does the model matter? I've been using analog madness 7.0 and absolute love it with all the testing it gets the best detail that I've seen
ah man I'm out of luck I can't use that sdxl because my graphics card only has 4gb of ram I'm using a laptop with a GTX 1660ti graphics card
ah
it has 6gb actually but it won't ever let me use a 6gb model every time I try it says it can't be loaded
yeah sdxl might not be possibe for you then. I'd stick with sd 1.5. i don't have any experience with 2.0.
the juggernaut reborn is good for 1.5, and zavy remix is good for 1.5
Here was the prompt I did.
transformer bumblebee with god powers with hundreds of children riding on top of transformer bumblebee. God rays, laser beams, smoke, glowing yellow eyes. Perched on a mountain of spaghetti.
it actually loaded for me in automatic1111 I'm testing a simple prompt now is there a particular sampling method you like more for that type of stuff? i've always used the default dmp++ 2m karras
sure that's fine with 20 steps. you can always go crazy and do dpm++ sde with 20 steps for more quality etc, but it slows it way down. really depends on what you're looking for. (speed / quality) etc.
upscaling with hi res fix often does more benefit than doing very slow samplers. so 20 steps of dpm++ 2m karras with a hi-res fix of 1.5x at 0.5 denoise with the same sampler and steps.
works well more than it doesn't.
there's a whole world of samplers and schedulers etc, but that simple config takes care of most things.
There are often information about prefered samplers and setting on the civitai page ...
(for every model)
That made me feel different in one second
A porcelain doll, with blue and silver floral designs, traditional Chinese attire, looking serene, holding a delicate flower. Background features a classical Chinese interior with wooden furniture and soft, diffused lighting. Created Using: high-resolution photography, intricate floral patterns, traditional Chinese artistry, studio lighting, fine porcelain texture, detailed craftsmanship, elegant demeanor, hd quality, natural look --ar 2:3 --v 6.0
draw a cat
You draw a cat.
Not via discord here.
You download it and run it on your machine.
Or pay.
Thanks
$prompt full body,
天空中浓厚如墨的乌云剧烈翻涌,闪电如银蛇般穿梭其中,一条体型巨大、震撼人心的神龙从翻滚的乌云中猛然穿出,它的身躯长达数十米,闪耀着璀璨的金色光芒,每一片鳞片都如同精致的黄金甲胄,反射着耀眼光芒。龙须细长而灵动,在风中肆意飘拂,龙睛犹如燃烧的火焰,威严而炽热,散发出令人敬畏的神秘气息
what checkpoint could this be?
Good morning!
danke
Don't lose your head 😄
kids, here's your homework - draw how you spent your summer
what kind of pokemon is this
Putinmon 🙂
not like that 🌚
Berlin Vibes without doner kebab ok...
I have made some Junkies fighting for drugs Ostkreuz station ...
But let me check "Döner Animal"
fight is problem for ai , its one promt only?
One Prompt and IPAdapter Transfer ... but only from my drug pictures ... so no fight in there ...
A.I. is so innocent ...
she thinks that if someone is naked it’s good if it doesn’t hurt him 😎
she has no idea what she's doing ... so we can't blame her ...
double standards from ai
is it not possible to create images in this discord server anymore?
I think A.I. is the complete wrong name for it ... cause it let us think it's more than a stupid software knowing nothing about reality ...
Marketing
What's the best way to color sketches? ControlNet lineart doesn't keep initial sketching style
what checkpoint is this?
an army of anthropomorphic foxes, dressed in medieval knight armour, fighting for their lives against an army of squirrels, ultra HD photograph, sakuga animation cel --ar 16:9 --s 1000 - @vapid atlas (turbo)
here your image what you requested
kek
Marketing ... the root of all evil 🙂
12 seconds? Cool!
well it's just looped.
with 2x interpolation with the rife node
i see that krea has key frame based videos, but it's really just animatediff morphing. i should get that going locally
I saw that later 🙂 But a good one
Animatediff can be pretty cool. Haven't played enough with it yet
Hope it's not to hard ^^
hah that's really neat
devil may cry 5
Not part of the prompt and I don't really know that game ... but nice if it fits 🙂
I don't really know ...
can i do this real quck
if human I wouldn't ...
you could delete quick 🙂
Haha 🙂
Fresh body ... if you need a new one ^^
But for any reason looks like she's good in something different ^^
lol i realized what you mean only after a minute of thinking
she ate all the food with her mouth closed
hehehe
how you do text?
Old idea I got watching Altered Carbon ...
it doesn't include the prompt cause upscaled. But it's been pretty hard. In real I wanted him like a chicken in a supermarket
@clever oar - You probably should delete those. Not only is it likely beyond what the server rules allow from a nudity standpoint, those girls are borderline underage and you're on the edge of posting just straight-up illegal content.
ok
Hello guys Im searching for a job of anything md me if you are interested
" wide shot, male human body in transparent wrapping film with printed selling Sticker saying: "Fresh human body" on it, in a futuristic labor, cinematic shot, ultra realistic, high details, 32k resolution " sometimes it's easier to create a new picture instead of searching for an old one 😄
mojo your photo like my
is it alright to request images here?
I don't have access to compute rn
I just want someone to try and generate something specific as a challenge and because I want to see what the model will do
There are many many models ...
You can ask...doesn't mean you'll get what you are anticipating.
Or try SD3 for free here: https://glif.app/@Oliveira/glifs/clw44qfbl0000m0zztwqk2tnf
I don’t agree about the age they are at least 18 years old
I'm aware, I don't really care about the specific model I'm just asking if someone would like to generate an image for fun
The image is this, what if Apollo 13 landed on The Moon?
You can say that. But since it is what it is, you can't prove that. So the idea here is that you should be closer to the safe side than not. That's why I said it was borderline, not that it actually was underage material.
someone in another server said it looked like it was AI generated and I thought it'd be cool to see if you could make something similar
is this mobile friendly? I'd prefer to use SD1.5 on my pc but I'm overseas rn with only my phone
I understand what you mean
I really didn’t think that anyone could see them as very young
It's a webservice so should work on any device. only tried once on mobile and it was OK
🙂
Deleted my picture, too ...
aww
Trying fights?
yes 😎
two men arguing which sampler is better
😄
the angels may not like it 😎
Thank god both are male ... females I got where almost naked ^^
Damn ... I'm waiting for the results of the creativity contest ...
To scary ...
haha aww
AI is a really bad name and the understanding of what it means is biased by hollywood
encoded intelligence is a better descriptor
lottery on xl too. even where it is purpose refined
i saw txt xl lora
https://civitai.com/models/176555/harrlogos-xl-finally-custom-text-generation-in-sd this is a gooder. but it's a lottery even here. it really improves the odds. there's a few
that moment when you see an announcement... but it's not the announcement you were looking for :3
fwiw nasa has a ridiculous number of images available via an api
all labeled, and relatively high res to uber res
they don't let you just paginate through the entire database ( think they stop at page 100 with empty search criteria), but you can search for terms instead

My mom thinks I'm handsome...
I'd forget about intelligence. Cause we only know human one it's to easy to missunderstand.
she is right
@nimble mason with the amount of images you generated, you can make your own model i swear :3
or lora
almost all this is from a dora i trained
on my own outputs actually
so yeah you are correct lol
i've generated over 110k images since dec 7 when i installed SD
what is the main difference between lora and dora again? i only recently started training myself and still learning all sorts of things
this is zavy
nice
dora, at least with kohya, has more layers
includes convolution layers
i've found it to be a hell of a lot better than a lora for style
the detail is much better, it's sharper, better colors, etc
ah, cause im using OneTrainer and it works nice for me, but i dont know if they have dora support, will check
also, how does dora load in comfy?
is it same as a lora loader?
there's no dora support in OT, which sucks
no locon anything
no convolution layers
that's the one big downside of it
yeah, same as lora loader
it uses more vram for whatever reason, but i just save commonly used loras as checkpoints anyway
loras slow things down, if you save it you get the same output with no performance price
the perspective onto the Castle I quite like
Got o bed, Elon Musk. You are drunk.
Ani workflow idea to animate this picture?
going there with a videocamera? 🤭
This has gone a bit to far 😄
A friend wants me to do some A.I. stuff ... but for me it's a bit boring not really inspiring ...
nah, Looks just about right
i c
You can't get any idea about the original store ...
Kinda sandstorm simulator?
thats a videoscreen 🙂
dusty old Dune cinema 🙂
Mother in law?
lol
Google on the road again?
Preparing for Dune 3?
Werthers Echte
Yummy! 🙂
🤭
Mit Karamellfüllung 🙂
But I hope they don't taste that dusty ...
how about futuristic Toffifee? 😄
Could also be a Ferero Rochez ...
yep, in its spacesuit
it´s cinamon
Desert brulé
2 hours to early to make me hungry ^^
thats funny shit
Old one but I still love 🙂
😛
Preparing lunch now ^^
figured out how to do text in sdxl
consecutive seeds
generated 7 of the clownshark ones in a row
i'm using nothing but sampling
no controlnet, no ipadapter, no latent images as guides, not denoising or pasting anything
that's the workflow
first try
Not even SD3.
"AUTOMATIC CANT DO THIS" not "AUTOMA TIC"
but it was more in reference to everything together
Or "AUTOMATC ," 😉
Oh yeah!? Well,
It may not be able to that, but Automatic can this.
My prompt (typo and all) was: colorful explosion in space, tendrils, ethereal paint, life dorm
[Juggernaut, SD Forge]
2 Three-dimensional images showing the connection of life sources, including water pipes
SD3
a teenage boy wearing dark green ferrari jacket and sunglasses and black bootcut pants and shoes in Haku art style
SDXL vs SD3, there is a clear winner here. 🙂
SDXL/SD3
I bet you don't use the original SDXL?
Hell no, who does? 😄
Might be better to compare them ... cause you don't have a community finetuned SDXL ...
SD3 still did far better for prompt adherence. I don't have the base XL in my model folder currently.
For sure we can do like that for fun, too ...
I can't even imagine what SD3 is going to be like once fine tuned.
Mine was approaching that as well. I realised there were so many I never used, so I pared it down to about 15-20 models 😄
Clever ... should do the same 🙂
I wonder what would happen creating a model with A.I. generated pictures ... there are millions and we have the prompts ...
A few models have already done that. Can't think of any off the top of my head right now though.
It could become great or just nuts ... on the one side we have so many information ... on the other it's all A.I. generated
advertising board with the inscription COMFY CANT DO THIS
what is it
The Pearl Girl thing? A friend wanted it ...
cool c/m 😄
looks like a porn movie ^^
promt -comfy cant do this
mojo
I ate a pancake with filling for lunch and was very hungry in the evening, not as usual
whis was with meat
Ahhh... sounds better!
meat and vegetables
Sounds a bit like a Lahmajun ...
Do you drink beer with German sausages?
quality is higher than quantity for you😎
It's like with cola ... I know it's not healthy but if I really really want it, it's ok
bought a box orion chokopie for dinner
For my country it is impossible to download chatgpt 40 without changing the region
You are talking about using ChatGPT and not downloading?
Bad luck ... had to use a VPN to use it some weeks ago but now also available here ...
it had a pretty good picture analysis ...
I do understand 😄
how I suffered when installing SD and eliminating dependencies
Where are you from?
i told you before
The country with it's own A.I.?
need country?
I think I remember ...
you joker right now or sirios//
kandinsky ai
Would start with R ... 🙂
Ohhh... strange ...
The Roman Empire
117AD? 😄
Just found a map from the roman empire 117 AD
my friend was banned here I want it to be canceled
Yes ... sometimes the rules are not 100% precise and clear ...
There's nothing I can do for you. I'm just a nobody here 😄
😄
Yes, sometimes it can be hard to manage different people, different languages and different cultures ...
I'll probably never learn another language
sounds like a quote from the book you're a millionaire
is true but you cant control all
I've had french at school and learned some Español ... but after a while you forget ...
Yes ... I think you can lose the focus ...
But English works pretty fine 😄
the problem is that it is difficult for me to speak even in my native language😃
A problem I don't think I can understand ...
Not a very flexible mind for speaking other languages🤭
Come to Europe ... we have Pizza! 😄
i love pizza
I hope this wasn't rude for you. My humor can be a bit absurd ^^
its very soft after my friends
I didn’t translate correctly how after this picture I want to go to a cafe
Not to late for cafe?
😄 litlebit
Just finishing this episode of Star Trek Discovery and going to bed afterwards ...
Have a good night!
bye
SDXL image masked in Davinci Resolve.
Girl with a gun in her hand and a knife
The modest dwelling of a woman earning 5000 RMB per month, a cluttered kitchen with mismatched utensils, a threadbare sofa in the living room with a frayed rug underneath, a stack of bills on the table next to a wilted potted plant, a faint smell of dampness in the air, captured in a raw and unfiltered style, highlighting the woman’s perseverance and resourcefulness, through the lens of a Fujifilm X-T4 camera, 50mm lens.
Getting closer 😄
But no place to sit 🙂
Phew ... just offered 120.- € for 3 ties 😄
Kinda ... made with steel ... bought one when I was 18 and still love it ^^
at the moment of the nuclear blast
They mutated before? 😄
SD3 can't do any hands since yesterday ...
I like where you were going with this image tho. Nice job.
hello hello, I was wondering if anyone here could tell me how this anime diffusions were generated... is this local SD or one of those websites like Niji, Leonardo or Novel? which model could they be using?
its been more than a year since I last got into SD so I am pretty outdated, forgive my ignorance 
I would like to achieve this clean and classy anime style in my diffusions so could anyone point me towards the right direction? thanksss
You can do that too. But masking those edges can be a pain in the ass if you don't have the right software.
Plus you want to consider 1920x1080 video. Or 1440, etc.
ha I'm not the OP, It was funny I just created that
Yes ... looks like my problem ...
Why do python programmers always wear glasses?
style 😎
🧐
this correct 🤓
Because they can't C 😄
He ate to much?
me when i need coock something
dont like coock
On the floor?
fell
Here we have quite good frozen food from "Frosta"
2 strange 4 me
don't judge a book by its cover
hes friendly
and just wants to be loved
you hurt his feelings
now he's gonna cry
Yeah but I don't really know whether there's reflection in the water and where's the light from?
He is hungry ^^
i think he love cockroach sauce
that last one was with 0.9, this is 1.0
working my way bakwards now with v1.0 through the same prompts and seeds
i think the v1.0 one is better?
it's a diff style though
but does seem better i think
and we're back where we started today
Neverending story 🙂
Life of Po?
lol
woops, meant to do this on another chan, but enjoy the spam lol
i've been training multiple versions of the dora
YaY! 🙂
images before this last set were on sdxl base v0.9 fp32
0.9 vs 1.0
I'm good in pretending to read your spam ...
Thank god I'm an A.I. and don't care 😄
The less realistic a look the harder to say wich one is better
Otherwise we would have to read you mind ... and that would be pretty scary ^^
WTF SD3???
expectation - reality
alright nevermind, now i'm on my desktop with a real screen
the fp32 v0.9 looks really good
I zhink iz's more important to hit your expectation and not the one of others ...
fp32 v0.9 ... yummy ...
Is this the power of 2B SD3?
I don't really know ... using glif ...
But her right arm becomes her left one ^^
Mojo
off, on
im surprised it has that much of an effect on photography considering i didn't train on it
Its beautiful 🥺
So cuteee
lol
due to time strike, a blocker appears with a power of 10
power adapter icon
change to icon
SD3 and hands ...
#🏞|general-with-images a big city in to the apple
thx is frog made of uranium glass 😎
Mojo
discordos 🙂
YaY! Warm enough? Nice weathere here, too.
very warm water
by the beginning of June the water is warm as in the middle of summer
Sounds good ... I bet it's still cold here
I thought in Germany the climate was milder and warmer
OK, but water can be pretty cold ... cause it needs some warm days and in Hamburg ist always changing 🙂
#🏞|general-with-images a schnauzer playing guitar
usually this is not normal at the beginning of summer and the water is not warmed up
but then there was a slight weather glitch 
Big parts of Germany are flooded at the moment ... to much rain ...
lol
I'm living 20 meters above the water ... so should be safe ^^
Here is the image you requested.
Is there also kinda dog-dope?
Here is the image you requested.
Its china?
A large and menacing gargoyle stands guard over an aged electric guitar in a dark and eerie alley. He is surrounded by fantastical disturbing creatures causing a scene of surrealism. Dead roses and rotted thorns adorn the guitar.
#🏞|general-with-images A large and menacing gargoyle stands guard over an aged electric guitar in a dark and eerie alley. He is surrounded by fantastical disturbing creatures causing a scene of surrealism. Dead roses and rotted thorns adorn the guitar.
You need healthier food 🙂
I asked to take a photo of me 😃
AR tells me it's A.I. 😛
AR what is it
Aspect Ratio ... size hight and width
she not like cheese?
Here is the image you requested.
Preparing for hospital ^^
Reminds me of that German cybercrime guy in the USA ... Kim Kimble or something like that
I didn't point out that he needed a bra. 😃
A.I. was clever enough ... learned from the first picture ...
I already wrote this here, sometimes it seems to me that AI predicts what color I need and even ignores what you indicated
It sometimes ignores my complete prompt 😄
what a dangerous attraction XD
Yes ... but it's not a Pen** ^^
is?
It's a Hotdog ^^
i try repeat it
people on a roller coaster in a hotdog-shaped cabin scream in horror
ok ?XD
The first ones try to eat 🙂
😆
This is not a booth shaped like a hot dog, this is a booth filled with sausages and hot dogs.
It knows what you want 🙂
like i said XD
But this one is cool
sd3?
😎
Fast food restaurant ...
hey guys, do you know what checkpoints are merged to get this checkpoint? it looks really good
Space Ship One to Earth Control ...
sd3?
That one is ella - In vibrant, hyper-realistic pop art style, a sun-kissed man in a bright orange vest and sunglasses stands in front of a psychedelic-painted burrito truck in a bustling parking lot, surrounded by excited crowds of people of all ages, as vibrant, swirling clouds of burritos, tacos, ground beef, lettuce, tomatoes, and jalapeño slices erupt from the truck's smoke stack, illuminated by warm, golden afternoon light.
"stars in the background, surreal and fantastical setting, intricate details in the gold splatters and textures of the clothing, flowing river, empty field with a deep, blending horror and surrealism, softly blended shadows from the fading sunset light providing a contrast against the calm forest background, and meticulously detailed textures"
omg...
the creativity is off the charts now with these settings
It could have been so easy ... 😄
😄
I didn't know you were a sadist🤭
lady demitrescu
our precious LOL
leonardo AI still counts as SD?
It's based on SD using SD Models ...
it went far with models quality, so just wanted to be sure
Yes they are pretting good addinf features we are getting for local use
love their leonardo anime XL combined with alchemy... the effects are amazing at times
Yes ... haven't found out how to use some features offline 🙂
I'm not able to even use XL models properly (since low vram) so making use of leonardo for now...
btw that's in subject of monsters
Good idea! I think if you can't do it local Leo is a pretty good thing
all of these made with one model... btw colors are amazing as well... almost too much... Do I sound like a fanboy? I almost stopped using local SD host now... one downside is I have to pay for leo :/
Pretty cool. I'm part of the Leo Discord community ... you might wanna join 🙂
yea, I'm spamming there from time to time
lol
I wonder why SDXL always adds a woman 🙂
Yes
A.I. Webservice ... giving away some free tokens every day but paid afterwards
it's a site, they've got their own models, loras, and misc tools to generate online
looks like it, but you won't find the models elsewhere
Sometimes they just change names ...
anyway it's the quality of their models that I love
and some tools are amazing as well (like the realtime canvas)
Over it? Me? Never!
I would like this functionality: pictures in minimum quality are generated in seconds and if you like the picture it improves to normal quality
sometimes I wait a very long time to check the result
they have some very fast models, and realtime canvas generates in realtime, and you can draw as well... though this tool has it's limitations
heh, I recall the times when I made 3d renders on an old PC... and had to leave the computer on overnight so I could see result of a single image in the morning XD
nevermind... "very long time" is subjective
but I know what you mean
ok i am glad XD
you can test leonardo free, just log in, and maybe check all the tools, or maybe check out their videos before registering
never tried that one
After his career as TV superstar Bart Simpsons ended as permanent guest in a dark bar ...
What model did you use for this images? they look so good and cute, the last one is my favourite 🤗
all the images I posted were made with leonardo anime XL with alchemy
these are awesome
wait, you mentioned leonardo. so is this sdxl at all or a paid service with their proprietary model?
it's a dedicated model for leonardo ai site, and yes, it's paid if you want more creds, but there is a free plan as well
I wouldn't mind having a model like that locally btw
Yeah that output style is awesome
Hallo
Unbelievable I already made 1k SD3 pictures 😄
not that many, no? 😄
Many if you use SDXL and SVD, too
🙂
Hopefully not on the api. That's a good chunk of change
I have used a free online service.
Oh ok
I'm not really a fan of the way they act so I don't wanna pay
Need to sleep. Have a wonderful and creative day/night/whatever 😄
attack of the Titans
Mojo
I decided to stop eating candy to lose weight😃
I have reduced the sugar some years ago ... I don't really miss it. But if I want ... I drink a coke ... I just found out I don't really want that more often than 2-3 times a year 🙂 Have a good time! Ciao
bye