🏞｜general-with-images

[kindred depot]

Ohhhhhh!!!

Lemmie check

[copper matrix]

Since i also use swarmUI, it uses ComfyUI in the backend so i cant imagine bad results
https://media.discordapp.net/attachments/1319568881781837865/1348432723848724584/0009-IllustriousXLobsessionIllustrious_vPred-910492484-1.jpg?ex=68310056&is=682faed6&hm=dd23a760df4ba181533e9a70e52c5993e2189377d20cedeb9e0b00536e169170&

[copper matrix]

Bc of the ui*

[kindred depot]

how do i do that

?

[copper matrix]

See that 1 above the "image history" stuff

And remove the lora from the prompt box the <lora

[kindred depot]

1?

OO\

[copper matrix]

Change that 1 to like 0.7 or something

[kindred depot]

omg

OMG

ITTTTTT

FUKIN

WOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOORKS

NOW I CAN MAKE THIS THING

[copper matrix]

Applying a lora twice also does wonky stuff

[kindred depot]

thank you so FUCKING MUCH MAN!

[copper matrix]

No worries, and if you get stuck theres also a swarmUI discord if you need swarm specific advice

[kindred depot]

Ok 😄

ITS NOT A MONSTROSITY OF COLORS!

YAY!!!!!!!!!

now to make OC DESIGNS

[copper matrix]

You can also try playing with the lora weights

[kindred depot]

this works so much BETTER then forge!

[copper matrix]

Lets say if you add another style lora keep both loras but on lower weight

Some interesting results can happen

[kindred depot]

Question

Could I ai generate this?

It’s a drawing I made

[copper matrix]

Hmm maybe not exactly but i dont see why not

[kindred depot]

I’ll try! 😄

[copper matrix]

You'd need a lora probably

Or very specific prompting like limited palette, black and white, line art, glitch

Not sure since i dont prompt for that kinda stuff a lot

[kindred depot]

ok

[edgy ravine]

oh oh 👀

[vast basalt]

I am following this and i don't know why is not working
https://www.youtube.com/watch?v=ReXaksRMyag

[vagrant dust]

took your drawing at let chatgpt describe the image. Used the input as prompt for a text to image model....
Results 🙂

[vast basalt]

Should i use different workflows or put all them in one like this?

[jaunty prawn]

Shes cute! I love 😄

reminds me a lot of my OC

[wispy nest]

Haha hell yeah! Flux is intensive as heck. Finally used almost all of my VRAM 😄

[edgy ravine]

try HiDream ^^

[wispy nest]

thats so cool

people will be making independant movies from AI soon

[dry blaze]

Downloading civitai lora's be like

[viral frost]

Bird of chaos

[sweet beacon]

animated my uni 404 screen

[still rose]

Relatable except I can see my usage on the UI I use on Mac but I use activity monitor

RayFlux 3.0 AIO with Topaz Gigapixel for upscaling and then PixelMator ML enhance for colour grading and stuff

[wispy nest]

amazing i need this haha

[wispy nest]

Wow, jesus! I need this 😄 That's so cool

[median saffron]

woah, these looks so real, have you used the word "hyperrealism" in your prompt?

[still rose]

I have not

[median saffron]

can you share sample prompts that can generate these type of images

[still rose]

"a beautiful ultra close up portrait photo of an old man, angle view, intricate light, art by David Fincher"

"Medium close-up portrait of a woman with shoulder-length, reddish-brown hair, caught in a moment of restrained frustration laced with quiet allure. Her expression is composed yet complex—brows gently knit, lips curved in a tight, knowingPortrasmile that hints at unspoken thoughts. Her wide eyes, glassy and intense, shimmer beneath the soft, moody lighting, conveying both simmering irritation and a magnetic confidence. A delicate sheen of sweat highlights her skin, adding a raw, intimate texture that deepens the cinematic atmosphere. Despite the undercurrent of tension, she holds the camera’s gaze with a quiet, seductive boldness—as if daring it to look closer. She wears a casually styled shirt, unbuttoned at the top to reveal a generous glimpse of her cleavage, adding a playful, confident air. Her posture subtly leans forward, shoulders gently rolled to accentuate her figure without overt dramatization—more charm than pose.
The lighting evokes a painterly quality, with delicate highlight roll-off and rich, muted midtones. Every surface seems to hum with quiet contrast and soft detail, reminiscent of Kodak Portra 400 film—low saturation, creamy tones, and a nostalgic softness that gives emotional depth to the moment."

first is for old man and 2nd is the woman
@median saffron

[shrewd prairie]

Ai Nicki Minaj from Pluto

💕

[shrewd prairie]

Ai beyonce looking into her ai soul 💕

[median saffron]

ohhh so can we say that the quality of the image generated was not dependent on how creative you are in structuring your prompt, it can be as simple as the first one or too detailed like the second

[still rose]

Ye

[undone sorrel]

vbhnj

[viral frost]

😆

[edgy ravine]

boy's like ...

[bleak tundra]

need accompanied audio of a really silly and poorly sang theme song. you know the one lol

[shrewd prairie]

What is this made with it's really hd lol

Ai Nicki infront of birkinnnnssss

[viral frost]

Thanks. It’s just Flux dev with an upscaling workflow. ComfyUI workflow is attached to the image.

[thorn forge]

https://cdn.discordapp.com/attachments/1077488030849441802/1376660230502613063/eb144d383073491787f6f6e6f11afb2ctplv-fhsjxsyzit-aigc_resize_2400_2400.webp?ex=6836cb00&is=68357980&hm=c9a16d69b81347891b2676270b4bacb59f9328d4cefdf4cad5078c952e68767e&

[still rose]

got RTX 6000 24gb and can use big flux workflows yay

[clever oar]

again parasites?

🤢

[viral frost]

Naw, it's spaghett!

Little dude has a backpack full of it!

[viral frost]

That would be a cool pet halloween costume.

[clever oar]

the cat takes revenge on the alien XD

[cinder coral]

^Nice. I love that boomer LoRA that makes it look like the image you've generated was a photographed screen from a camera. Pass me the link.

[copper matrix]

Scam lmao

[shrewd prairie]

Hmmmm might have to try it

[solar tide]

Hey guys! Sorry to ask out of nowhere, but I have a favor.

I’ve been trying to make a LoRA for like 2 months now, and I’ve learned a lot, but I just can’t get any results. Right now, another person is helping me and said the reason I can't make a LoRA might be due to captioning problems. We’ve been going back and forth on many possible causes for my lack of success, not just captioning, but so far nothing has worked. Tbh, I’m just feeling a bit demotivated right now because I’ve put in a lot of effort and still haven’t seen results.

So my favor is: could you let me borrow a set of training images + captions that you’ve successfully used to make a LoRA?

That way, I can figure out whether the problem lies in my training images or captions, my parameter settings, or something else entirely. If I use your images and still can’t get a working LoRA, then I can safely assume it’s not the dataset and can cross one more thing off the list of possible causes.

I just want a taste of success. I feel like I’ve been heading nowhere despite trying hard, and I want to know what it feels like to successfully create a LoRA at least once, even if it’s using someone else’s training images and captions.

Would anyone be able to help me? (sd 1.5, automatic 1111)

[sweet beacon]

I am using this prompt for wan 2.1 manga schoolgirls classroom, black white line art, two girls standing, black hair girl and white hair girl, plaid skirts, school uniforms, detailed anime style, gentle breathing motion, light skirt flutter, soft breeze, eye blinking, subtle fabric movement, chest rise fall, minimal animation, preserve line art quality, pure monochrome manga aesthetic, no color, grayscale only, realistic physics and it keeps adding yellow highlight

[nocturne oak]

The word monochrome can mean more than just "black and white", it can also just mean something with tones of one color.

[sweet beacon]

oh

I think I will try " light skirt flutter, soft breeze, eye blinking, subtle fabric movement"

using fun in p model

[copper matrix]

Image to video? Or text to video?

[sweet beacon]

image to video. I am using this image

[copper matrix]

Ah image to video is rather easy in that case,
Try this as a prompt:
gentle breathing motion, light skirt flutter, soft breeze, eye blinking, subtle fabric movement, chest rise fall, minimal animation, realistic physics

Wan image to video doesnt need you to describe the whole image, just what you want animated

[bleak tundra]

also, add greyscale to the prompt, sometimes that reinforces black and white as the monochrome scheme. at least it has for me in the past. depends on the model though i supposed

bro you gotta do better

another one hits the ban list

[sweet beacon]

will do that

[sweet beacon]

still epic fail

the head tilt is like dafaq

[bronze yew]

生成一幅庆祝端午节的图片

[vagrant dust]

Your source image is not black and white but has some colors in it. Try to convert in completely to black and white (saturation = 0).

[sweet beacon]

Didn't even notice it

[copper matrix]

Well 2d isn't wan's strong suit but it's also trail and error

[earnest meadow]

hey guys, quick question: how do you guys use clip strength?

Afaik, clip strength of a lora adjusts the effect the lora has on the "word interpretation" of the model.

Should this be set equal to the lora strength or equal on all loras to have a balanced "mix" of the effect on the clip?

[sweet beacon]

edited it so the base image is 100% black and white

[shrewd falcon]

SolFlare From Kanchigai no Atelier Meister NoobAI Apex_Ai

[viral frost]

Bacterial actuary

[quick ravine]

io

[edgy ravine]

Ro presenting pieces of the 2590's collection - HACS Fashion TM

[royal solstice]

yeah, really enjoying Flux so far

[timid laurel]

Design a luxurious high-definition wallpaper sized 2560 x 1600 pixels (16:10 aspect ratio) optimized for a Lenovo IdeaPad 16-inch laptop. The background should feature a dark, highly detailed, and textured tiger face—majestic and powerful, with deep shadows to evoke mystery and elegance. Overlay the text ‘allwin ar.ij’ in a rich, deep red color using the Old English Cloister Black font. The text must be bold, refined, and centrally positioned with subtle shadows and highlights for depth and luxury feel. The overall image should radiate exclusivity and sophistication, with intricate textures and a dark, moody palette emphasizing every fine detail.

[sage isle]

Does anyone know how to fix this real photo with AI? Messing with compression and upscaling ruined the details of the flowers and vines

[sage isle]

How did u get so much detail on the flowers and vines?

[sterile kiln]

a simple img2img

[viral frost]

😆

This one feels a little ominous.

[proud glade]

chat

which models are the best for landscapes or stuff like that

(besides flux)

ang honestly id like to find model with style of this image
like semi realistic dark fantasy ish anime
cuz im tired of refining every image on low strength with 2dn latentlence
it has very strong style but it sucks so badly with background and everything else besides character if you generate image with only it

[bleak tundra]

Could check civit for a "Dark fantasy" or "fantasy background/landscape" lora and as long as it's the same base as whatever model your using just slap it in.

[proud glade]

just used 3 damn models, webside and local ui software for that picture
flux civitai generation with first prompt
80% refine with wai in my ui with second prompt
20% refine with 2dn latentlence

[proud glade]

is 10-20 images really enough

how do i train lora then
is civitai lora trainer works good enough

[elder tapir]

Hey guys. Can someone help me? I'm trying to place my speaker on a wooden platform, but it always makes duplicate of a speaker behind it. I'm using img2img inpaint upload

[edgy ravine]

😮 👋🏻

[tawny lance]

ahoy!

@edgy ravine Did my old haunt become the musigreg show?

[edgy ravine]

nah.. SkyWolf owns it basically

[tawny lance]

Oh I meant gen chat

lol

[edgy ravine]

nah .. he in the basement

[bleak tundra]

lol

[viral frost]

Sometimes I like to think about AI image generation like peering into an alternate dimension. People are tip-of-the-tongue familiar, but remain unplaceable. You can't quite tell what the purpose of certain objects is. It's normal for people and monsters to go about their business together in odd places. The written language is smeary and almost-readable. People randomly have a photo of a creepy child or a toilet framed on their living room wall. Beans will explode out of someone's mouth without warning. Birds fly underwater. It's a strange world, but much like ours.

[still rose]

rayflux 3

[balmy reef]

A surreal AI render with realistic lighting, featuring a robot and a giant mechanical animal head in an urban setting. Inspiration for this photo?

[grave sleet]

#🏞｜general-with-images

Draw a statue of an anime god : "raise the level alone" contrast photo with highlights, without background

[bleak tundra]

wrong channel there chief

[grave sleet]

Help me please! I’m don’t understand where is true channel for generation

[bleak tundra]

https://discord.com/channels/1002292111942635562/1237460360634830939

one of those

read the rules or whatever for it

it aint free i think, best of luck

[viral frost]

It’s just a randomized prompt! You can get surprising results. Prompt and workflow are attached to the image.

[proud glade]

guys pls help my flux doesnt understand anything i want from him :((((((

city is not on branches, dude is not grown up, only sky is kind of divided, tho its still not a crack i want

this isnt even fantasy

[copper matrix]

Your prompt could use some work the "this image is made in the style of" can go (just the style is fine), "grown up human male" why not say "adult man"? Try changing the crack in the sky part into "theres a crack in the sky splitting it in two, orange on the right, blue on the left, theres a sunset on the right side"

Should improve some aspects i would assume unless you got some finetune

[paper knot]

This error is occurring in Stable Audio with non-original songs

[bleak tundra]

dont click, thats spam and phishing

[paper knot]

Is there no way to solve this problem?

Like it didn't happen before, it started happening today

[hoary anchor]

but holy that's beautiful

[balmy reef]

love this one, mind me asking if how long did you render this animated photo?

[celest edge]

so im trying to figure out how to use a model installed from civic
why does mine look so bad
im day 1 stable diffusion so i really have no idea on what im doing

@wheat girder im new here but this is suspicious so i thought id report it sorry for the @

[solar tide]

@celest edge whats the name of the lora model ? I'll try it out. im really noob too, 2 months in and i get confused by how lora works. this is just noob advice from me, so confirm with others too, but maybe you should add better prompts, use trigger keywords, change the lora weight( e.g., Guts:1) (1 is the highest) and play around the denoising strength

[celest edge]

it has nsfw so idk if i should share it here
JANKU v4 tho

[celest edge]

for the prompts, i copied what the reference image had but it didnt give me same results and not even close

[solar tide]

@celest edge yeee ive had the same problem too, the loras i downloaded all look similar but not quite the same. we need to wait for a real pro to answer xD

[celest edge]

i uped the sample from 20 to 130 now
should i change any of the setting

not even close

[solar tide]

yeah thats the problem im trying to fix right now. is the artifacts is always different and idk how to fix it. im trying to "inpaint" it but its not the exact artifact used in the lora

@celest edge did you change your clip skip in settings to 2 btw?

[celest edge]

someone said to copy the generation data and paste it and i did, now lets see if it works

[solar tide]

im not sure what webui youre using im using sd 1.5 automatic 1111 but afaik youre not suppose to put the parameters in the txt2img

the seed for example you should press the thing next to the dice

[celest edge]

oh wait maybe i should put the settings instead of in the prompt😂

[solar tide]

you see "Seed -1" press the thing next to the dice and the number will change

yeah xD

[celest edge]

my pc crashed 😭

[solar tide]

xD i know how you feel my pc is very weak too

i wanna go to sdxl but thats just a dream

[celest edge]

it worked ig settings was the issue and someone helped me

[proud glade]

nah, fuck that
this is impossible to generate

[proud glade]

comfy users be like

[solar tide]

ohh congrats man !

[balmy reef]

Man, this shot’s clean,,, crazy detail, dope lighting, and it hits that perfect blend of real and cartoon. 🔥🎬

[proud glade]

does anyone know what the hell is DPM++ SDE CFG++

this sampler even exists?

[edgy ravine]

jup .. it's for defluxing ... get some nice sampling back into distills ... comfy extra samplers, i think ^^ (never personally used it cuz i don't do distills)

[proud glade]

are you talkin about this?
it doesnt has it

[edgy ravine]

i doesn't? 🤔 where did i get the thing....

look up comfyUI cfg++

https://github.com/CFGpp-diffusion/CFGpp

could be this ... it is the tech ... idk if it's the right one tho

[proud glade]

i already have that

maybe i just didnt installed properly cuz like yea

[edgy ravine]

was it the clownShark? ... the clownShark sees all

https://github.com/ClownsharkBatwing/RES4LYF

[proud glade]

ill check now

[proud glade]

seems like it wasnt

[edgy ravine]

shit .. i can't find it .. must be in a pack

[balmy reef]

I’ve been playing around with Savro for a few days now and mixing in some Stable Diffusion for touch-ups.

Still figuring out how to keep the skin, eyes, hands, and facial expression consistent but curious what you think of this one.

Open to feedback or collabs!

[solar tide]

Hello guys, can you tell me what is wrong with my pose for OpenPose Editor? I’m trying to create a "superhero landing" pose, but I can't get the output to match

[feral panther]

@solar tide Just make this guy wear a boonie hat. That's it. Just to test some stuff

a plain green boonie hat

[solar tide]

just to be clear smthing like this ?

[feral panther]

that or just a plain green clolor like this one

to make it easier

[solar tide]

so right now i can only get it to do this, seems the ai is having a hard time recognizing what a boonie is. imma try more tho and let u know how it goes. im also a big noob tho so maybe someone better can come along with better results xd @feral panther

[feral panther]

what is the example prompt you used? and for impaint, I think itse self explanatory that you have to erase the top of his head and then add a hat to it?

[solar tide]

for inpaint press "only masked" not "whole picture" unless youre changing background

higher cfg: prompt is more important

higher denoise: more creativity

you can use different checkpoint for styles that better match your image too

it also might differ for UIs

from my experience the avg cfg is 7 and avg denoise is 0.6 so play around that and lower and higher it according to your needs

[viral frost]

Flux does discount Willem Dafoe

[bleak tundra]

both spam, dont click

[feral panther]

is it this bad in this discord, do better mods /s

[solar tide]

Hey guys, I'm looking for someone who can inpaint this character's hand for me. I'm having a hard time and would really appreciate it. 😄 the 2nd image with the black hand is kind of my desired result it doesnt have to be exact. just want the guys hand to touch the floor naturally.

[proud kindle]

I am not so fond of doing stuff for free for random strangers. specially when said strangers didn't even bother to say the word "please" in their message. but I am up for sharing the love and teach you how to do it

[solar tide]

then dont. i said "i would really appreciate it" with a smiley face. i was being very polite. if you want to be offended then go do so

[proud kindle]

I offered you to teach you how to do it. if that is not something you are interested in then fair enough. just don't get surprised why anyone who values their own time wouldn't do free work for some one who is clearly just being lazy

[solar tide]

i dont want help from such a rude person. i asked helped in a polite way and youre criticizing me and now youre calling me lazy. anyway youre clearly being hostile to me for some reason so im not gonna engage with you

[balmy reef]

POV: You trusted your sibling with the pizza

[feral panther]

so I have been following this guide to create 2 characters in an image https://stable-diffusion-art.com/regional-prompter/#Installing_Regional_Prompter_extension

I've read this guide but not sure if its outdated or not, and not sure if it includes how to include extra prompts to add more items, settings, or actions. ON the example given here with the correct settings in that guide, I was able to create this sample image

with the settings similar to the break settings in the guide:
a man and a woman
BREAK
an asian man with black hair
BREAK
a woman with blonde hair

However, I'm not sure where to add extra prompts to change the settings, action, clothing etc. every time I add it, it kinda just breaks it

Note I'm using the default settings "common" recommended in the guide

[bleak tundra]

you would put the clothing stuff for each character in the section for that character. as for how to change background, you either put it in the first prompt section (the overall image portion) or change your regions to allow for more than just columns or rows, but a mix of both. its not super intuitive and still not a guarantee to get what you want (nothing with AI is, its all iterative) but it will help.

alternatively, you can get a flux model and just be extremely specific in your prompt. i dont use flux personally, but apparently its prompt adherence is top tier so

[feral panther]

so if I wanted the setting to be in a coffee shop, the man wearing tie, and the woman to wear a maid outfit. It would be like this
a man and a woman, coffee shop
BREAK
an asian man with black hair, suit
BREAK
a woman with blonde hair, maid outfit

[bleak tundra]

yeah pretty much

you can also insert character loras into each region and it should properly use them (more of an anime thing, but useful to know)

[feral panther]

character loras?

[feral panther]

also where would actions and character expressions go go (overall and individual). Like I want one to be happy and the other sad or both happy or sad. While actions would be a hand on the shoulder by the man and the woman cleaning a table.

[bleak tundra]

i believe expressions should be set in individual regions, as for posing, good luck. i know things like hand gestures can be set to individual regions, but to have the characters interact with each other is trickier. and for complex interactions and poses, you have to add controlnet to the mix which is a headache

[feral panther]

yeah i would think so, how about physical body characteristics. For example I want to make the asian man fat while the woman to be skinny just to test things out. So far this is what I have for testing (both are at a cofee shop, man is wearing suit + happy, woman is wearing maid outfit + sad) seems to be working okay so far
a man and a woman, coffee shop
BREAK
an asian man with black hair, suit, happy
BREAK
a woman with blonde hair, maid outfit, sad

[bleak tundra]

any characteristic of the character that you are trying to generate in a region, goes in that regions prompt. that includes hair style, hair color, body proportion, eye color, etc etc. think of each region like your prompting for a solo image, everything you would put in there for that character in the image would still apply here. be aware, the more lengthy it gets the more risk of "style bleed" where traits will bleed over into another region. thats where people usually use inpainting to fix it

[feral panther]

I see, the shorter the better? and impainting involves manually drawing whatever I need?

[bleak tundra]

you mask an area by drawing over it, and then prompt for the AI to replace what you covered with what you wanted to be there

you can mask faces, outfits, backgrounds, sections of the image, bad hands, anything

sketch lets you draw a mask a bit more specifically and prompt for something like :eyepatch" or whatever, and have it not affect the surrounding image in any way. if your mask with inpainting is too big or covers areas you didnt want to change, it might change those areas as well

[feral panther]

so if I wanted to mask the above picture to have the man be angry, how would that work

[bleak tundra]

you would mask his face, then in the prompt put the expression you want, and set your denoise to a medium amount so it will change the expression without completely changing the face

[feral panther]

when it comes to the format above, can you change the common prompt on top to be something else like (a male priest and a businessman or something like that) or would that have to go into the breakers

[bleak tundra]

the common prompt is the common prompt. you could put it as two people in a setting, and the put things like priest and business man in the regional sections, or you can detail less in the regionals by putting things like priest and businessman in the common

end result is the same, its preference

[feral panther]

so is there a good tutorial on that?

[bleak tundra]

youtube friend

its what i used for 90% of the post processing stuff with A1111 and Forge

[feral panther]

Yeah I just did a test with a random fallout 3 pic from the web.

first one isi before, one is after. I just did an impaint sketch over her lips to not make her smile.

a little bit stretched tho

I guess this is a good way to edit photos you already have/created from txt 2 image?

and by denoise medium are you talking about the denoise strength?

[bleak tundra]

yep. set it to like 0.4-0.6 so it changes just enough without completely changing the face. i.e. turning an asian man into a white man, etc

theres a bunch more settings and parameters that will help clean it up so its not obvious you inpainted, but youll have to watch a video and follow along for all that to make sense. its too much for me to explain here and frankly, i dont inpaint much

[feral panther]

ignore previous question, does impaint sketch discriminate between realistic and anime models? for example if I'm currently usinga realistic model and not an anime one, would impaint sketch not work on an animated non realistic picture if I tried to mask stuff?

[bleak tundra]

as i understand it, it doesnt matter what model you use. you should try it out and see what happens.

i dont use realistic models, so i cant really comment on the efficacy with them

[feral panther]

some don't work with original "masked content" settings and had to be set to fill. vice versa for mask mode.

Also sometimes I see a white/gray smear on the spot I masked. I assume that mask transparency needs to be increased, same for mask blur?

for example, I can see a noticeable white smear on the lips here

[feral panther]

Never mind worked with soft impainting with default settings, seems to be working okay now

[sweet beacon]

need some help I need to replace the robotic finger with real human finger then merge with the farm.png. then I can feed it to wan 2.1 to anime it like someone just receiving the conscription letter

[lapis basin]

Do you guys know what the art style is called where it's like water color, but not all of the image is colored. It's like splashes of water color that don't necessarily color within the lines? I'm looking for a prompt term for that, or if you guys know any Illus. models/loras for it

[solar tide]

@lapis basin im noob but maybe if you have an example image you can input it to chatgpt and ask what the artstyle is called

[lapis basin]

o that's a pretty smart idea ty

[solar tide]

@lapis basin you can also download a prompt reader to read what prompts were used to generate an image by someone else

[lapis basin]

o hm ok gotchu that's pretty useful

[quiet current]

Those infos are stored in picture's metadata. You don't need some "SD prompt reader" specific software. Your average picture viewer should be able to do it.

[solar tide]

@quiet current ah ok didnt know that. thats why i prefaced by saying I was a noob xD glad a more knowledgable person could correct me

[potent tangle]

I saw your complaint hit me up for more info!

[quiet current]

np, I'm not here to brag or whatever. But I'd rather people use known software rather than some sketchy unknown ones

[quiet current]

lol don't worry bot scammer, I'll hit you up

Time for the daily PSA, there is no "official external support discord" / no "official support" done in DM. Don't get scammed folks.

[proud glade]

why it takes so different time
its flux fp8

dow do i fix it

[potent shard]

A 2D anime-style illustration of "Zen Yi", a new character from Zenless Zone Zero, wearing an office uniform with a name tag, black stockings, in 8K resolution.

[bleak tundra]

wrong channel chief. try the artisan channels

[sweet beacon]

need to figure out how to anime this coz wan 2.1 is failing me

[fair saffron]

I gave ChatGPT these two images and asked it to create something based off the underlying styles

[sweet beacon]

you mean sora

[fair saffron]

I asked for two consecutively and got these

[fair saffron]

No… not sora. I don’t make videos

[sweet beacon]

I use sora to create that image I posted earlier

video feature requires payment and I was like hell no

[fair saffron]

Lol yeah you can make animations for free, given you have some semblance of hardware

[sweet beacon]

I have 3060 12gb, wan 2.1, hunyuan video and I can't quite get the animation of that pic quite right

like hand shaking upon reading the cosncription letter

what I generated so far

[sweet beacon]

don't know why it reminded me of neet army videoes like this https://www.youtube.com/watch?v=d895GqaTU-w

[balmy reef]

awesome smokey wolf, how u came up with this idea? what prompt u use?

[celest sigil]

Thanks, I got the idea from some of the images on Civitai and just adapted them. https://civitai.com/models/1330309?modelVersionId=1884429

[tawny lance]

hai

[feral panther]

when using the impaint sketch, is it pretty easy to create text bubbles? or is that dependent on model used or something?

[sweet beacon]

hunyuan image2vid prompt: shaky hands upon reading the conscription letter Dafaq is the new hand coming from

[delicate salmon]

Give me photo of india culture

[balmy reef]

appreciate it! how long did the process take? was it time-consuming?

[celest sigil]

No, about 1.5 to 2 mins

[quick helm]

3 beautoful women wearing a white oversize hoodies by the beach and rocks by the sea in Sweden, windy weather. Full body image. Natural light. during golden hour, taken with a DSLR camera, using a 35mm prime lens with cinematic lighting.

[jovial solar]

https://tenor.com/view/low-tier-god-banned-mods-smite-gif-4866200577432974147

[neat hull]

#👥｜roles电脑照片

[edgy ravine]

like... does this server have mods at all?

[vital blade]

@balmy reef 生成几张关于取快递的AppLogo

[edgy ravine]

@Moderator

[latent tree]

Can i put all same and replace this image with real doctor nick face...

I dont know where this indian guy made this....probably he ia here...but says me he dont know auto or forge ....how did he made this

[viral frost]

What the heck is up with all these Mr Beast scams?

[balmy reef]

love this one as well, would u be open to sharing the prompt or idea you used to create it? i’m curious how you approached it.

[celest sigil]

character intro of a {male|female} White Mage {|with white hair} in a society where one’s status is determined by dreams, shallow depth of field, action dramatic amazing composition, Pale __colour__ colour grading

[charred knoll]

Minimalist devil head sports logo, sharp horns, glowing yellow-orange eyes, flat design, red and black colors, dynamic composition, no text, hockey team branding --v 6 --style raw --ar 1:1

[hollow dagger]

AI image generation is a constantly evolving process, and I am really enjoying learning new techniques. You can see what 6 months of self-education and continual tinkering has done for my art.

[balmy reef]

is face-swapping more reliable than generating the same face from scratch each time? I tried using savro recently, and it kind of handles that automatically gives you a consistent AI model right away not perfect but so easy

[torn gulch]

generate a image of an alien

[inner surge]

Me: "Generate something random…"
Nimbus AI: "Bet."
[insert that pixel eclipse image]
Me: "What is this? A moon? A planet? A pizza? A glitch in the matrix?" 😂🌀

He Can Also Make Sum Better Images to 🤔 I think

[vast basalt]

first one is with vision x vace, second one with veo2, it is possible to generate something like that in local?

[vagrant dust]

Wan2.1 image2video and rotation Lora would be my approach to generate such videos.

[vast basalt]

but i need the bg to be rotating also, the loras will make only the character rotate like this
https://civitai.com/models/1346623/360-degree-rotation-microwave-rotation-wan21-i2v-lora

[vagrant dust]

https://huggingface.co/Remade-AI/360-Orbit

[hearty violet]

@sullen perch on this opions:

• fill: best to remove
• original: to correct
• latent noise: to substitute
• latent nothing: to create

@sullen perch some models like this one that I use a lot, on the page give all models and I only choose, like normal or inpainting

but always when was a inpaint model it will be writed

[balmy reef]

thiis is cool, ur art has that some next-level engineering right there👏

[night crescent]

why am i discovering this months and months after i started using it?

lol i'm dumb

[still rose]

Question on sum isn’t really tech support but people using chroma what is the “best” sampler and scheduler combo to use for it?

[balmy reef]

this kinda weird and cute at the same time, btw like ur art style :>

[flat sand]

really cool artstyle. Is that a specific model/lora?

[formal tiger]

Hey everyone,

Newbie here and was hoping for some advice on image restoration.
Does anyone know how a result like the 'After' image (on the left) can be achieved? I'm amazed by how it restores details like the skin texture, mask marks, and even the subtle rash so perfectly.
I've tried a few methods myself (like img2img with 4x-UltraSharp/R-ESRGAN) but can't get anywhere near this quality.
Any tips on the possible workflow or models would be a huge help!
Thanks a lot!

[vagrant dust]

Would suggest you look for detailer (upscaling and generating new content)

[celest sigil]

It depends on the type of image, workflow and your personal taste. Euler/beta seems to be fine for most things.

[still rose]

I see I’m not used to comfy at all so I just use the template workflow for chroma u can get in comfy

[charred grotto]

hey,

am new here, can i get a hell yeah!

dang it, those rules are pretty relaxed.. i think i will fit in here nicely 🙂

[tawny lance]

So, long shot here, has anyone messed with HSV frequency separation nodes for Comfy?

[edgy ravine]

frequency? ... like .. light? ... separate blue->do thing?

[tawny lance]

Here I'll show

High frequency

Low frequency

Combined

[edgy ravine]

aaaah .. DETAIL frequency .. interesting

[tawny lance]

(with upscale)

[charred grotto]

ahh right, like that is it..

[edgy ravine]

yours? ^^ ... #🏞｜general-with-images message

[tawny lance]

Yeah it's private to like 3 people, I just made a shitload of stickers and emojis

Ah, the jokes from the good ol' days.

[charred grotto]

those we're the days..

[edgy ravine]

make a pack ... so ppl can use it ... if that is a thing on discord 🤔 .... they good ^^

[tawny lance]

I heard Civ's in the shit now and the general SD IQ has gone down a lot.

[charred grotto]

i often want to go back also, so many things i would change, in other peoples lives.. my life was cool af though.. 😄

[tawny lance]

I MADE MODELS:

They're absolute fire right now

[charred grotto]

aight..

[tawny lance]

Yeah going back in time would be fun

I'd have words with myself if I could

[charred grotto]

do you guys have a community partner, who frequents here?

[edgy ravine]

they come for our toys .. civ is just a symptom ... also has an identity crisis .... fight for the people🤷🏻‍♂️lay low and wither ... but ... this is the WRONG place to talk about the people ... closed source SAI and all 😄

[tawny lance]

I don't even frequent here

[tawny lance]

Good times with Visa®️

[edgy ravine]

rely on crazy - get fuc**d

[tawny lance]

yeah.

[charred grotto]

im not one to argue..

[tawny lance]

So what's going on in the server that I was banned from (via underhanded means)

[charred grotto]

i dunno i was banned also

[edgy ravine]

via public mud wrestle you mean 😄 not much ... lots of complaining cuz they have to downscale all the time ... they sprung the trap set up years ago ... and i told everyone ... too loud ... too quiet .. whatever makes ppl not listen 🤷🏻‍♂️

[charred grotto]

always, ppls think they know best without even trying shiznat.. it is the way..

[tawny lance]

WHYYYYYYYY

Is Lin still going insane?

[edgy ravine]

Lin? ... OH.. was she ever sane? ^^ ... not seen her in a while ... Divine is absent too

[tawny lance]

Divine? Really? He's the staple that holds everything together!

[edgy ravine]

right?!

[charred grotto]

tbh, i don't know you guy's im just interjecting.. .. i'll leave you to it..

[tawny lance]

Yeah we just know eachother pretty well; more or less a chance encounter

[charred grotto]

🙂

[tawny lance]

You can stay, I'll allow it

[edgy ravine]

those are great ^^

[tawny lance]

lel

[charred grotto]

thx

lol

[edgy ravine]

Mars pulled his shit off site O_O ... he vanished too

404 on Strigoi

[charred grotto]

nice to meet you all anyway. 🙂

imma character guy, i like stuff.. without using lora..

lot's of stuff, prolly like you guys..

no doubt, you guy's can teach me how to do better..

so i made my own wf, and i make my own nodes, im all newb with the nodes stuff.. but im slowly learning

im into realism, but this stuff intrigues me tbh..

nothing is complete without one of these tbh..

❤️

i would love to learn from someone.. who know's what they are doing 🙂

couple of..

think i called this one becky from the block.. can't remember, i always forget these girls...

there has been so many!

nope, it's jackie... sry becky..

nice tats though Jackie...

ahh, becky, there you are, with that cute smile..

teddy can be ur best friend, WE love teddy.. look at that cute smile..

🙂

last one go nuts.. 😐

[charred grotto]

title says "It's general chat, but with images! Drop an image you'd like to chat about here and get that discussion going!" well, please do 🙂

don't let that i'm new here bother you.. imma nice guy..

2nd's 🙂

they are flux no lora.. 😉

oops last one is good though.. did'nt mean to post that, but have it anyway..

crit the shiz outa all of that.. 🙂

but then i said, NO jackie, the speed is not worth the quality.. and to this day, i stand by that jackie..

[charred grotto]

nice, 6 months time lapse really helped here.. "good journey" 🙂

took me 2 months from start and ongoing to get my character prompt something good..

random early stuff..

🙂

he's cute, he is furry!

top 5 images ive done..

did some tay tay..

hulked it out one time.

did a girl with a gun.. not perfect

tried this...

did these one time

was found wanting

so did these

tried this..

and this

and this..

did some of these

liked this one

from another guy, i did this..

i did this,

from someone else got this prompt.. gave it a go..

eek

did this..

did this also

did all these

did these

messes about n did these

couple of things..

last ones..

the prompt is strong with this one.. 😉

wow, i just did that wall..

[charred grotto]

what they talking about?

[viral frost]

Inst ication difthon trat momther yoller visch, obviously.

[charred grotto]

obviously

i got you.. 😉

in english, prolly means, that fkin ant man n iron man taking all our glory!

again!

prolly not, but hey, we make make stories where we can..

my whit is my own worst enemy, :/ you either get it or you don't.. it is what it is, ill try tone it down somewhat..

so i figure, if i can get one good character from an image its all good, i can cut that char out of the image and put it into another image..

101

should know also, every image i post you can have they low rez, and 2nd's..

here is a few high rez, :/

2 more

i just said to the Mrs, i've found us a new home, She rolled her eye's and said "oh god, here we go", it is orrible though is'nt it joining a new place where no one knows you..

[whole cliff]

hi

[charred grotto]

oh you are here!

[whole cliff]

yep im here

[charred grotto]

ok les chill here for a while, see what these peeps have got..

[charred grotto]

god!, i hate starting a new..

[charred grotto]

nice images but, so over rated.. 😉

were flooded with this stuff..

[charred grotto]

grrr

i can post scary images like this cant i?

[tawny lance]

🤷

[charred grotto]

i think so..

[hearty violet]

Can I rotate the building on this image? I never did this kind of stuff

[charred grotto]

it's just stupid stuff.. not provocative in any way..

[charred grotto]

with wan you can, not image gen, but you will have to do it many times to get a clip you require..

me personally, id use wan, rotate around it, take the last frame, control net that last frame, get the image required, upscale it..

wan will prolly require multiple gens to get the correct dimentions and end image..

someone else might know how to it better.

[hearty violet]

what is wan?

[charred grotto]

video..

wan2.1 is free for you use locally..

[hearty violet]

i don't know about it, I will make a search 🙂 thanks!

[tawny lance]

[charred grotto]

you can either use comfyui or pinokio

yeah, what he asked, doing clips is intensive... and takes time..

[hearty violet]

man my noteboook is poor

if I use was be renting a GPU

[charred grotto]

5 second clips on avg on normal mid range game gpu is 5-10 mins.. per gen

[hearty violet]

I cant do it on forge?

[charred grotto]

clips is diff to image gen

[hearty violet]

I never did a video, I can be done on a11/forge/reforge or only on comfy?

[charred grotto]

you can, but i saw you said you on a laptop... so prolly not for you..

[tawny lance]

Comfy is just... Leagues above any video offering forge has if I'm not mistaken at this time.

[charred grotto]

it is

[hearty violet]

I rent on vast

[charred grotto]

see if they support clips.. either wan of veo or something else

[tawny lance]

goat.

[charred grotto]

with wan, you can 100% rotate around the building,

there are other models tailors for this also, which you can use the image and use blender,

but you gotta have the system to support this..

a laptop wont cut it..

[tawny lance]

he uses Vast so it doesn't matter even if he has intel.

[charred grotto]

i think that one was against the rules

[tawny lance]

https://tenor.com/view/what-are-the-rules-iasip-gif-27514763

[charred grotto]

No harassment, hate speech, rage baiting, or discrimination.

No spam, excessive self-promotion, irrelevant links, or repetitive posts.

No NSFW or explicit content. it is said..

i just wanna be good here. one ban from a server is enough for me.. lol not for images..

just.. talking

[charred grotto]

hold on i got you

some early ones..

[tawny lance]

:O

for like, the week

[charred grotto]

do you like war hammer?

[tawny lance]

Sadly I never got into it

[charred grotto]

😦

[tawny lance]

I'm spoiled by a few notable names in SD who make a ton of 40k content

[charred grotto]

lol

[edgy ridge]

where did u guys learn how to properly prompt, or just test and error

[charred grotto]

test n error for me my guy

[tawny lance]

Mostly by talking to people like me. I just chucked a prompt into the ether, people tore it apart and I learned.

[charred grotto]

i looked on civi also for images and took some of thier prompts and amended..

[tawny lance]

I mean, that's another way but it doesn't teach teach

[charred grotto]

once i get a good prompt, i use co-pilot to enhance it..

[edgy ridge]

thats what i was thinking, only started today so a lot to learn

[tawny lance]

And frankly

most prompts people post on my wall are just deadass wrong

like they use pony tags in illu

no booru tags

[charred grotto]

ive done alot of my own prompts trial and error, once i think i have it, i make gpt enhance it

then move things around withint the prompts.

[tawny lance]

so it also depends on the model you're using and architecture

[edgy ridge]

i was trying to do super realistic but couldnt make it work hahaha

[tawny lance]

Yeah that's fair

You gotta get the right model

[charred grotto]

i straight ass just use flux with no lora.. so tags aint in my prompts.

[edgy ridge]

do u know a good one

[charred grotto]

2 months in, im doing this..

[edgy ridge]

i downloaded the realistic vision from civit

[tawny lance]

Unfortunately realism isn't my bag... Let's see.

[charred grotto]

just random stuff.

[tawny lance]

this one looks okay. StableYogi is a bit of a hack but sometimes he comes out with gems

https://civitai.com/models/974693/realism-illustrious-by-stable-yogi

[charred grotto]

realism is easy, getting what you want to be real is the hard part

[tawny lance]

add masterpiece, best quality, newest, absurdres, highres, to the front of the prompt and then prompt what you want

and if you're really trying to get gud, read the model pages, just some of it

[charred grotto]

just char..

[tawny lance]

Illustrious is where it's at as far as model types

or NAIXL which is just illustrious+

[edgy ridge]

thanks, and another thing i never understood is how do u "save" a character u create

for example i get a result i like and how do i work it

[tawny lance]

You make a lora of them or recycle elements of the successful prompt

[charred grotto]

i want all my images in this format..

[tawny lance]

eg, if I use one model, and I prompt masterpiece, best quality, newest, absurdres, highres,

1girl, white hair, black lips, eyeliner, backwards hat, white hoodie

I almost always get this girl:

[charred grotto]

like this format..

my guy, you do know you can save an image from here to yuor pc, load it up in comfyui and produce the same images?

my images implicitley..

[edgy ridge]

okay that makes sense

[tawny lance]

I sure do:

[charred grotto]

do you want my nodes?

[tawny lance]

Open picture in browser, drag to comfy grid

[edgy ridge]

i was looking for sort of get 1 general image and then keep the character consistent into other images

[charred grotto]

rrd nodes arent in the manager yet, so i can post here if you want them..

[tawny lance]

nah mine's a bit more... robust.

wait wat

oh interesting.

[edgy ridge]

and the lora concept was confusing since when u get 1 image its not enough for the lora to learn i assume

[tawny lance]

I would have just used impact wildcard processor

[tawny lance]

correct

You need like... 20-50

The absolute best you can get

[edgy ridge]

so i would need to create a lot of images close to what i want and then the lora would need to get the general idea and get its own result?

[charred grotto]

so now knowing that, you can grad an image from here, and slot it into comfyui, if they use comfyui, you can dl the required, and reproduce that image or similar..

[tawny lance]

yes, but there's a bit more of a process

[edgy ridge]

yea i got a lot to learn

[tawny lance]

You need to learn how to "tag" properly as well

[edgy ridge]

tought it was going to be simpler xd

[tawny lance]

I feel ya, but the journey is a good time imo

[charred grotto]

only tagging for lora though..

or use lora:theloraname:1.5 etc

if you aint using a lora loader.

[edgy ridge]

im not using anything for now hahaha

[charred grotto]

if you use a lora loader, you can set params in that and not worry about tags in your prompt..

[edgy ridge]

was just doing some prompts to test

[tawny lance]

we're talking about making loras

[charred grotto]

nods

i'll shh then, thought we we talking about lora tags in prompts..

apologies

[tawny lance]

both

really

[charred grotto]

😐

[edgy ridge]

ram do u mind if dm u?

[charred grotto]

no ill shhh, i want to learn this also

[edgy ridge]

dont want to flood the chat with baby questions xd

[charred grotto]

we all here to learn my guy

[tawny lance]

I'd hardly call this a flood

[edgy ridge]

and so i could send some examples of the idea i was trying to achieve

[tawny lance]

I'll flood this place by myself, you can ask Maxfield

[charred grotto]

which point are we upto thern, the start?

[tawny lance]

Psycho meter is at 0%

[charred grotto]

copy that

les go then, cause i really want to know how to do lora also..

[edgy ridge]

hahaha

[edgy ridge]

the problem is i dont even know where to start the more i tried to do the more questions i had

[tawny lance]

But yeah with loras for characters, I was told to tag everything you want to be changeable (clothes, accessories) and untag everything you want to be consistent (hair, eye color), stuff like that.

Basically it goes like this -> Load your images -> Tag your images -> Bake your images

[charred grotto]

cool 🙂

[tawny lance]

civitai.com has a good lora trainer.

[edgy ridge]

ah that makes sense

[tawny lance]

they also kinda walk you through it

[charred grotto]

ahh

[tawny lance]

bbl

[edgy ridge]

can i send the sort of result im trying to achieve to you? we can discuss the rest here, its not anything nude but prob to much for here

[tawny lance]

sendit

Just don't send me anything sick

lol

[charred grotto]

like my old prompt ```(MH_ImageType) (MH_GenderType) wearing elements of (MH_ClothingTypes) clothing.

Use realistic human proportions and accurate anatomy with realistic hair, hair highlights, hands, stout chin, body, face, skin, eyes, feet, include subtle imperfections like freckles, blemishes, scars, warts, spots and wrinkles. ensure accurate, detailed and intricate clothing, nothing in hands. Ensure they are photo realistic.

include (MH_HairTypes) types of hair style,
include {blue|green|brown|red} coloured eyes,
include (MH_BodyTypes) types of body,
include (MH_HeightTypes) types of height,

ensure they are (MH_CharactorAge),
ensure the facial expression is (MH_ExpressionTypes),
ensure the posture is standing up with hands by side.

use only photorealistic photo-realism. Use natural light and soft shadows to create realistic lighting. Ensure smooth colour transitions. The background is dark grey with no seams. Use a wide camera field of view (FOV) that shows the entire body. Ensure that they do not overlap and are clearly separated.```

i use vars in my prompts. but you can see.

[tawny lance]

I use chatgpt and have 20 wildcard documents for it to refer to

lol

it makes the lewdest and nastiest shit

[edgy ridge]

stuff like that i sent

[tawny lance]

I love it

[charred grotto]

i did a wildcard node {1|2|3} but also, (varname)

[edgy ridge]

create one general idea and then switch fits and stuff

that was what i was looking for

[charred grotto]

vars creature beast monster person human

[edgy ridge]

but when trying to prompt couldnt get even close to that

[charred grotto]

but not limited to..

[tawny lance]

niceeeee

[charred grotto]

its in that wf you have..

take a look

top right 🙂

[tawny lance]

oh I just switched profiles and don't have SD up

just my 134,500 images

[charred grotto]

lol

but yeah, vars via text replace works well before the final prompt..

i work on a dynamic node to replace the x5... so you will just need one..

[edgy ridge]

btw do u guys use vast

and what machine

[charred grotto]

i personally dont

[tawny lance]

I got a 4090 so no need

[charred grotto]

i have a spare 4060ti machine, so i use that

[edgy ridge]

i was doing on my 6800xt but couldnt bother was having some problems

was getting frustrating xd

[charred grotto]

i think prolly ai these days spec image gen is geared for nvidia..

[edgy ridge]

yea when i got the gpu was not thinking i would render anything

[charred grotto]

none of use were 2+ years ago, yet here we are NOT gaming but generating images

who would have though tbh, and i im not taking the pss..

its a true fact..

[tawny lance]

literally same.

lolz

[charred grotto]

lol

it's so crazy

now im like, which card to buy, not for gaming but for faster gens loool

abit of gaming 😉

i figure though, if its fast at gens, it's 100% fast at gaming..

so winner winner chicken... yeah..

[edgy ridge]

yea hahaha

[charred grotto]

frames win games "apparently" BUT power gens faster.. thats what i need

[tawny lance]

I just couldn't swap two cards with 16gb vram

it felt unholy

so I bumped up

[charred grotto]

to what?

[tawny lance]

And honestly I'm glad I did

4090 from 6800xt

[charred grotto]

fk, thats a jump

[tawny lance]

yesh, I wanted the video power

[charred grotto]

4sure

[tawny lance]

I made some especially insane shit with animatediff back in the day

[charred grotto]

oh?

[tawny lance]

Stuff I haven't seen anyone else make to this day

[charred grotto]

nice.. got any links? lets have a look my guy

[tawny lance]

alright let me hunt them down

https://civitai.com/images/6081416

[charred grotto]

awesome!

[tawny lance]

https://civitai.com/images/5740894

https://civitai.com/images/7364806

[charred grotto]

nice, so they took your idea, for those kinda vids?

[tawny lance]

https://civitai.com/images/7802078

Old old old shit

All these are mine lel

I have so many

[charred grotto]

the last tree one is the sht

[tawny lance]

some are great some are donkeybrains

[charred grotto]

first one is good also

[tawny lance]

but now we have kling.

[charred grotto]

we do, we moved on from splicing

now we got full video..

[tawny lance]

lel

[charred grotto]

i love the dark arts.. im like the snape of it..

[charred grotto]

lol

make it an ice kitty, would be puurrrfect

i own 3 cats

i like cats :/

[edgy ridge]

same

got 2

[charred grotto]

have 7 in this house but i only own 3

it doesnt smell..

[tawny lance]

I got 2

So as far as vram goes

[charred grotto]

yep

[tawny lance]

go with something by nvidia that has 16gb vram at least

set your resolution to 1024x1024

steps to 35

euler karras

[charred grotto]

100% 16gb min, but 8 is ok, if video for 5 second clips..

[tawny lance]

video gonna be a journey for you

[charred grotto]

we all cant afford 16gb cards, so a budget 8gb is ok..

it will still work ... it will just take longer and not be as long..

4070ti is your beast..

work towards that or better..

🙂 at a mininum.. but 4060 8gb will do aight..

1 min 1024 image gen, 5-8 min video..

[edgy ridge]

ill get a 5090 or 4090 on vast just so i dont have to run into my 6800xt problems hahaha

[charred grotto]

4060 is like 240£ right now, 300£ for ti..

[tawny lance]

4080 on vast will do the job

[charred grotto]

yeah

[tawny lance]

for images

[charred grotto]

4080 will do any job lol

my system is fked tbh, mine any my partners, we got ddr5 systems 7900X3D with 4060TI

its aight, but we so bottlenecked..

a lil tip, i use co-pilot, so ill tell cp to generate me an image of blah, it does reallt good images, i will either use that image via control net or ill dl the image then reupload it and tell cp to describe and create me a prompt to recreate the same type of image.. try..

try it 😉

[charred grotto]

s few iterations later you can say to cp, enhance this prompt..

[tawny lance]

I make some degen shit but please don't assume the worst

lmao

[charred grotto]

😐

[tawny lance]

yessss

your suspicion feeds me nourishment

[charred grotto]

im assuming, 😐 i dont want to.. 😐

[tawny lance]

haha

:|

[charred grotto]

think good thoughts, think good thoughts.. roses n nuns, roses n nuns..

[tawny lance]

Yesssssss

[charred grotto]

lol

[tawny lance]

Hi Maxfield!

Just waiting for me to slip up

[charred grotto]

is we about to get banned? ban him, ban him, ive been good.. 😐 no wait.. dont ban any of us..

[tawny lance]

https://tenor.com/view/not-yet-mace-windu-star-wars-gif-9797353

[charred grotto]

https://tenor.com/view/yoda-the-force-star-wars-gif-15992603181964778142

jokes aside, has a mod arrived

hello!

[edgy ridge]

ill just accept my long ride and a 4090 is not that expensive since i can use it only on demand and its hourly

[charred grotto]

thats the thing with subs

imagine in one year if you go hard at it, the cost! now just buy a budget pc outright, and think nothing of it..

[edgy ridge]

i mean im not expecting to do like 1000 hours of renders

xd

[charred grotto]

and wait 30-60 seconds longer per gen..

well, why have it, if your not going to use it..

you can knock out some really good stuff..

ala, yeti and gorilla.. and shiela..

if i had 24gb mem gfx card, i'd defo be putting my char to use.. right now i cba waiting for 5s clips..

redo redo redo..

girl is ready..

these borg got messed up..

it's low rez, but its so good..

gotchoo..

with that good night, 🙂

[edgy ravine]

https://tenor.com/view/jeremiah-johnson-robert-redford-nod-of-approval-yes-you-got-it-gif-3521589

[manic vale]

realvis theshidder vhsrealisticxl a frontal photograph of muscular dr Manhattan at the front sitting on the toilet, defecating in distress, in pain, highly detailed, epic, 8k, masterpiece, best quality negative score_6, score_5, score_4, text, censored, deformed, bad hand

[formal tiger]

Thanks a lot! I will try this today🥳

[sweet beacon]

one of the most memed location in wireframe

[acoustic latch]

Ultra‑realistic vertical selfie, 9:16 aspect ratio, taken from a high POV with extreme fisheye (8 mm f/1.8 style barrel distortion). In a small bright living room with white walls and soft natural cinematic light. Centered on a smiling young woman, surrounded by Shinchan, Doraemon, Nobita, and Naruto—each making silly, exaggerated faces. Stylish realism: anime characters appear photorealistic and naturally integrated. Sharp focus on all faces, slight vignetting at edges, cinematic color grading, 8k, ultra‑detailed. --ar 9:16 --v 5 --quality 2 --lens fisheye

[balmy reef]

the way the light and shadows play together in ur art bro is just awesomeee

[jovial solar]

https://tenor.com/view/ltg-gif-40123618314459872

https://tenor.com/view/low-tier-god-banned-mods-smite-gif-4866200577432974147

[edgy ravine]

JEZUS guys... get some mods in here

[tawny lance]

I guess... who do we call?

I have no idea

But I know how to get someone in here...

BOOBS

BOOBS

[edgy ravine]

did you see popcorn gurl?

[tawny lance]

I LOV POPCORN GRILL