#🍥|anime
1 messages · Page 39 of 1
friday here


12AM for me is 17:00 for you!
well, because my timezone is my timezone 
is that really working?
what do you mean? 
america is in the past
you status is this depended on the watcher?
yup yup 
imagine you travelling 5 years trough space with high speed and for you 5 years passed but on earth has 6 years passed
MIND BLOWN
it's probably 2338 for you now right?
0:39 tbh

general relativity is the goat
he dont calculate summer time
time travelers are real
its possible to travel into the future
if you fast enough kekw
absolute fire
if you drive with train and throw down a ball, the moment you catch him up is like a minimum in the future, its almost not measureable but thats it
well, actually whatever you are looking at, it's actually in the past
because light takes time to travel
they made 1 test, they let a dude fly but before they matched his watch with the perfect time of the atomic clock. After he landed his watch had different time then the atomic clock. To be sure they matched it again and testet it the same time without flying, there was no difference.
true
i mean the difference was really really small but yeah they was able to approve Einsteins theory
"Dark" series vibes

lol
welcome to germany xD
When you accidentally leave your denoise too high 😉
xD
Can we get much higher
i love german series. im watching babylon berlin
thats a good series. The sad thing is, its almost like the reality in europe since 1-2 years bc some clans growed up
I need Kindred
2 min

the preview ayayaya
dont shit on me SD
Boo womp
😳😳😳
except 1 finger (again) its perfect
oiii nice look
can y'all like my comment on a youtube channel?
i know that this is completely random but idk where to ask
Why does this happen? you can see the outline where the controlnet line was, on the character
If you want to take the risk, I'll like it if I like it and dislike it if I dislike it.
ok
https://www.youtube.com/watch?v=ilr398uhdjE&lc=Ugz4fAz201KNJD1J1TZ4AaABAg.9ozSv2eKXEe9ozUDfVLblq
video and comment
Download Viking Rise to WIN $500 CASH: https://vr.igg.com/event/promotion/?code=eng_wizardsandwarriors
Valhalla is calling! Lead your Viking warriors across new lands to plunder, conquer, and unlock a $200 gift pack FREE for a limited time only!
#vikingrise #valhallaiscalling
Wizards and Warriors starts a new animated alternative history docu...
i´m already to tired to search that comment sorry xD
oof
is fine

do you see the wolfe
lmao

i mean, i would have that face as well
like "sheeeeesh"

is this critical content already?=
how does anime work
i would say, best to remove that 2nd pic just to be on the safe side
I think wolf want to eat her.....
how to report?
you mean in general how anime work?
right click > app > report to staff
i selfreported me xD
I call this piece
lets see what mods say
idk man
yes
this is.... interesting?
Like, i guess it's not really how anime work? But just japanese cartoon?
but how it work
imagine the mods punish me for selfreporting to ask if this content is okay or not xD
why is anime a thing
Why not
Well, I would usually just ping one of the mod tbh
🫶🏻
why is anime not a thing? 
Ikr anime is life
If u don’t watch anime then what are you doing
do you guys use tiled diff/tiled vae when you hiresfix?
it's good imo, but probably not the best
its not workign for me sadly ._.
It’s a good anime but I would go as far as saying it’s the best anime
But hey everyone has oppinions
I don't always use tiled diffusion, so nope
It's annoying when it starts hiresfixing and boggles my pc while I'm trying to play games
the series gave more depth to the universe of Cyberpunk


@finite herald are u here?

That may be so but still wouldn’t say it’s best
Ight now ama try get some sleep my body is aching 😭 I hate being I’ll
aww thats nice
Goodnight 
Elythris, a time spirit
sleep well 
OWO that colours
All the time spirits are colored like that. but its pretty
except the amorphous one, but they dont play by anyone's rules
i dont know about them but the pics are very nice
@scarlet kelp 1st one is good jus had to take down bc the 2nd
okay no problem, thats why i´m asking thank you 

They look more like cosmic entities
thanks for the service
yup. they are
naww cute
Anyone have suggestions for getting a good Japanese school background. Hallway or outside in front of the school
Struggling to find one that looks decent
like how to prompt for that?
I had thought here was a lora for it, and apparently i was correct https://civitai.com/models/20289/school-building-scenery-lora
Miku tea

ohh i think i cant post my next pic here
i think Zyra has to much boobys
is this a result of low cfg ?

i can send u private if you want
she is really pretty
send and I'll send you my girlfriend
possibly, or maybe just lora interfering, or too much weight on any prompt
nani
🤝
rendering the last of 3 then i send
aight
removed weights and upped CFG
no lora was used
goood luck 
also, extra fingers growing out of her forearm
I want pokemon pics
OHH ITS WORKING
Is that Zyra!?
idk is this critical already?
😳
yes
oh momma
Model?

i think it's a little too much cleavage shown

Make a teemo pic
Seems ok to me, I wouldn't go any further though.
Disclaimer: I'm not a mod and have no authority here.
still good generation tho
they was the last one for today i going to sleep soon

@finite herald we need you again
ohh u already here
what do you say, is it to much already?
well i guess you got your answer 
yea..unfortunately lol
i guess 😄
sadge
Just be cautious of busty ladies
ik thts jus how she looks
And her outfit but yea D:
yeah its not ez if they look like this xD

Words to live by right here folks.
let me try smth else

true
no shit sherlock xD
im a nautilus main. sorry i win 😄
i´m lillia main
a hook guy

are u sure? 
meet me in the rift rn
Picture of a cat eating a big mac
No bot here sorry
im waitin!
(╯°□°)╯︵ ┻━┻

arent we all degen? a bit
i mean
Well...we are in the #🍥|anime channel
if i havernt find SD i would play league all day XD
i used to play a lot during lockdown but then i too found SD lollll
then we post league animes and talk about league kekw
LAS

Que linda
i know...3rd world country
i'll just make a smurf acc ez
of...
p
.
i was kidding. i havent played lol for a long time. too toxic for me
idk what you're talking about... I'm innocent.
understandable i think i like making fun of league more than playing,, been some yrs for me haha
AH KAYOOOOT
This style is so pretti
cutie
I get more tilted in valorant
if you taking league to serious, you will not life happy xD
oh i havent tried valorant. but i tilted too much
I'm going to have coronary diseases due to the stress that valorant causes me
😆
okay i will take my leave now, to sleepy and too late
thank you all for the day, sharing and pictures
Gnnnn!
Sweet dreams
have a good night night
too much adorables
Thats so cute
or how Lillia would say:
"Humans cast a spell every night. Every time they fall asleep."
i ded
Is bababoey a spell cast?

same tbh. rip us
👀
Emote making squad
I mean theres a lora for that....
This absolutely blursed LORA is brought you to in concept and post by me, while having been beautifully trained and supplied to me by a fellow Proj...
Weeeeeee very nice
Im experimenting horrids lora with different models
Ghostmix with it makes nice art
it seems to work pretty well on Aurora.
I dont know that one
It's a must have, I've been daily driving it for almost a month now. https://civitai.com/models/40199/aurora
ooo its luna's i see yeah ill deff check it out
tempted to play with deforum, have you played with it before?
No AI animation is one of the few things I haven't dabbled in yet.
Fizzledorf has a guide for animation that seems pretty thorough https://rentry.org/AnimAnon
LETS GOOO
I know at one point @olive perch was working with the deforum devs. I'm not sure if he still is or not.
I love this
❤️ here's the full sized one after touchups incase you want it.
Awesomeee
What lora u using for that
adding expressive to the prompt gives some fun results on @real nexus's EmoteMakerLora.
OMG, thank you so much!
Would probably work pretty well with that. I made an accidental one earlier (manually added the tears though) #🍥|anime message
Yes, sorry for the late reply I was at work
I can't believe my Lora can do memes
Lol that one wasn't even intentional. I just added expressive to my prompt.
@stone onyx tempted to load it into wallpaper engine lol

If you could mask the character and up her on the top layer and make the effects behind her
It'd be so good
also GIVE PROMPT 
imma try to make a mega wide one
Did png info from when you snagged it earlier not work?
cute
Im im bed rn while deforum is interpolating i can send you the full prompt and settings afterwards ^^
oh I didn't snag it 
I assume barely anyone uses stealthpng
DW about it, I'm going to bed soon as well anyway
I'll ask tmrw

If youd like I can dm it when I have the chance idm 
For some reason when I use highly detailed in the prompt with the lora above 0.7 strength it tends to give me multiple images in one like that.
here have my husbando~
Ooh idk who that is, sorry! He's from a sci-fi novel I'm writing. His name is Odion
I've been slowly making renders of all the characters just for fun 
I love his jojo-esque muscles omg
😂
I'm about to go to bed, I'll definitely look him up in the morning!!!

re deforum pls idk what any of this means 
Error: local variable 'value_is_number' referenced before assignment. Check your schedules/ init values please. Also make sure you don't have a forward slash in any of your PATHs - use / instead of \.
But "/" is a forward slash? Surely they mean backslash?
i knew i was tripping when i tried understanding that
gonna try clearing ui cache ig
reee
tomorrow me problem
i was playing with this earlier today and i could of swore i could select and manipulate the canvas and for the life of me i cannot seem to do it anymore and idk if im going nuts or mixing up extensions 
latent couple lets you draw the areas for your prompts. is that what you're thinking of?
i think thats latent couple, and i want to say they recently added something like it to regional prompter, but i ahven't played iwth it in a hot minute
maybe thats multidiffusions' regional prompter?
there's like 4 versions of this, lol
That sounds just like latent couple. You draw the zones, let it analyze them and it gives you a box to type each prompt into then it sends it to your main prompt formatted correctly.
nah latent couple is wonky
this is baby
wiht region 1 and 2 on
ezpz
i personally it so much better to use than latent couple
oce finding the fucking thing
plus latent couple is just 2 regions, this lets up 8 for whacky split wallpaper potentional
latent couple lets you do lots of zones. I've done up to 6 even before the change from numbers to specify zones to the canvas.
With the canvas you just have to use a different color for each zone.
Guess im too dumb to use it properly lol
I like this visualization, allot easier for my.smol brain 
Soon™️ ill start participating in cow
ree i keep getting double on the second idk why
evenin'
there we go
Tonight was rough for me. I had a complete meltdown and felt everything was about to overwhelm me. Then I realized I had enough of my own brain telling me I couldn't do things, and after working through stuff, I had a moment of catharsis I'm still feeling now. I wanted to capture how I felt, so I made this. It's as close to the feeling of "freedom" and "peace" as I can convey. The ocean has always brought me peace, and twilight is the most beautiful thing to me. Took like 25 successive generations because I wanted it to be perfect.
beautiful
thank you ❤️
man I'm terrible at this animation stuff
That Pika's model? 
No, it's a combination of things
Oh, fair then 
I'll have to see Pika's model at some point
I use Rev Anime V11 and FlatMerge
and nyan mix sometimes
She looks like a girl
anyone trying deepfloyd for anime?
It seems pretty bad from what I've seen. Doesn't even know who Hakurei Reimu is despite her being popular enough to have the second most images on danbooru, only behind Hatsune Miku.
I should clarify, I mean bad for anime. Probably pretty good for not-anime.
h-haha yeah who wouldn't know w-who she is r-right? 
Doesn't really matter to me anyways. I could barely fit the text encoder alone at 16 bit in my 12GB of vram so it isn't anything I'll be able to run anyways.
oh yeah, her. touhou girl
yep
once it gets publically released it'll likely be pruned and mixed and stuff
just right now its not great
Yeah, once it's fully released I'll keep an eye on it. I'm not holding my breath for it to be practical for people with anything less than a 3090/4090 though.
if SHARK can eventually run it on my 6800XT it may give me a reason to actually use SHARK 😛
Xtra Long
SO BEAUTIFUL OWO
ohh hi lotta, thank you for making nice Models i´m using some of yours, great work!
hope it is part of the outfit 
did anyone of use dreambooth already?I got an error i dont understand :c
google prompt guide for stable diffusion
Like, my method is already good and all.
I just wonder if you guys actually used descriptions at all. Here's my example:
too many brackets lol
I'm guessing you just used weights for these.
Also, do I still use the description at the beginning, or its just fluff?
I dont think "perfect detailed hands" is going to do anything
theres many useless keywords here
go on danbooru or any other booru site and see the tags the images there use
I just prompt using booru tags and hope for the best
will do, thanks for the info.
Should I still use the beginning description, or nope?
"A 10 year old modern fire mage holding a silver staff with a red sphere orb wearing a fantasy fire-themed robe and a pink ribbon bow standing on the halls of lit fire temple, with the fire from her red orb on the staff illuminating it" This?
Thats like the most important part of the prompt/image isnt it
It could still work
yeah that bit could probably work
the stuff after that is mostly redundunt
a simple "masterpiece, best quality" will likely be enough
like "(((aligned eyes)))" theres practically going to be 0 images in the training dataset which will have that as a tag
?
did anyone of you created its own Lora model already?
Seems like automatic1111 is back
seems like he made a dev branch and is merging all the pull requests on that
AUTOMATIC1111版とvladmandic版の速度比較してみました
CPU:AMD Ryzen 5 3600
MEM:32.0 GB
GPU:RTX4070
Batch count:4、size:1で4枚出力
AUTOMATIC1111版:xformersあり
拡張機能は両方に同じものをインストール済
341
someone smarter than me can probably figure out what's going on here
seems to be the jp guys found out about the vlad fork
I tried to have multiple characters in a single image rendered in 3D, then put through SD with img2img, controlnet, and T2I adapter. Initial image 100% in Maya 3d:
Output generation:
Not bad for a first try. The T2I adapter really kept the color consistent. Before, the colors were bleeding and mixing all over the place
But it needs more detail
Do you know of a danbooru tag that would bring out the semi-realistic anime?
Probably realistic
with words
why do you use {{}}? thats for dynamic prompts
an i dont see any other dynamic syntaxes so abit confused lol
probably from a novelai prompt
well, that was my old attempt, and I am trying to improve my prompting
oh then first suggestion dump {}
unless youre using dynamic prompts
() with weight is better
(example:1.2)
this extension can help with formatting if you're too used to many brackets, there's nothing wrong with them but a numerical value is more accurate https://github.com/uwidev/sd_extension-prompt_formatter
2d in negative prompt
how do you guys get like "clip skip" and other category on top of model
like dropdown VAE models and stuff
refresh ui after
hello,Is it possible to put 2 LoRA characters in one image? I tried but it generated a character with the characteristics of the 2
hellooo :3
Drop those brackets? What do I use for descriptions then?
https://github.com/hako-mikan/sd-webui-regional-prompter this extension will help you
you'd need to use that oen
() with weights for position or [] with weights for negative
{} is novelai's way
of weighting
wow
D: is that a spoiler
sorry
I mean I'm fine with it, its just that I saw massive spoiler preview images when I was looking at it on civit
I'm like 700 chapters behind on one piece anyways
bruh one piece is just filler lol i hate it lol
what, just filler ? '-'
dont mob me for disagreeing with my opinion pls lmfao
never
dw I couldn't stand the fillers in the anime either
ims the same with any filler
bleach's filler actually pissed me off cos it was os bad lol
but i love bleach
the big 3 at its prime was goated
pretty
torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 20.00 MiB (GPU 0; 8.00 GiB total capacity; 7.18 GiB already allocated; 0 bytes free; 7.27 GiB reserved in total by PyTorch)

Personal merge, it's this one
https://huggingface.co/sleepotimer/Model_A/blob/main/1A.safetensors
Thankies!
So something like this?
A spacious Nave in a church filled with stained glass that was dimly lit up by multiple torches with the camera facing the exit with multiple columns aligned across the hallway with chairs adjacent to it as the fire staff from the middle of the scene lights up with a warm glow, (absurdres:1.8), (highres:1.8), (realistic:1.5), (Nave), (church_interior:2), (beautiful:1.6), (dim_lighting:1.4), (Masterpiece:2), (best quality:2), (chiaroscuro:1.6), (hallway:1.8), (venice:2), (stained glass:1.8), (fantasy:2), (modern:2), (medieval:1.6), [light rays:0.6], (staff:1), (red orb:1.6), (red theme), (fire:1.2), dark shadow, (rim lighting:1.2), (two tone lighting: 1.2), (smoky atmosphere:1.1), (torch:1.6), (night sky:2), <lora:epiNoiseoffset_v2:1>, <lora:finalFantasyVIIRemake_v1:0.4>
I think those would fry the image no?
a bit too wordy and heavy on the weights i agree
yeah, the descriptions deepfried my images
No need to have too much weights on the prompts
cmoing from midjourney is a huge learning curve
and i defff speak from experience
been with them for 2 yrs, tried SD once and immediately swapped lmao
Case in point:
😢
yeah too much weight
I think there's no need to put too much weights unless you want a certain tag to show up 🤔
1.5 is usually the highest I ever go and like sleepti said, only emphasise what you really want, no need to put weights on everything
It's looking somewhat coherent...
`absurdres, masterpiece, realistic, A spacious Nave in a church filled with stained glass that was dimly lit up by multiple torches with the camera facing the exit with multiple columns aligned across the hallway with chairs adjacent to it as the fire staff from the middle of the scene lights up with a warm glow, , (Nave), (church_interior:1.3), (beautiful:0.8), (dim_lighting:0.9), (best quality), chiaroscuro, (hallway shot), venice location, (stained glass), (fantasy:1.2 (modern:0.8) (medieval)), [light rays:0.6], (staff:1), (red orb:0.6), (red theme), (fire:1.2), dark shadow, (rim lighting:1.2), (two tone lighting: 1.2), ((smoky atmosphere:1.1) (torch:0.9) (night sky)), lora:epiNoiseoffset_v2:0.8, lora:finalFantasyVIIRemake_v1:0.4
@random valley
example on how cleaning up your weights can achieve
didnt use your loras tho,
where can i do my prompts?? and my images?
but overweigting of loras is usual cause of frying an image, and using composable loras for more than 2 loras is nice
with stable diffusion web UI ran locally or on dreambooth site
Ok thanks
yw
absurdres, masterpiece, realistic, A spacious Nave, church, filled with stained glass, mulltiple columns, chairs adjacent, warm lighting, (Nave), (church_interior:1.3), (beautiful:0.8), (dim_lighting:0.9), (best quality), chiaroscuro, (hallway shot), venice location, (stained glass), (fantasy:1.2 (modern:0.8) (medieval)), (light rays:0.6:0.9524), (staff:1), (red orb:0.6), (red theme), (fire:1.2), dark shadow, (rim lighting:1.2), (two tone lighting: 1.2), ((smoky atmosphere:1.1) (torch:0.9) (night sky))
notice how we getting more detail with less wordy but more descriptive words
this looks like shit cause im just doing 512x512 and nothing else higher lol
Did a lot of people change nicks or is it just a lot of new people here?
I've noticed there has been an influx of new people in the last few weeks
Easy way to tell is new people will have a little leaf looking icon by their username
im ngl im still new, i joined march 25 and learning everyday
crazy the amount of knowledge i learned in a month
Been playing around with AI since the leaked NAI days lol
So you use `description here` to make it work?
NICE!
i have been with midjourney since they had version 2 but asap i found out about being able to run this locally free i couldnt justify paying them 40 a month anymore
YOOOO
it worked!
Here's my prompt:
Still need to emphasize the bluish night scene like akin to a scene from some cult order of sorts.
😄
less wordy words, more descriptive words is better
like your whole thing about the torches lighting the room? easy warm lighting shaved like 40 tokens or so alone off tha tlol
reminder: tokens are grouped in 75
so each group of 75 tokens gets equal weighting
this is 179 tokens -> 3 groups
I was currently trying to shave off some redundant tokens
all tokens are a bit diluted at that point
i didnt know that thanks eface
so
lets ay we are
175 tokens
it'll break up in sections and apply those segements one after another?
it cuts hard at 75
k
the original spec was 75,
they figured out they can send multiple groups, but at 75 tokens, it starts a new group of tokens
If you want dim lighting I think it's better off using an offset lora since most anime model generates bright pics
i did see an offset lora in there but heavily weigtted when they asked
loras dont need much weight
i usually keep them below 0.85 or so
isn't the locon/lycoris a lot better?
no idea tbh
O, good Civit update, theres categories now, thats nice 
nice prompt my man ❤️
who me?
anyone, i cant get deforum to work for me 
thanks man
this is the internet -> everything is a man until confirmed otherwise!
or ungendered
expression, did not assume gender
im teasing ya
Ain't that supposed to be neutral at this point?

if male -> good choice. if female -> probably even better choice
if non-human -> best choice
heh
My best shot
not sure what youre aiming for but looks nice nonetheless
why did i have to make my model forget how to use booru tags... oh right
it's me i'm talking about, i hate using those
Hello there
hi nerf 🙂
This is my accidental harvest when doing edgerunner's Lucy, feel this composition is rare
Aaaa cool
I modded it a bit now at 61 tokens
I'm not looking for full fire. I am looking for the same description, with darkness flowing through the temple, with only the red orb lighting it.
aight
that should be doable
eh. oh yeah i forgot, i accidentally deleted all my models folder
no more lora's
@random valley i tried man 😦 guess hentai models aren't great at this kinda stuff? or i am a lousy prompter xD
masterpiece fr
I keep seeing way too much light even at supposedly dark pics that I've prompted
Let me check if there was a way to do just that...
90' style
@random valley now with the lora
nice, same prompt as without LORA?
yeah
what was the prompt anyways?
bandaged young mage girl, torn fire themed robe, holding staff, silver staff, illuminated by red orb, , (standing),dark church interior, torches, (hallway:1), (venice), (modern fantasy:1.1), (two tone lighting: 0.6), (smoky atmosphere:1.2), (floating short hair), (pink hairbow), stained glass window <lora:offset_p-0.5-0.1:0.7>, dark night sky
AI never failed to impress me
Woah cute
AI with ai???
Luminochemical princess, snake-like crown, gold, elf ears, small fibrous horns. neonsynthetic lapis blue eyes, illuminated red dreadlocks, spider-crouching on top of cyberpunk train, cyberpunk art, futuristpunk aesthetic infusion, apocalyptic style, cinematic lighting, dramatic angle, masterpiece, best quality, HDR, High Contrast, Low Saturation, 8k
Try this but toy with weights yourself
Im not home to try it
can also throw in 3d digital illustration, digital art at start or before** cyberpunk art**
@queen void@wispy canopy
@sharp bluff you too
lemme get on pc
https://civitai.com/models/6209/dark-gemini-v3 especially if you have my latest release
DARK GEMINI Version 3.0 (I recommend using Waifu Diffusion 1.4 VAE kl-anime 2.) ( Not far down you'll find information regarding multidiffusion ups...
which after surviving surgery and having a full resection of the cancer, you might even get a version 4 or another type
@cyan burrow how is the treatment faring?
cool, without weights right?
yeah, just copy pasted it
I've got a weird eye for prompts
for some reason they mostly come together
(Luminochemical princess:1.19), golden snake-like crown, elf ears, small fibrous horns. (plasma-synthetic eyes:1.14), illuminated red dreadlocks, spider-crouching on top of cyberpunk train, (purple desert badlands in background:1.05), (3d digital illustration:1.15), (digital art:1.12), (cyberpunk art:1.15), (futuristpunk aesthetic infusion:0.90), [apocalyptic style::0.60], cinematic lighting, dramatic angle, (masterpiece, best quality:1.10), HDR, (High Contrast:1.11), (Low Saturation:1), (8k:0.95)
Currently, there is no bot on the server that generates images. However, there are plenty of other ways such as the official https://beta.dreamstudio.ai/ website or by running Stable Diffusion locally using your own hardware! Check out #1080946152318443610 for more details! You can also stop by #1025467151206854736 for any issues you experience while using DreamStudio or #🤝|tech-support for any problems you encounter while installing it locally!
there we go @wispy canopy
ok trying now
the latest is with weights
@shut wind I know where to prompt, I'm giving other people prompts
I'm not new to this 😛 but thank you
sorry
Luminochemical princess, snake-like crown, gold, elf ears, small fibrous horns. neonsynthetic lapis blue eyes, illuminated red dreadlocks, spider-crouching on top of cyberpunk train, cyberpunk art, futuristpunk aesthetic infusion, apocalyptic style, cinematic lighting i usually remove the masterpiece best quality stuff... it doesn't do a whole lot in my model
nah dont be sorry for trying to be helpful mate @shut wind
cool, what about with the updated one?
it didnt want to make a purple desert?
yyyy
good shit!
hehehe now there's luminochemical
(Young Shadowcypher Woman:1.19), golden snake-like crown, elf ears, small fibrous horns. (plasma-synthetic eyes:1.14), illuminated red dreadlocks, spider-crouching on top of cyberpunk train, (purple desert badlands in background:1.05), (3d digital illustration:1.15), (digital art:1.12), (cyberpunk art:1.15), (futuristpunk aesthetic infusion:0.90), [apocalyptic style::0.60], cinematic lighting, dramatic angle, (masterpiece, best quality:1.10), HDR, (High Contrast:1.11), (Low Saturation:1), (8k:0.95)
cool
try the shadowcypher 😛
I can't wait to be home
hopefully by wednesday
you had surgery?
third time my entire abdominal is cut open completely
mmm apocalyptic gritty

so who's still the reigning prompt king?
how you holding up?
I'm tired, in pain, uncomfortable, depressed for being in a hospital for the Nth time
and bleeding over myself on occasion
thank you
too bad i'm not in sweden, else i'd give you a visit man
techno-lamia grabbing luminescent boobs. art+culture ❤️
appreciate the thought, though visits to this ward are heavily regulated
yeah, if that's an open wound, i can image
from groin to rib, from above navel and 20 cms to my left
that's the incision
then inside they cut away a lot to get the tumor
they did
how did the doctors feel how it went?
now I just hope the lungs aren't where its spread
the surgeons got it all, they're highly skilled and they work solely on sarcomas
especially rarer kinds like mine
lets hope this is gonna be your last visit in a long while
oh I doubt it
this cancer might come back every few years and I have to battle the odds
its unfortunately that kind of reality
anyway gotta go, cba to sit by laptop
thanks for the well wishes all
cya later"
cya @cyan burrow
eyes are meh... need to retrain
Wait you're doing a Hoshino Ai LoRA?
Are character loras of simpler designs more consistent because less things can go wrong? For example, a single color sweater (simple) instead of a business suit (complex)
I think so
Do any other tags other than "looking_at_viewer" exist, cause god damn I cannot get this character staring anywhere else
https://danbooru.donmai.us/tags tags with tag search
o ok
But god damn they do nothing
XD
I feel like I probably need to make a LoRA for them
looking up, looking away from the camera
Using lineart controlnet, img2img, and image with lineart removed for t2I adapter color pallet, images are more consistent now. Initial image generated from 3D:
Line:
Color with line removed:
how did ya make the 3d image? blender?
Output
I used Maya with the Pencil+ 4 plug-in. They both cost a lot but I found it...free...somewhere... Very recently, a Blender version came out so it may be possible to find the free version someday
"I found it...free...somewhere..." 
I copied Luffy and sailed the high sees and it was in a treasure box
Makes sense 
why not use koikatsu/koikatsu sunshine?
Koikatsu is good for making it img2img but Pencil+4 makes lines based on geo, not image. Although now that the preprocessor for anime line-art is out, that can be used now. I haven't tried but it could be good
its really good in my opinion
yes
but there are like 20 hoshino Ai loras already
Make one with the best eyes
Because the geo of koikatsu is all the same, it's just textured on top. I wonder how it would do, putting a koikatsu character in blender and extracting the line-art from it to use on ControlNet, and also using that image for img2img. And the T2I adapters should also help in color consistency. I saw someone run a koikatsu character through img2img so the poly lines and 3d look went away while still being consistent, but took off the backgrounds
Time to crank up the unet
i always wondered what tf T2I adapters are
It's similar to ControlNet, and it's models can be loaded from the Controlnet menu. They're way faster too, and versatile as well. Some are the same like depth model. Some are different like the color control
This video thingy is pretty neat.
thanks for the info, but on the technical site.. what is better?
Fair :o
I don't know, but there are some things that only one can do that the other can't. But if they can both do it then I just use ControlNet
mm this one didn't turn out as well 😛
When I generate the image from the 3d to SD, I didn't use any upscaling. You can tell because the eyes are pretty bad. Practicing and experimenting now. Hopefully it will be good, but it messes up the hands because the controlnet isn't there in the upscale
Need to figure out how to use both at the same time or something, apparently its possible
Wouldn't it be nice if they could just team up and work together on 1 UI. That would put all the brainpower dedicated to a single project instead, rather than doing similar stuff
ok done fartin' around with this 😛
Book

https://youtu.be/iE39q-IKOzA Also is that the same used for these?
Patreon to support this channel, Hi-res images or just for prompts: https://www.patreon.com/user?u=87233464
Instagram: https://www.instagram.com/demonflyingfox
Twitter: https://twitter.com/demonflyingfox
#ai #harrypotter #balenciaga
Man I'm bad at this.
I mean, thats just mostly RNG isn't it? :o
Is it possible to improve the quality of many images in extras instead of going 1 by 1?
you can use the batch process or batch from directory for multiple images
/
ok thanks
With which public model can I get something like this?
I'm using Aurora with my lora merge
what extension do you need for this "[purple dress:blue dress:50] " again?
model?

what extension do you need for this "[purple dress:blue dress:50] " again?
Not sure if it's this
https://github.com/ljleb/prompt-fusion-extension
thx
Prompt editing? That's built into Auto's, if that's what you're using.
how tf does it work? it doesn't seem to work on my side
[from:to:when]
Anyone know why my inpainting doesnt really inpaint at all
trying to add scales and it just doesnt do anything
thx
I've had the same issue forever. Part of me thinks that it doesn't like to inpaint things that are too small when, after changing nearly every setting there is, it still won't inpaint.
So mask more?
Fair, appreciate the effort nonetheless
try @real nexus's settings (get pinged nerd)
worked for me now
u the nerd
what lora?
horrid's
rate my controlnet image(for a Visual novel) please and send suggestions!
os. never really used controlnet that much
i hardly used controlnet myself, still trying to understand the canny and lineart models
settings?
This one has pnginfo and for cnet just anime lineart
thx
what is lora:LoraMerge_dim128:1?
My custom merge
can i have it?
stocked image from previous gen at home
Loving the art everyone 🫶🏻
crown prince Wahalakk of the Kothia Dynasty
im gonna photo myself, anime myself in clip studio paint, then img2img myself with a control net and see if I can tranfer my anime appearance onto a chiselled over 9000 warrior
just dont f*c$ yourself 😛 you never forget it
Btw, anyone here a above average voice actor with at least a prosumer audio recording setup (high quality microphone, pop filter, shielding, sound proofing (at least the immediate area, pretty cheap to do) and knowing how to record high quality? ^^
sorry
you can do Ai voices now, you dont need to be actor, just enter text
you should probably delete that fuck part, mods have gotten like extreme
they will track down and delete and warn you about anything leaning sexual
not the word, the sentence
They should set up an anime 16 or 18+ channel too, not for hentai. But for unfiltered chatting within reason and so you can post rather violent content and stuff without going deep hentai or deep guro/vore
Dont need to go euphoria
agreed
please
could even just be a subchannel to here
so we can jump back and forth should the image or topic come up
When I get home, it's time to...well sleep, eat a tiny bit and slowly so I don't tear the stitches and go sepsis
and @subtle crown using any t2v or v2v that's very very good costs money I don't have
If you accept such work for me it would be a passion/fun project, with a possibility of future endeavors but I wont promise them or have them in a contract. That's legal chaos war there
since I can't pay
Chubba the Hut, Queen of the Tri-Arc Bolon System
BARK AI is a groundbreaking text-to-audio model that allows you to create incredibly realistic speech, music, sound effects, and more, all on your local computer! In this video, I'll guide you through the simple process of setting up BARK AI on your PC in 3 different ways and I'll demonstrate how to use BARK AI to generate stunning audio experie...
when youre broke, this are your options 😛
once something is out open source where I have meticulous control over inflection, tone, pitch, formant, dialect, clarity etc that's what I will use
no when I'm broke I can ask people that love voice acting or want to try it on
I can use t2v from replica studios on the credits I have. Then replace em. I will only use em for special events
once everything is done and if there is a paid version that's cheap enough where I can do all those things. I'll probably use that
then I will mix my own amateur music and run it through an AI to beef it up
afk sleep now
nn
we are scawy
okay I'm back. Hi :)
en tralla
can make happy again with mikoo tho 

dont ever set your repeats this high guys.......
You spent day on this?
no im checking other loras in my library
Guys remember to water your chibis
so we might get text in models soon TM... looking at that DeepFloydIF model... now we need hands kappa
someone smart enough, probibly will extract the weights for text, hands are hard 😛













