💬│general

[scarlet copper]

Bro click on #👑│introductions and introduce yourself in that channel and a bot will guide u

The go to #✨│priority-early-access and fill the form and after 24 hrs you can generate

[turbid frost]

@scarlet copper you can search fro this prompt: Ice Queen with crown. !/Masterpice/!=

in the search thing

i

swear that it was my prompt

[shadow lantern]

Thanks bro

[torn pendant]

Lol I love that

[fast tiger]

Hi!
Please go to #👑│introductions to make an introduction and #✨│priority-early-access will be unlocked, you must fill out a form and in 24 hours or a little more (due to demand), the invitation will be sent to you by email. 🙂

[jagged shard]

what's the best negative prompts?

[silent raven]

Anything you don't want included in your generated image!

[hard torrent]

⚠️ @torn pendant Hey there! Please be informed that Leonardo.Ai is currently only available as an independent web application. /imagine and prompting for image generations do not work in this community server.

If you have yet to get access to the web application, kindly consider making an quick self-intro in #👑│introductions and read the automated reply for further details.

Thank you!
- The Leonardo Team

[scarlet copper]

I finished my internet and now I'm running on 2g ... So i can't do it now ... Forgive me bro

[jagged shard]

ok

[torn pendant]

Ah woops wrong chan

[civic violet]

Hello there! Please go to #👑│introductions to make an introduction and you can read the automated response by YAGPDB.xyz for further details on how to access the LeonardoAI Website, if you haven't done that yet!

[turbid frost]

I am crying rn

[scarlet copper]

Mutated body parts , extra limbs , two heads , blurry , 3d render , cross eyes , disfigured, ugly ,

[turbid frost]

Everyone, you are cool if you like my newest post! on god

[scarlet copper]

My internet will be back in 2 hrs

[jagged shard]

ok

[whole gate]

Looking forward to trying Leonardo.ai -- I'm an educator & would love to update our visuals from the 90s, lol 🙂

[vernal condor]

how do i know if i get the early access?

[scarlet copper]

Click on the #👑│introductions and introduce yourself and a bot will reply you ... Now early access will be available go to #✨│priority-early-access read instructions and fill the form... You will have access in about 24 hrs

[silent raven]

Fill out the form in #✨│priority-early-access then wait roughly 24h

[scarlet copper]

You will get a mail

[vernal condor]

but i did and i didnt get any email

[whole gate]

@scarlet copper already did, thanks 🙂

[civic bone]

how to fix this problem...😂

[whole gate]

@scarlet copper - was just chatting 🙂

[silent raven]

Should get it sometime soon. They are improving infrastructure so it may be a little longer

[scarlet copper]

It takes 24 hrs

[civic violet]

Go to your email and type leonardo.ai in the search bar

[vernal condor]

i sent like 2 days ago

[silent raven]

Check spam folder or promotions

[vivid heath]

still cant figure it out 😫

[scarlet copper]

I had same issue ... Search for Leonard ai .... And there will be a mail saying you r in for next wave

[shadow lantern]

Heya.. it's probably the model you're using.. set ratio to 1:1 and try it again with the same seed.

[civic bone]

I need 16:9. is there any way to fix this?😕

[civic violet]

You already filled the form in #✨│priority-early-access right?

[vernal condor]

am i getting discriminated because i picked the hobbiest role

[civic violet]

If you already did it then you can open ticket for that, because it shouldn't take 2 days to get approved.

[scarlet copper]

No bro

[vernal condor]

and i have no company or money behind me?

[civic violet]

Bro i am a student here ;-;

[vernal condor]

or they just give priority to large companies?

[silent raven]

No? Most of us here are solo

[low forge]

Is it down for everyone?

[scarlet copper]

They don't care about that

[proven bloom]

your name has now transformed purple

[vernal condor]

i mean i want to use their art just for testing the AI not to make money or career out of it

[civic violet]

Magic it is.

[shadow lantern]

Sure.. you create in either a different model, or use (((single --whatevah--))) or create the 1:1 then use the image to image function to add to it as a 16:9 under a different model.. make sense ?

[civic violet]

Did this for fun 😄

[proven bloom]

the wizards of sorcery

[vernal condor]

so you have access?

[silent raven]

Just have to be patient 🙂 There has been a massive influx of people recently so they are working on improving their systems.

[scarlet copper]

Hey bro did u fill the Google form

[proven bloom]

what model you use?

this looks good

[vernal condor]

yes i have

[civic violet]

Leonardo Creative

[civic bone]

I... I'll give it a try 🥲

[civic violet]

Please open ticket in #989166677910519850 regarding your whitelist email.

[proven bloom]

bet

[silent raven]

I mean idk if you need a ticket for that... they said there will be a new wave soon

[shadow lantern]

Sometimes some Prompt Engineering knowledge helps.. you'll get there.. if you cannot do it RIGHT NOW.. then play with some other image ideas and come back to it.. yeah?

[civic violet]

I mean their form got delayed for 2 days its not supposed to happen

[silent raven]

Not rly technical support tho idk

[scarlet copper]

Try again with another Gmail Id

[fast tiger]

Hi, could you click on sign up that would help you get on the platform please.

[civic bone]

okay okay thanks

[civic violet]

Yoss, @vernal condor submitted a whitelist form 2 days ago and still hasn't gotten their approval email.

[shadow lantern]

okay okay.. yah welcome.. be nice to share a screen to show folks.. but hey.. I get to be annoying for now 😉

[nimble anvil]

Ok...another random question, whenever I make any kind of image that includes a woman, they always have a humongous chest area...what prompts/negative prompts can you use to change that? 😅😹

[civic bone]

😂

[civic violet]

big breasts?

try put it in there

not sure if iis forbidden word or not

or just put flat chest in your prompt

[shadow lantern]

play with different models perhaps ?

[vivid heath]

i cant get the background to change

[civic violet]

dk if that'll work

[civic violet]

You can use Inpainting in Canvas to change the background.

[nimble anvil]

I'll try a few different words

[vivid heath]

i tried but i think im rubbish at this because it wont change

[civic violet]

Uhmm....

[scarlet copper]

Did u use canvas editor

[low forge]

Yes it worked out, now I need to pick myowesomeusername it says😅

[silent raven]

Wooo

[vivid heath]

i used the canvas yes, has earaser and mask tool etc

[scarlet copper]

Good luck

[scarlet copper]

Use mask tool and select background

[shadow lantern]

Sometimes a model will force certain things to appear.. like backgrounds or sets of colours etc.. try a different model or make sure you don't have something in your negative section that might cause it

[civic violet]

PioneerOfDreams sounds dope

[vivid heath]

okay thank you

[fast tiger]

Perfect!!

[vivid heath]

and thank you

[dusty forge]

How i access to this ai ?

[scarlet copper]

Click on the #👑│introductions and introduce yourself and a bot will reply you ... Now early access will be available go to #✨│priority-early-access read instructions and fill the form... You will have access in about 24 hrs

[civic violet]

Please read the information in #✨│priority-early-access

[proven bloom]

why is this community so good

I'm amazed everytime I come here

[fast tiger]

Hello!🙂
By the way...
If you have already filled out the form just wait for your invitation. In 24 hours you will receive it in your email.

[civic violet]

Why i can't feel asleep 😭

[silent raven]

Because you're typing??

[civic violet]

🤔

[scarlet copper]

I think ur the Batman 🤣

[civic violet]

I don't think so.

[shadow lantern]

Because you're running the caffeine model still ?

[proven bloom]

life

[civic violet]

I dont even drink coffee 😭

[shadow lantern]

Tea, coke and sugary sh!te will do it too 😛

[civic violet]

Don't drink any of those today

[low forge]

Wow, it really is, can I take it ?😅

[civic violet]

Of course

[low forge]

Thanks😅

[civic violet]

Yw

[shadow lantern]

@fast tiger can you mute @civic violet for 4 hours so he buggers off to bed?

[dusty forge]

Does this AI also work on Descard like Midjourney?

[silent raven]

No, it's a web application!

[civic violet]

😰

[fast tiger]

Do you want me to do that?

[civic violet]

💦

[shadow lantern]

It's a friendly intervention.. don't want him dying on us

[rose star]

Sleep is important y'all 😄

[midnight cloud]

😂

[shadow lantern]

all them arts and helps.. and he carks it ? 😉

[low forge]

Noo, don't do it, he invented an awesomeusername for me

[proven bloom]

you're just like Ayu now @civic violet

[civic violet]

Like this one first

🤣

[scarlet copper]

Don't worry @civic violet will sleep in the classroom tomm 😂

[civic violet]

Will work in work tmr

i mean

[bronze zodiac]

can someone explain

[civic violet]

Will sleep in work

[bronze zodiac]

whats the concept of steps?

[civic violet]

More Steps = More Details

[bronze zodiac]

i see

thanks

[silent raven]

The more steps theoretically more detail but you can overdo it so be careful

[civic violet]

yw

[fast tiger]

hahahaha

[shadow lantern]

Too many steps == unintelligible mess

[silent raven]

I've found the default of 30 steps to be pretty good

[scarlet copper]

More steps ==more tokens 😆

[civic violet]

i usually use 30-50 depending on the mood 🤣

[shadow lantern]

I tend to use 60 steps.. depending upon the model

[scarlet copper]

Cgf 8 and steps 40 Are ideal

[keen nebula]

I understand that they are working on a mobile app but where to report bugs? I can no longer enter text into the prompt, it always brings up last prompt. Started just 2 hours ago.

[proven bloom]

that's the biggest problem

[silent raven]

You can open a ticket in #989166677910519850 for bugs!

[stone horizon]

what does that mean? The image becomes too far deviated from the original?

[keen nebula]

Thanks!

[civic violet]

Too much details added = unrecognizable image

[shadow lantern]

Yep !!

[stone horizon]

good to know thanks

still waiting on access so I figured I'd get some info here first

[shadow lantern]

my experience has been that it goes into weird colours and shapes..

[stone horizon]

gotcha -- so it soudns like 30-50 is a good range

[shadow lantern]

yup

[stone horizon]

thank you

[scarlet copper]

What happens when you shapeen a image more and more .... The same happens here

[stone horizon]

oh I see

this is an upscaling function?

[bronze zodiac]

Prompt Magic? whats that about?

[civic violet]

Makes your prompts more accurate

[scarlet copper]

Try with and without... You will see the difference

[stone horizon]

how does the output compare to midjourney?

[proven bloom]

see some stuff in #🖼│image-share

[shadow lantern]

the algorithm is a mathematical function that removes 'noise' that isn't part of the image you've prompted it for.. each step taken removes more noise.. remove too much noise and you start to go into "wtf is that?" zone

[stone horizon]

thanks i'll check it out

[silent raven]

You can select from any specially trained models which will run on top of an SD or Leonardo model!

[civic violet]

Never compared it but here's the result of me using Leonardo Creative

[brazen wyvern]

thats a nice one

[scarlet copper]

Midjourney is for beginners

[stone horizon]

these look pretty nice -- can't wait to play with it

[civic violet]

Our fellow Youtuber is here 🤣

[brazen wyvern]

I have arrived

[civic violet]

How's your next video doing?

[scarlet copper]

Did u use magic

[civic violet]

what magic?

[silent raven]

magic prompt

[civic violet]

I forgot

But i think i did

[proven bloom]

midjourney requires little effort to produce a good image

[scarlet copper]

Prompt magic

[scarlet copper]

Yes

You can do better in stable diffusion but requires more skill and hardware

Unfortunately I have 4gb GPU .... But now I have Leonard

[shadow lantern]

Midjourney is the way it is because it has been around far longer, has more models/data sets to pull from and is pay lots to use..

[silent raven]

Also has massive backing

[proven bloom]

definitely more skill, I could type the most elaborate prompts, the best negative prompts and still get bad results

[bronze zodiac]

same over here

lol

[shadow lantern]

Midjourney also has a huge language parser for the prompts

[scarlet copper]

Yup ... U need the use brackets

[proven bloom]

their datasets are massive

[shadow lantern]

Leonardo.ai is growing and when it starts to get into its 3d stuff.. it -should- take its own place in the AI image generation scene

[brazen wyvern]

agreed

[scarlet copper]

Biggest drawback of midjourney is blood can be used as Keyword

[proven bloom]

definitely agree, aiming for the #1 spot

[brazen wyvern]

I should make a youtube video testing midjourney against leonardo

[civic violet]

Did you already thought a good title for your next vid?

[proven bloom]

that would be good

[civic violet]

You could definitely do that

[scarlet copper]

Give your vid link

[civic violet]

And add tutorial on how to use Leonardo because its different from MJ

[bronze zodiac]

would this website ever add more models?

[scarlet copper]

Sure

[civic violet]

There's a lot of models to explore that were made by other users

[runic pecan]

I created an awesome model to get a magical forest environment in Leonardo, To use my model just select my model "Mr. Rushy's Enchanted Forest" And type it In front of your prompt. Hope you all like it Creators. 😉

[brazen wyvern]

this is the first Leonardo video I made yesterday: https://youtu.be/wsE14BCHdJA
I'll have more today / tomorrow as well

[shadow lantern]

it doesn't need to be #1.. it just needs to keep growing.. creating its own niche market place

[humble axle]

Hi. How to use seed? I want to generate a series of images with same style but variations in background and actions.

[scarlet copper]

Oops my internet is over ... I'll watch tomm

[shadow lantern]

I understand that would be down to the model you're using

[civic violet]

Tutorial on how to use Dataset Training
MJ and Leonardo Comparison
Showcase of different models by Leonardo and models created by other users
Canvas Features Showcase

[versed flicker]

My only concern is if the designs we get here are not copyrighted. I mean, if we can use them safely for commercial use, especially in platforms like Amazon Merch or KDP without being banned.

[shadow lantern]

You can create your own models btw

[runic pecan]

Let me know What kind of model you expecting, I'll give my best to give life to it in Leonardo

[bronze zodiac]

id love to see something close to anything v3

[civic violet]

You can use the image you generated for commerical purposes.

[silent raven]

leonardo.ai/terms-of-service

[humble axle]

Please elaborate

I've generated this 1st image and want to use this image to generate more similar but with different background, actions etc. Any idea,how to do that

[civic violet]

#1070595459963433030 look at the anime girl topic maybe that is the one you're looking for

[shadow lantern]

I'll leave that to others to explain.. suffice it to say that A) I haven't gone there yet and B) Leonardo.ai has a model training function for you to play with 🙂

[silent raven]

https://app.leonardo.ai/faq gives a nice overview of creating your own models 🙂

[civic violet]

#📌│faq

[scarlet copper]

What voice is that

[brazen wyvern]

my own

[scarlet copper]

Come on bro .... That sounds la an ai voice

[brazen wyvern]

watch my other videos
thats my voice

[shadow lantern]

2:30AM here.. g'night all.. GL 😄

[brazen wyvern]

I'm youngbuffett on tiktok and youtube you'll see me talking

[civic violet]

Have a good sleep

Its 1:32AM here

[silent raven]

😮

[scarlet copper]

Btw i was able to watch only 20 secs becoz of slow internet

[civic violet]

1.15k messages in 1 day

🤔

[proven bloom]

what's your speed??

[proven bloom]

sounds like you need to step outside

[scarlet copper]

30 kb ....

[civic violet]

in 7 hours.

[scarlet copper]

No bro... I have exhausted my daily limit and it will be back to speed in an hour

[proven bloom]

I feel your pain

[proven bloom]

enjoy some nature

or minimum take a break

[civic violet]

Im enjoying this cat.

[scarlet copper]

Mosquito bites

[proven bloom]

are you using data or your isp has a data cap?

[proven bloom]

this cat is in the ocean, the ocean is outside

[proven bloom]

lmao

[civic violet]

And my screen can show me the outside.

[vestal hinge]

i still haven't figured out how to make a decent sprite sheet with this. it's tearing me up. :/

[proven bloom]

fair play

[vestal hinge]

hopefully everyone's having a good day

[silent raven]

🙂 Thanks

[vestal hinge]

yeah 🙂

[scarlet copper]

Is anybody a gamer here

[proven bloom]

thanks man, hopefully your day is also going well

[vestal hinge]

yeah, it's alright. gonna have a chicken sammich. lol

[vestal hinge]

i've been playing Hogwarts lately. does that count?

[civic violet]

I play roblox for a bit

[scarlet copper]

Yes

[proven bloom]

I used to play paladins

[vestal hinge]

it's a fun game

[scarlet copper]

I still play gta 5 online, battlefield 5

[proven bloom]

battlefield 5 is good

[vestal hinge]

another game i like, it's an older one...is the Legend of Grimrock

cool solo player dungeon crawler

[proven bloom]

I used to play battlefield hardline back in the day

[civic violet]

cries in Celeron CPU

[proven bloom]

upgrade your rig

[scarlet copper]

Yes but battlefield 1 is cinematic

[civic violet]

No money 💀

[civic violet]

2 Cores 😭

[brazen wyvern]

[radiant lantern]

Is it possible to make different pics with the same face?

[civic violet]

You could train your own Dataset with that.

[scarlet copper]

I use my friends laptop... I don't even have one

[radiant lantern]

Only in that way?

[civic violet]

Im pretty sure training your own Dataset will work better for your case.

[scarlet copper]

Make a video on how to get a voice like u 😸

[brazen wyvern]

[sterile oyster]

how long have you been waiting to access?

[silent raven]

Roughly 24 hours once you fill out the form in #✨│priority-early-access

[scarlet copper]

Do text to speech voice videos get monitized

[sterile oyster]

Thank you)

[turbid frost]

darshan whats your leonardo.ia name

[scarlet copper]

Darshan

[bronze zodiac]

hi guys, got a question, i dont get the image to image thingy, like i uploaded an elon's picture with a grey suit, and typed, white suit, and some complete random ppl came up with white suits

[turbid frost]

latest prompt?

[steel glen]

https://www.facebook.com/110615081939094/posts/123379107329358/?mibextid=Nif5oz

[scarlet copper]

A Greek girl , blond hair with flowers , flower dress

[scarlet copper]

Something like that

[bronze zodiac]

any ideas?

[turbid frost]

no the search bar needs the excact prompt

[nimble anvil]

You'd need to use the Canvas mode for that, mask out the suit and then say a white suit etc

[civic violet]

Gonna go sleeping now

[bronze zodiac]

ooooh

[turbid frost]

gn

[bronze zodiac]

okayyy

[silent raven]

gn

[bronze zodiac]

ty sm

[civic violet]

Have a good day, you guys!

[bronze zodiac]

gn ty sm for the help btw

[nimble anvil]

Gn

[turbid frost]

nangin the discord sweat

[scarlet copper]

Rn i can't I'm using my mobile, the prompts are in my friend laptop

[civic violet]

Had to go to sleep cause i dont have food rn 🤣

[turbid frost]

make 1 prompt with just 3 words

and I´ll see your acc

[vagrant coral]

Who use stable diffusion automatic 1111? After i update now i get message cuda of out memory on rtx 2070 Super

What i need to change in scripts and where?

[vernal condor]

man i want to try it but the devs wont let me

[turbid frost]

A

[scarlet copper]

Wait my internet has to replenish... It will take 30 mins

[turbid frost]

takes some time to get the mail

[scarlet copper]

Can't we search user by name or something

[turbid frost]

do you still know one of your older prompts? or did you wrote them in here?

[scarlet copper]

Is it possible to change the user name

[silent raven]

Yes, go to settings

[vagrant coral]

What graphic card is best to buy i mean all is expensive what is best option to wait lower prices?

[turbid frost]

the search bar doesnt see user names

[scarlet copper]

I use lengthy prompt ... I have all in the laptop

[weak oxide]

How to make better image 2 image?

[vagrant coral]

Install stable diffusion and buy rtx4090 😂

[turbid frost]

Mutated body parts , extra limbs , two heads , blurry , 3d render , cross eyes , disfigured, ugly ,

?

[scarlet copper]

Wait 20 mins i will DM u

[vagrant coral]

Copy from somewhere negative prompt which include much as possible things

Or ask open ai to write this for you

[silent raven]

There is a video on how to train your own data set in #🐤│twitter-feed !

[weak oxide]

Can someone please help me with image 2 image?

[vagrant coral]

I can ask open ai for writing prompts but you need to explain him what you need and for what, then he will make prompts like crazy

[sullen cipher]

What do we do here in this channel?

[vagrant coral]

We speaking 😂

About Ai

😂

[tired rapids]

hey guys i wonna replace faces from 2 images what should i do ???? ( important )

[sullen cipher]

Really... 😝

[brazen wyvern]

you could use your own dataset or use photoshop or something

[vagrant coral]

For faces i open Photoshop just 😂

[sullen cipher]

I have the same question... I wanna put a photo and put some armor on it

[vagrant coral]

Go on Fiverr find me there and i will make this for you

For 10$ 😂

[tired rapids]

dude we all designers man 😂

[vagrant coral]

I collecting money for graphic card 😂

[sullen cipher]

🙄 oh yeah... I'm going right now

[brazen wyvern]

I would first train a dataset on the face you want and then make the image from the dataset

[sullen cipher]

By you guys

Bye you guys

[scarlet copper]

I create a gig bro... But when I search i can't find it

[vagrant coral]

How?

You going on preview option under tab in my gigs

[scarlet copper]

Yeah I can see my gig there

[limpid raven]

How does this work

[sullen cipher]

Where I can start creating images in Leonardo AI

[silent raven]

Welcome! Feel free to introduce yourself in #👑│introductions and follow the prompt to get started!

[vagrant coral]

Or you find onself by name from another device like mobile

[limpid nimbus]

Hi

[silent raven]

Welcome! Feel free to introduce yourself in #👑│introductions and follow the prompt to get started!

[tired rapids]

its a nice idea but im not sure it will be the same face or not

[scarlet copper]

But if i switch to buyer and search my gig is not in the first 10 pages ... And i lost the patience to search

[brazen wyvern]

I'll have a youtube video out later today about how to do this haha stay tuned

[vagrant coral]

Hi to all we all are happy in this room and waiting professor engineer to say what we need next 😂

And when we fly to Mars 😂😂😂😂

[tired rapids]

ok im waiting but send me the link plz

[vagrant coral]

You need to type tags in your gig when you creating gig

And also help that you have more images on gig or video which explain what you doing there

Then you will bi higher in search

[turbid frost]

anyone knows the max amount of letters in a prompt?

[vagrant coral]

Where is promt i just registered

[scarlet copper]

I don't think there is a limit

[brazen wyvern]

You also probably don't want to flood it with a bunch of fluff

[turbid frost]

I will go wild with that information

ik but it will be fun to try

[brazen wyvern]

haha true

[scarlet copper]

@turbid frost i got that prompt i will DM u

[turbid frost]

ok

[vagrant coral]

After how much hours you get prompt here ?

[turbid frost]

ok nice

[silent raven]

You need to write an introduction in #👑│introductions and follow the prompts! Then, wait roughly 24 hours

[vagrant coral]

Ok thank you👍

[silent raven]

np 🙂

[slim lion]

what leonardo ai gives me

vs what midjourney gave me

[full thorn]

Midjourney only scored there...but it lacks other areas

[silent raven]

Stable Diffusion also typically requires more description than what you provided to get intricate details

[jagged shard]

how i can get pro version?

[silent raven]

There is currently no pro version. They are currently working on it and it may release next week or sometime thereafter

[full thorn]

I can get a better racoon than Midjourney as well. Not even from here

[eager dawn]

I'm excited to try this out. Anyone know how this ai works? Is it based on stable diffusion? Or completely seperate?

[full thorn]

Kinda both

[silent raven]

Is is based on stable diffusion but there is a collection of trained models on top of it from Leonardo and community models!

[full thorn]

You can train your own models which makes it unique

[eager dawn]

I see. Thanks for the response. I assume that as stable diffusion upgrades so will leonardo?

[silent raven]

Presumably

[sterile sable]

Pessoal existe limite de imagens

??

[silent raven]

200 imagens por dia, pero #🗺│language-select para futuro

[frank bear]

Can I get infinite tokens by paying for the service?

[silent raven]

There is no paid plan right now. They are working on releasing paid plans in the future 🙂

[frank bear]

...I finish the tokens in 4 hours

[silent raven]

they will reset 🙂

[grizzled cape]

I’m new what do I do to start?

[slim lion]

hmm... well, the best I was able to get out of stable diffusion was this. And only because I used img2img with a lot of descriptions

[silent raven]

Welcome! Feel free to introduce yourself in #👑│introductions and follow the prompt to get started!

[frank bear]

I know but I would need it for jobs and stopping every day is not a good thing....

[full thorn]

@slim lion not bad...it's all about correct prompts in place. No matter the application

[wicked yew]

Pleased to be part of this community, look forward to imagining new things 😊

[silent raven]

😄

[fast tiger]

Hello!🙂
By the way...
If you have already filled out the form just wait for your invitation. In 24 hours you will receive it in your email.

[wicked yew]

Not yet, thanks 🙏

[heady root]

does leonardo have image combining, or is it planned for the future? like mixing two or more images together.

[fast tiger]

@mild walrus any suggestions for this?

[mild walrus]

yes, not implemented yet but see here https://twitter.com/Ethan_smith_20/status/1616266391065067525

[heady root]

awesome thanks!

[scarlet copper]

Nice

[brazen wyvern]

quick question, any one run into this issue? "failed" training & dataset?

probably resolution issue on my end

[mild walrus]

yeah, we're getting a lot more traffic than our servers can handle at the moment (which is a good thing!) but when the week starts we're going to try to get it sorted out, at the moment i believe training jobs suffer the most in delays/failures

[brazen wyvern]

ah gotcha, okay no problem
I was trying to record a youtube video tutorial on training & datasets

[mild walrus]

ah heck we'll get it sorted out real soon, our users just unexpectedly doubled in the past 2 days haha

[brazen wyvern]

good problem to have!
no prob I'll make a video about Canvas instead

we should make a channel where youtube content creators to post leonardo videos for everyone to have quick access to

[mild walrus]

@analog nebula 👀 👀 👀 👀

[heady root]

any reading material on the planned 3d generation available yet? I have yet to see any AI that does this well yet, but have high hopes for leonardo!

[brazen wyvern]

@analog nebula under -information-
a, "videos" channel where you guys post our youtube videos
OR you give verified content creators a role on discord to be able to share their videos in that channel

because the more eye balls we get on those leonardo videos the more the youtube algo pushes it
(easy marketing)

[vivid heath]

how do i make the editor square bigger to fill the design i made? im trying to add or replace the background but the square to small and when the square renders it only replaces the square section not entire background of design

[trim scaffold]

#💬│general

[silent raven]

I believe you can change the size of the square by changing the "image dimensions"

[vivid heath]

ohhhh

thank you

[silent raven]

😄

[analog nebula]

#💞│self-promotion : )

うぉおｐｓ

Apologies Japanese keyboard

[wooden parcel]

Hows everyone doing today?

[analog nebula]

I took a break. Overworked 😅

[hollow parcel]

Does "Community Models" also granted the same rights?

[silent raven]

We can't view channel as non "Content Creators"

[analog nebula]

Yes they do

Ah my apologies!

[hollow parcel]

Thanks @analog nebula

[analog nebula]

Permissions fixed

[silent raven]

🙂

[analog nebula]

I had set it in place earlier but had forgotten to set the right permissions

[vivid heath]

hi Ayu!

[analog nebula]

@warped sapphire , feel free to share your content in #💞│self-promotion as well 🙂

There's a timer however to prevent people flooding it

[wooden parcel]

@keen breachu what are you working on? if you wanna share?

[analog nebula]

The community 😉

[analog nebula]

Hey there!

I'm sorry if I missed everyone's questions

I had a bit of an fatigue situation earlier and had to rest unfortunately

[woeful kelp]

hello there

[silent raven]

It's ok 🙂

[woeful kelp]

so hype for this!!

[silent raven]

Welcome! Feel free to introduce yourself in #👑│introductions and follow the prompt to get started!

[full thorn]

I told you to rest Ayu...

[vivid heath]

oh my gosh 😧 your okay now yes?

[analog nebula]

Somewhat!

[analog nebula]

I did

Took a few hours away from the server 🙂

[full thorn]

Ok 👍

[vivid heath]

must rest Ayu 🙏

[shut agate]

Hi gn all

[silent raven]

Gn gn

[wooden parcel]

Best of luck

[vivid heath]

when i lived in miyazaki, miyakonojo prefecture it was very different climate to osaka but i finally got used to it and now i love miyakonojo 😊

[scarlet copper]

@analog nebula hi you r back

Good to see u again

[serene frost]

yo

[silent raven]

Welcome! Feel free to introduce yourself in #👑│introductions and follow the prompt to get started!

[shadow lantern]

So, how to access to the website?

[silent raven]

You can introduce yourself in #👑│introductions and follow the prompts to get signed up!

[scarlet copper]

Did u fill the form

[analog nebula]

Thank you so much everyone for helping! 🙇‍♀️

[silent raven]

🙂

[cerulean depot]

Hi, this is Osama from plusplustutors.com

[wooden parcel]

Np

[shadow lantern]

Hey there. Amazing tool. I'm having soo much fun with leonardo!

Quick question: Can you somewhere play around with ControlNet Model settings? Just saw a youtube video and it seems amazing.

[analog nebula]

ControlNet features coming soon!

We're definitely working to implement them as soon as we can! 👍

[full thorn]

Interesting

[fathom kettle]

Hi glad to be here thanks for the invite

[shadow lantern]

I'm staying awake until it does 🙂

Super excited about it!!🔥

[full thorn]

That's been spoken about too

[wooden parcel]

i filled out the form is it gonna take 24hr?

[silent raven]

Welcome! Feel free to introduce yourself in #👑│introductions and follow the prompt to get started!

[scarlet copper]

Yes

[wooden parcel]

beautiful thanks

[scarlet copper]

But keep checking ur mail

[spark cosmos]

automoderator in self-promotion seems broken

I contacted the team already

[wooden parcel]

Will do, thank you for the reply

[scarlet copper]

Welcome

[turbid frost]

thats amazing! included in pro?

[rapid salmon]

Is there an editor FAQ somewhere that I can read to understand how to use it and what can be accomplished with it?

[spark cosmos]

canvas or the whole generator?

[rapid salmon]

CAnvas

[analog nebula]

Likely to be for everyone.

[vivid heath]

when we run out of tokens do we purchase more or do we get a new lot the next day?

[shadow lantern]

Just did that

[rapid salmon]

I get the generator stuff, I think what makes Leo unique relative to other platforms are the extra tools like canvas etc

and not banning NSFW

[turbid frost]

how great

[silent raven]

They reset every 24h! You can hover your tokens to see when they reset.

[vivid heath]

thank you!

[shadow lantern]

Thanks a lot did both and now where I will receive the invitation?

[spark cosmos]

mask or erase the parts where you want to change the image (erase is more destructive).

Partially overlap the canvas with the image (around 35-50%) to do outpainting.

anything else?

the team also posted this video #💬│general message

[silent raven]

You will receive it in your email! Estimated wait time is 24h but it can be less/more.

[analog nebula]

Just to add, I added an nsfw channel since we may get an influx of people wanting to well.. share more sensitive content

[shadow lantern]

Thanks a lot
Can't wait

[silent raven]

🙂

[analog nebula]

Do note that we reserve the right to ban or timeout people who abuse the channel or post highly inappropriate content.

That is all 🙇‍♀️

[vivid heath]

@analog nebula can we have a childrens channel for cartoon creators etc

[analog nebula]

I'll create one in #1070290451560665158 perhaps

[turbid frost]

best way to make a full nsfw chanel for people above the age of 18

[rapid salmon]

A little confused. For example I see I can make a selection box ontop of an already generated image... then there's a prompt. Does this prompt just modify whatever is in the selection box? For example if it looks good, but I want more photorealitic textures, I'd prompt with that?

[analog nebula]

Already done so

[silent raven]

Correct!

[vivid heath]

thank you 😊 and you must come visit moyakonojo prefecture

[analog nebula]

The selection box is the boundaries yeah

but if you painted an mask

[vivid heath]

miyakonojo

[analog nebula]

[rapid salmon]

That makes sense. I've never seen t his on any other platform.

[spark cosmos]

yes, you need to ask specifically for what you're changing and trying to change. The AI will look at what's in the canvas to determine what to draw (basically how far it "looks" within the picture)

[analog nebula]

Ah do you want access to the Japanese channel?

I can add the role for you 🙂

[vivid heath]

ohhh yes please

[rapid salmon]

i've mostly used midjourney until now. This is a game changer for fine details. I think you guys are on the right track

[analog nebula]

Well I won't say that others do not have it but we are aiming to make ours as good as we can 🙂

[rapid salmon]

CAn it fix things like multiple limbs/digits?

[silent raven]

You can try adding a negative prompt

[rapid salmon]

awesome.

[spark cosmos]

not the easiest of tasks but yes

[scarlet copper]

Use negative prompt

[analog nebula]

Also please do excuse me

I need to check our ticket requests

[turbid frost]

where?

[rapid salmon]

Ayu> Since you know the tech better than most, I have a pre-suggestion question.

Is it possible to introduce more

oops hit enter too fast

[silent raven]

Also, if you want to get access you can introduce yourself in #👑│introductions and follow the prompts

[spark cosmos]

@hoary pendant try again now

[turbid frost]

I see, never seen something like this to get roles

[analog nebula]

Yeah we are an massive server

Managing it can be a bit of a pain 😅

I need that jump to channel shortcut

ah ctrl k works

good

[rapid salmon]

@analog nebula Is it possible to introduce more/optional syntax to prompts to steer the AI the direction we want for characters... for example... Char2 could mean '2 characters' then fill in the details for 1 and then the other... for example. "Char=2, Char1 natural hredhead girl smiling wearing pink and grey dress, Char2 muscular Chinese kickboxer wearing pink and grey shorts with white hand wraps"

That's with the tech, not the current features available now

[turbid frost]

pro mod

[spark cosmos]

the issue with that is that the model would need to be trained on a dataset that specifies this information and I don't think there's anything like that currently available

[rapid salmon]

hmmm OK. It can clearly make multiple characters in one image. I'm sure as technology advances it will be possible. It would be nice to direct specific sections of the image broken down by character, object, or whatever.

[spark cosmos]

I think canvas is a good solution for now however, you can just take a plain canvas and paint your character on that

[spark cosmos]

It can yeah, but the images aren't necessarily tagged "two people" or "three people", usually it's "person" "people" "group" "party"

[rapid salmon]

For the time being graphic designers are not going to be run out of business lol... still need photoshop for composition pieces 🙂

[spark cosmos]

yeah! but if you can make a quick mock up, img2img can very easily transform it into the final piece

[rapid salmon]

I think Leo will make big $$$ if they keep NSFW stuff on the platform since there are so many people trying to create it by dodging the banned words on other platforms. It's nice not to get an error if I type "sexy" or something.

That said, that's not all I do. It's just the thing that stands out here (aside from the tools). I like creating new races for fantasy worlds that are kind of like ours but in an alternate universe...

[hoary pendant]

Ty!

[analog nebula]

@mild walrus would be better able to reply to this one, as he's one of our ML developers 🙂

[analog nebula]

Oh there's still a filter to be honest, but mostly to prevent questionable and possible underage accidents/abuse

[mild walrus]

would require a different training protocol, we're exploring other options :)

[rapid salmon]

I haven't done any programming related to this kind of thing, but I imagine it'd be like directing the AI to create multiple image prompts at once, and then combine them (short of completely new training of course).

[shadow lantern]

hello guys, i want this AI for designing logo/emblem of my company.How can I do this

[silent raven]

Welcome! Feel free to introduce yourself in #👑│introductions and follow the prompt to get started! In terms of the logo, you can try the community model for logos

[rapid salmon]

I was a little disappointed that I couldn't create blood drenched vampires, and zombie morgue stuff

[mild walrus]

the idea is we'd like to have it so you can do that, but have it so its filtered so that by default users will not see that unless enabled. we would like to allow all ages here, also censorship is complicated in that its okay to make the kind of regular nsfw stuff, but it can easily trickle down into absolutely illegal content, so for now we are bit more conservative

[rapid salmon]

^^^^ Horror fanatic

[fossil elbow]

/imagine

[hard torrent]

⚠️ @fossil elbow Hey there! Please be informed that Leonardo.Ai is currently only available as an independent web application. /imagine and prompting for image generations do not work in this community server.

If you have yet to get access to the web application, kindly consider making an quick self-intro in #👑│introductions and read the automated reply for further details.

Thank you!
- The Leonardo Team

[scarlet copper]

If you are new click on #👑│introductions and introduce yourself and a bot will take u to #✨│priority-early-access fill a Google forms there and you can get access to generate in 24 hrs

[rapid salmon]

yeah totally not against filtering... honestly I think a lot of people who want to make NSFW would appreciate the filtering so no one can look up their kinks lol... just the fact that you aren't stopped from making it is welcome though

[hoary pendant]

@spark cosmos Thanks again, good to go now! 👍🏻

[analog nebula]

There's filtered stuff for potential sexual violence etc

[analog nebula]

Apologies for the inconvenience 🙇‍♀️

[rapid salmon]

I guess you guys are trying to avoid gore? I wasn't able to make anything involving the word corpse (zombie stuff), and... I think bloody.

[scarlet copper]

Kindly Check ur DM if u r free i sent some cool generations

[hoary pendant]

no worries at all Ayu, I know you all are working hard over the weekend! 🙂

[fossil elbow]

/imagine

[hard torrent]

⚠️ @fossil elbow Hey there! Please be informed that Leonardo.Ai is currently only available as an independent web application. /imagine and prompting for image generations do not work in this community server.

If you have yet to get access to the web application, kindly consider making an quick self-intro in #👑│introductions and read the automated reply for further details.

Thank you!
- The Leonardo Team

[rapid salmon]

murder might have been blocked too dunno. Is there a list of banned terms?

[fringe arrow]

is there an affialite program for this product?

[rapid salmon]

Just saw this... made me laugh. We need proper emoji reactions in the feed to enhance the sense of community.

BRBAFK

[stuck plume]

Interesting, it says "If you would like to get an invitation to the web application much faster please do take a look at No Access for further details.". The No Access is a link that can't be clicked. 🙂

[silent raven]

Make sure to introduce yourself in #👑│introductions first! and then #✨│priority-early-access will open.

[stuck plume]

Ahh i see. Thanks. My bad. 🙂

[silent raven]

np 😄

[mild walrus]

not fully sure but @analog nebula may know

[worthy frigate]

Hello everyone 🙂 New here I had a question. Is there to get more tokens?

[silent raven]

It resets every 24h! You can hover your tokens to see when they will next reset for you.

[scarlet copper]

Im gonna sleep bye

[silent raven]

bye

Ooo 31k members now!

[lime crag]

any suggestions for funny/unlikely places for a skeleton to show up?

[jagged rampart]

Is this the best channel to ask questions about the model or is there a dedicated channel for that?

[full thorn]

Yes

Ask away

[jagged rampart]

I notice a lot of generated images now include negative prompts. How are those generated and how are they used?

[silent raven]

Negative prompts are things you don't want to appear in your image.

[jagged rampart]

No, I understand that

This is the first time I've seen them (I use SD, MJ and Dall-E)

[full thorn]

The negative prompts help balance your initial prompt more successful. Plus they have an added feature called "magic prompt" that can tune your original prompt with extra spice.

Try to experiment with those

Including the canvas feature

[analog nebula]

Have looked at them 🙂

[jagged rampart]

Are you generating negative prompts in ChatGPT or is there a set somewhere? I notice they are all very similar.

[lime crag]

Negative prompts : low resolution, text, word, wording, watermarks, watermark, error, cropped, worst quality, low quality, jpeg artifacts, ugly, duplicate, morbid, mutilated, out of frame, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, dehydrated, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, fused fingers, too many fingers, long neck, lowres, ((bad anatomy)), ((bad hands)), text, missing finger, extra digits, fewer digits, blurry, ((mutated hands and fingers)), (poorly drawn face), ((mutation)), ((deformed face)), (ugly), ((bad proportions)), ((extra limbs)), extra face, (double head), (extra head), ((extra feet)), monster, logo, cropped, worst quality, low quality, normal quality, jpeg, humpbacked, long body, long neck, ((jpeg artifacts))

[jagged rampart]

Thank you so much!

[analog nebula]

#❓│platform-questions works too 🙂

[full thorn]

No most negative prompts typically have the same idea what to not have included

[jagged rampart]

Ah - I missed that - thank you!

[full thorn]

So people copy and paste

[jagged rampart]

Gotcha

[lime crag]

feel free to tweak it, words at the front seem to have more "weight"

[analog nebula]

Negative prompts definitely are very useful. Got several massive ones I've seen

[rugged tusk]

New here, only just found out about Leonardo AI and looking forward to trying it out...

[analog nebula]

They're pretty general purpose

[silent raven]

Welcome! Feel free to introduce yourself in #👑│introductions and follow the prompt to get started!

[turbid frost]

have fun!

[analog nebula]

You're fast lol

[jagged rampart]

And magic prompt is autogenerated by the Ai?

[full thorn]

I suggest trying out what you think seems best for your render

[silent raven]

Speed demon

[proven bloom]

does Leonardo have a limit to negative prompts? in SD I think the limit is 40 words unless it's not true

[analog nebula]

magic prompt does things in the background yeah

[analog nebula]

Don't think there's a limit at all

I've been using some decently long ones but don't take my word since I don't count

[full thorn]

Magic prompt is the kicker for Midjourney....

[analog nebula]

Might be replaced for something else

[silent raven]

haha

[analog nebula]

Can't confirm yet however

[full thorn]

Lol oh I know

[brisk badge]

what are some of the coolest uses you've seen for leonardo.ai?

[proven bloom]

oh lol, well I could experiment

[full thorn]

@brisk badge everything to be honest

[silent raven]

There's a lot of cool ones in #🖼│image-share

[jagged rampart]

Thanks all..... off to go and play. I may be some time 😆

[full thorn]

@jagged rampart yall come back you hear

[brisk badge]

are you using it for a specific purpose?

[tough leaf]

Made my own customers model today and it came out great👍🏾

[jagged rampart]

Absoluteomondo!

[brisk badge]

Agreed!

[worthy frigate]

How do you do that?

[silent raven]

There's a visual guide here: https://app.leonardo.ai/faq

[worthy frigate]

ty

[full thorn]

@brisk badge I use it here and there. I recommend tweaking a few prompts to figure out which works best.

[prisma gull]

Hi im a creator and would love to try leonardo i really appreciate if inm able to get an early access to this spectacualar software thanks

[silent raven]

Welcome! Feel free to introduce yourself in #👑│introductions and follow the prompt to get started!

[jagged rampart]

You've arrived at the right time!

[brisk badge]

This stuff was science fiction when I was a kid - amazing to see how far we've come

[full thorn]

Indeed

The fun has just begun!!

[brisk badge]

Now to connect this to a 3D printer

[full thorn]

Ooh that's the idea

Sell

[spark cosmos]

this technology currently generates 2d images (or soon applies a texture to a 3d image), it is not 3d printer ready

[brisk badge]

yet!

[full thorn]

It will be coming soon this year

[shadow lantern]

has leonardo a way to remember characters?

I'd like to use a person in differen scenes with different poses, but don't know how to prompt it in a way it keeps the character intact.

[full thorn]

Especially the video aspect

@shadow lantern you can train it to do so

[jagged rampart]

Let the Sigmoid Curve begin!

[spark cosmos]

finetune a model on the character, otherwise there are techniques that require advanced prompting (still a WIP)

[shadow lantern]

Thanks guys! I'll try that!!

[tulip glacier]

i just found about this one. Curious to know when it wil be available?

[slate yacht]

hey guys. I cant seem to upscale images, anyone else having the same issue?

[silent raven]

It may be because the servers are overwhelmed with the influx of people.

[tulip glacier]

where do i actuall make the images?

[slate yacht]

Okay, that makes sense. thanks!

[silent raven]

🙂

[analog nebula]

Server load at the moment!

[analog nebula]

Welcome to our community! If you're new, drop by #👑│introductions to say hi, an automated reply will give you further details 🙂

[tulip glacier]

ok thanks guys

[slate yacht]

what does "leonardo style" do?

[turbid frost]

changes the style of your image

[tulip glacier]

are we able to create unlimited amounts of images with this?

[silent raven]

You have 200 tokens per 24h. you can see when they will reset by hovering them

[spice verge]

Those that have extensively used this and MJ, how does it compare and what differences have you noticed between the 2?

[magic token]

Hi

[silent raven]

Hi 🙂

[magic token]

I get confused, how it works!???

[spark cosmos]

Leonardo has better upscaling, with more types coming, more precise prompting (especially with prompt magic), inpainting, and outpainting.

I feel like MJ has better style \ composition at times, but I still finish MJ works with leonardo's upscaler because it's just that good.

[silent raven]

It's a web application! Have you filled out the form in #✨│priority-early-access ?

[spark cosmos]

Also leonardo is free 🙂

with paid tiers incoming

[deep shoal]

Oh hey, are you the guy who makes awesome YT vids?

props for all the trippyness!

[proud elm]

why do all the ai things have a problem with text you would think that would be easy

[spark cosmos]

because you need to train the text encoder specifically to make text if you want that, but then it's not very good at doing anything else

[proud elm]

9 million fonts to train on

[silent raven]

It is scanning a bunch of photos so it's not exactly reading the text like english

[proud elm]

ah

[deep shoal]

There's a model from Stability coming soon that can do text

[magic token]

I just sent it

[analog nebula]

it does generate what looks like text but not legible text in the sense

[analog nebula]

Will be processed in a moment 🙂

[echo rivet]

Can I just use the upscale for an image I upload?

[analog nebula]

There's a delay so we don't have our server load overwhelmed

[spark cosmos]

text does not survive the VAE at the moment, so diffusion cannot really do proper text

[proud elm]

ya it would be nice if you could have like coffee shop on the ocean and a sign that says coffee and have the thing actually spell coffee

[deep shoal]

This is usually because of something wrong with one of the images in your dataset, do they all open ok locally in image preview or whatever?

[analog nebula]

Yeah it's quite hit and miss

[spark cosmos]

no, but you can img2img at 0.9 Init strength with low steps and cfg and a prompt that properly describes the picture and it will come very close 👍

[deep shoal]

The infra load shouldn't be causing training jobs to fail, just process slowly.

[analog nebula]

It does what looks like japanese text and kanji

but it actually is unreadable lol

[deep shoal]

Yeah, when DeepFloyd drops, assuming it's open source, we'll integrate

That'll give you accurate text

[proud elm]

how long till I can run something like this on a mid tear PC?

and be decently fast

[spark cosmos]

let me double down on the prompt being precise, because the upscaler listens to the prompt you used

[analog nebula]

It's run though an web browser and all generations are done on our hosting services 🙂

[deep shoal]

It's working, it's just backed up. We're going to need to bring in upscale token limits to better manage the load.

[analog nebula]

lol yeah you onboarded 1220+ users in one go 😂

[spark cosmos]

they developed their custom solutions on top of existing works, so the issue is that even if you have a powerful enough machine, you don't have a company doing R&D for you

[analog nebula]

I was a bit shocked

[proud elm]

oh nice deep floyd huh there twitter shows some good text thanks

[deep shoal]

yeah I did XD

[analog nebula]

I was going to have an delay in between batches honestly

[deep shoal]

probably a good idea

[spark cosmos]

oh no, my upscaling addiction is coming to an end!?

[analog nebula]

1220 is an massive surge 😂

[analog nebula]

You haven't upscaled me

[spark cosmos]

you are perfectly fine the way you are 👍

[frosty pond]

Unzoom Ayu

[deep shoal]

early contributors will be taken care of 😉

[spark cosmos]

https://tenor.com/view/blush-anime-shy-cute-girl-gif-16149781

[lime crag]

One thing I have noticed is that there does not seem to be a lot of interaction with the Community feed. The "trending" images have been basically the same ones since I started and only a handful of images have triple digits likes. The Vast majority of the Images have 0 likes. you are seriously lucky to get in the double digits.

Love one another

[proud elm]

trendings all people trying to make hot chicks :/

[spark cosmos]

I can tell you that 3 days ago (before everyone new showed up) was completely different, to me it looks like it's changing a lot and very quickly 🤣

[lethal void]

hello

[deep shoal]

agree, it could probably be better but it seems to change quite a bit to me!

but it's basically 90% hot chicks lol

[silent raven]

haha yeah

[lethal void]

could anyone help me, i would like to get into the beta for

[deep shoal]

almost feels like we should counter that algorithmically or something

[frosty pond]

Hello, be sure to introduce yourself in #👑│introductions

[lime crag]

@spark cosmos COOL. Glad to hear that, im one of the new people that showed up 3 days ago 😉

[silent raven]

Welcome! Feel free to introduce yourself in #👑│introductions and follow the prompt to get started!

[spark cosmos]

maybe just increasing the decay for trending will be enough, boys will be boys otherwise 😌

[proud elm]

just follow the beta invite stuff Torch

[spark cosmos]

although maaaaaaaaaaaybe, a slight negative weighting wouldn't hurt

[lime crag]

ngl, ive made some "trending" images

[spark cosmos]

actually it's a good point, I know how the stable diffusion subreddit ended up and it was kind of toxic with the single-minded focus for scantly dressed characters. If it's not too much work I'd like to see that happen

[lime crag]

Ill 2nd that

[frosty pond]

MJ is too lol

[analog nebula]

Already doing that bit 🙂

[frosty pond]

I mean, I like scantily dressed characters as much as the next person, but variety is the spice of life and all lol

[analog nebula]

I will end the toxcity if I need to

[analog nebula]

Yes.

[spark cosmos]

you are the toxicity

[frosty pond]

shots fired 😮

[analog nebula]

Back in the day all it took were 2 likes

[lime crag]

Thats all it takes now

[rancid pier]

Hey

[silent raven]

now it takes 30+ i'd say

[lime crag]

2 likes on a image would be huge because most of the images have 0

[frosty pond]

Hello, be sure to introduce yourself in #👑│introductions

[analog nebula]

By the way just in case people need to vent their 18+ shares do check out

[analog nebula]

Thank you so much for helping by the way 🙂

[frosty pond]

https://tenor.com/view/obama-pew-deal-with-it-finger-pistol-gif-18056062

[analog nebula]

We notice things ^^

[near rain]

i have my doubts

what's all this token fuss?

[analog nebula]

Just an limitations on how many images one can generate

[silent raven]

You can hover to see when they will reset

[analog nebula]

because of the amount of users we have, we will run out of computational power if everyone had 0 limits

[near rain]

i see, i've been missing out since im back now from my vacation

[hollow parcel]

Run out of Tokens and also Dataset, can we get more Dataset? more than 2, since we can train 10 model within 10 days? but only 2 Dataset?

No mastepiece have been made yat just newbie trying how awesome the tool.

[spark cosmos]

I never had any issues with the amount of datasets, I have 10 atm, what happened?

[hollow parcel]

I only able to create 2

let me re check

the add more button nowhere to see

[willow bridge]

How long take to upscale image?

[hollow parcel]

sooryy

[silent raven]

It may take a little longer due to the influx of people.

[hollow parcel]

it look like my fault, about datasets

[glad slate]

I guess I can't create images today

[willow bridge]

On average? It is now 15+ minutes

nothing scary

[spark cosmos]

community recently more than double in the span of a few days, it's going to be quite slow while the team tries to improve the infrastructure.

If you like the project and would like to see it grow, consider talking about it on social media 😁 it's an important moment for Leonardo and the team needs all the attention they can get!

[spark cosmos]

can you send me a screenshot of your page?

[willow bridge]

I can, but it is 1st generation 😄

nothing to see

[hollow parcel]

Sorry, my fault, it was my mistake, the button is there, I just missing it.

[spark cosmos]

replied to the wrong person

[spark cosmos]

no worries, I was talking to someone else

[willow bridge]

I see

[hollow parcel]

@spark cosmos Thanks, my kstake about more datasets

mistake

[spark cosmos]

no problem!

[forest nova]

I really like how everything is structured around this server!

[frosty pond]

Its the wrong ar

[light minnow]

Hey guys, a couple of questions to see where your heads at!
1- Which use cases of AI are you exploring?

2- Which use cases of AI are you most excited to see?

[turbid frost]

it is

[cloud solar]

Hi, everything I'm making is having the body cloned multiples times and such like this. How can I avoid it happen?

[compact halo]

Got my account today and, I'm quite impressed with such a simple prompt. This project is definitely going to go places and I am glad I am part of that journey 🙂

[quasi snow]

yo yo

[frosty pond]

Check the community feed and try stealing the negative prompts from someone else

[cloud solar]

I'll try

[shadow lantern]

@analog nebula

[cloud solar]

Thx

[shadow lantern]

Why when making a model

why isn't it recommended to generate over 30

Ik it's cause it won't look good but

other models generate over like

50 images

[turbid frost]

you can post you stuff in #🖼│image-share

[compact halo]

Thanks and nice name, hope you win the race 🙂

[turbid frost]

lets see, art3mis might get to the finish line before me

[frosty pond]

Dang, when I made a model on here, I figured no one would use and I'd be like "Yeah, seems accurate" but wow, 480+ pictures made with my model. Probably not a lot in the scheme things but its nice to see people actually using it

[hushed geyser]

I see so many people doing the mistake of adding "malformed hands" to their negative prompts... Guys it doesnt work like this, this will just negate the image in the dataset that contained the string "malformed hands" to start with. IE. no image at all. All you achieve is having the AI hidding "hands" in general behind their back, or under a clothes... It doesn't achieve what you are expecting.

[shadow lantern]

how much images you used

[frosty pond]

Still a win, lol

[turbid frost]

whats your model? xD

[frosty pond]

29 images

[shadow lantern]

I'm using 33

model is creating rn

[silent raven]

which model @frosty pond ?

[turbid frost]

I know, this is one of the reasons why I dont use negative prompts

[haughty salmon]

Where can I find some good prompts for Leonardo?

[shadow lantern]

Is 33 images good?

Ik I went OVER 30 but

I want to get every detail possible

[turbid frost]

@frosty pond , you can make models public now?

[fast tiger]

Hello!🙂
By the way...
If you have already filled out the form just wait for your invitation. In 24 hours you will receive it in your email.

[frosty pond]

https://app.leonardo.ai/models/6b20ff7d-9106-4bab-87e2-5564cb18a980 Its for making paintings mostly, but it didn't come up as nice as I hoped, so I just made it public and moved on to try to make a new one with more variety in the data set

[hushed geyser]

There are very good use for negative prompts such as "monochrome saturated" to control the amount of color.

[shadow lantern]

How long did it take for your model to make

[haughty salmon]

I already have access, just not having good luck with my prompts. I would like to see some examples.

[shadow lantern]

minutes

Hours?

[frosty pond]

Yeah, private by default, but if you go in the settings, there's an option to make public, and a forum to share them

[turbid frost]

I know, but it wents wrong for the most time

[silent raven]

You could always go the to "Community Feed" to see what some other people have done

[frosty pond]

Definitely hours, and its probably worse now as the server is getting busies

[shadow lantern]

Any alternative to making models?

I have like 200 images already

But I can't put em in leonardo

Any other solution?

[fast tiger]

Oh well, you can use this guide will help you a lot:
Prompt Guide: https://openart.ai/promptbook (It's for SD but applies to Leonardo)
https://prompthero.com/stable-diffusion-prompt-guide

[haughty salmon]

I don't see a "community feed"

[frosty pond]

https://app.leonardo.ai/community-feed

[shadow lantern]

Anybody?

[silent raven]

Don't think so

[shadow lantern]

Crap.

[turbid frost]

I am about to make my own model rn, gonna be fire

[cinder pond]

cant wait to give this one a try

[shadow lantern]

I don't think it'll be as good as the others that could generate it with datasets containing 200+ images

The max for this ai is 30

and the advanced settings is behind a..

vomits

[vestal hinge]

really? it was 40, wasn't it?

[shadow lantern]

IT WAS 40?

[silent raven]

i think 30

[vestal hinge]

when i trained my model, i had 40

[shadow lantern]

No the max is 30

[frosty pond]

Technical max is 40, recommended max is 30

[turbid frost]

trust taco, trust

[shadow lantern]

I'm so mad now

[vestal hinge]

10 images can make the difference

[shadow lantern]

Dude check other stuff

You can see

[vestal hinge]

from a nose to a mouth

[shadow lantern]

How good it is

[turbid frost]

I KNOW

[shadow lantern]

EXACTLY

There has to be another way

THERE HAS TOO

[vestal hinge]

i'm just patiently waiting for any development on sprite sheets and the like

[turbid frost]

check my stuff

[shadow lantern]

Don't tell me I downloaded

200 damn images

for NOTHING

[vestal hinge]

lol

[turbid frost]

just drag them into your dataset

[shadow lantern]

It'll be crap

[turbid frost]

its just that ez

[shadow lantern]

No watch

Max is 30

[turbid frost]

if theyr way to different, probably

[shadow lantern]

you could upload more but it'll harm the quality

[turbid frost]

yeah cancel your job

[shadow lantern]

yeah

Imma do that rn

gimme a sec

[turbid frost]

make a new one with just the best

[shadow lantern]

No I aint wasting 200 images

I need to find another way

Wait..

alguem poderia me ajudar?

You can't close models

[silent raven]

#🗺│language-select

[shadow lantern]

so my damn model is still training

on empty images

LMFAO

como eu tenho que escrever pra Ia gerar uma imagem??

I can't take this anymore

[austere rain]

hello!

[shadow lantern]

This gotta be a joke

com " / "?

[fast tiger]

Hi!
Please go to #👑│introductions to make an introduction and #✨│priority-early-access will be unlocked, you must fill out a form and in 24 hours or a little more (due to demand), the invitation will be sent to you by email. 🙂

[silent raven]

Welcome! Feel free to introduce yourself in #👑│introductions and follow the prompt to get started!

[analog nebula]

send me your Leonardo username

Under settings

[supple beacon]

may I suggest a clear button and also a separation between the desired prompt vs the negative prompt with a gap or generate & clear button. Currently, the negative prompt is too close and I have made several mistakes during my cut and paste prompts.

[shadow lantern]

eltaco6969

tks

[analog nebula]

okay

[shadow lantern]

man.

200 images

[analog nebula]

Will let you know if i can get it sorted 👍

[frosty pond]

Yeah I downloaded like 100 paintings for my model and it was like "Error: Max 40" and I was dang

[analog nebula]

Might just reset your count

[shadow lantern]

it sucks

[analog nebula]

Yeah I had that before 😅

[shadow lantern]

200 images gone down the drain lmao

[vestal hinge]

could always split them up into 5 models, lol

[shadow lantern]

Hell no

[analog nebula]

sadly we can't make model training any larger without sacrificing our server capacity

[vestal hinge]

could always just write your own machine learning code

[shadow lantern]

This is why

[silent raven]

it's a matter of server capacity

[shadow lantern]

Yall should let people donate their gpus

[supple beacon]

may i also suggest another feature whereby we only enter one prompt and it would render with multiple ai models for comparison? This way we dont have to go through each ai model manually

[vestal hinge]

train it on your gpu or rent power out

[silent raven]

Well with thousands of people and no funding...

[shadow lantern]

You mean my

intel hd 4400

lmao

good one

[vestal hinge]

https://www.youtube.com/watch?v=nd_uvlhnvE4&ab_channel=OldiesZone

😉

[analog nebula]

They probably would have considered it during conception

[vestal hinge]

otherwise, deal with what you've got access to.

sorry duderino

[shadow lantern]

Conception?

[frosty pond]

I mean I don't think the image limit is too bad. The results still came out pretty good. Plus I heard some people from Stability said with 2.x over specialized small models are better then big chunky all around models anyway

[vestal hinge]

for the image limit, it comes out amazing.

it's a matter of having the proper images though...

[supple beacon]

we can also request donation or membership fee for those that want to train larger models

[shadow lantern]

Also question

if my images

[vestal hinge]

seems reasonable, lambda

[shadow lantern]

have a large transparent logo over it

Can I still use it?

for training

[lime crag]

Did Upscale Break again?

[silent raven]

You can, it might affect some outputs though

[shadow lantern]

damn it..

[native river]

Hi, wy the upscale modify the image ?
Do there is a way to upscale without re interpreting the image ?

[frosty pond]

not right now. Soon

[vestal hinge]

i've found that even base models on Leonardo will produce watermarks...

just depends on what you're producing.

pretty sure it was dreamstime water marks i was seeing

[spark cosmos]

dreamstime and watermark in the negative prompt can help, but yeah happens

it's a residue of SD 2.1 unfortunately

[supple beacon]

have you tried entering negative prompt i.e. no water mark

[vestal hinge]

i have

doesn't always work

[shadow lantern]

Oh screw you dude

literally screw you

Oh get a job lolol

[supple beacon]

have you tried specifying small watermark? This way you can easily remove it in photoshop?

[frosty pond]

https://tenor.com/view/its-always-sunny-in-philadelphia-jobland-job-cannon-get-a-job-charlie-kelly-gif-15634737

[shadow lantern]

DAMN RIGHT

I hate these damn people

Can't stfu sometimes

Think everything so easy