#🍥|anime
1 messages · Page 104 of 1
I like your PFP girl 🙂
Ayyy, thanks 
I haven't generated her in a while though, I should generate her again 
dont click me
Oh my
i want that milk
😄
I sorta um reworked it. 😄
not much
changs the model to a more animeish one
I would watch SHAD ow WAnLes
Dio lu wuolo numildio
which model is it
Hey, is it ok to use pixiv in bot channel?
what do you mean by that?
My model with a mix of LoRAs <lora:loraslayernoiseoffset:0.2> <lyco:LoconLoraOffsetNoise_locon0501:1> <lora:drawingsmallerv2:0.8>
https://civitai.com/models/100505?modelVersionId=126070
@weary solar ^
thx
Rip hands
Hello very nice anime pictures.
thanks owo
grabs
does a merge... GRIN
hope you don't mind a merge with the Irismix_yours_cocomix 😄
now lets try just yours same prompt. also yours but different neg
yeah I'm in a good playful mood today
changed up the girl and used a lora for her
lol very cute
a mix of my model on civit, lottamix
Anyone have a good general-use negative prompt?
one of those are negative embeddings
Thank you. A lot shorter nowadays. Well, people probably still use 200-token negative prompts.
c:
couldn't get metadata so I guessed and made my own , sort 😄
tho thats using a two merged and a 3rd to make a new model. more anime drawn style
trying to make face rounder but I think the lora is messing with it
better
ponders a different prompt
so cute! c:
I like this one too
am i the only one who cant replicate older images? like i put my old image in pnginfo and put it in txt2img and still the image is a lot different
cross check your model/ LoRA / seed
also if you used ancenstral sampler that could be hard to replicate
already did, all the same
its not an ancestral
not too sure then
yeah ive tried a lot lol
How old and did you change GPUs in between? Different GPUs can generate different noise so the seeds are not necessarily consistent between them. There have also been some changes to WebUI that would create differences, I think all of those have options under compatibility or something like that to revert to the how things used to be done. If you were using any extensions that influence the generation when you made them, then you could have to revert the extension and possibly WebUI to earlier commits.
about 1 month, didnt change gpu
specifically july 11 lol
also, is there a way to see a model's merge recipe without having it downloaded? i accidentally deleted a model that i merged and forgot the recipe
i don't think so, if you didn't save backups anywhere then you might be out of luck.
isn't the first time it happens. so many models lost with the sand of time lol
damn the only thing i have is its model hash and the name lol
Are you using Adetailer?
If not then it could help with faces
Or maybe those are XL models
then I don't know if Adetailer works
think I prob forgot to turn it on
for no reason this looks cool
Johan liebert
For a lora of a character
I have 118 images
How many steps should i use?
to train the lora
2k steps-ish should be enough with 10 epoch
i put 25 epoch
it's too much?
Woa hold up
for 118 images
A total of 2k steps I mean
Not 2k steps per epoch lol
is not an anime character
117k steps is waaaaaaaaaay too much
so i should put 10 epoch?
there is 118 images
and is for sd xl
Ah, sdxl lora
Not sure if it'll work the same as 1.5, but definitely try a lower step first since you don't need hundred thousand of steps for a character lora
https://www.youtube.com/watch?v=N_zhQSx2Q3c&t=624s&ab_channel=Aitrepreneur
In this video, he has 20 images and he use 4800 steps
In this video, I'll show you how to train LORA SDXL 1.0 using YOUR OWN IMAGES! I spend hundreds of hours testing, experimenting, and hundreds of dollars in cloud computing training to bring you the ultimate LORA training guide for complete beginners and experts alike. SDXL is incredibly easy to train as long as you know what you are doing with t...
so if i have 118 should i use 4800*4?
3 instead of 20?
4800 from that video is the total steps of the training
I still don't get where that 2340 steps from
Rename the folder to 3_lily collins
For 3 repeats instead of 20
Like this?
Yep
okey thanks
Is it advisable to use more than 100 images or will it give me the same result as 20 images?
In the end I raised the repeats a bit and there are 5850 steps in total, I don't think there is a problem, but is this normal?
steps: 0%|▏ | 6/5850 [13:36<220:57:10, 136.11s/it, loss=0.154]
220:57:10
This loks like you're probably out of VRAM and suffering having it swap between VRAM and system RAM. Lowering batch size, caching latents to disk, or lowering training resolution are the simplest options to reduce vram usage.
(:
is not enough with 12gb vram and 32gb ram?
for 1024x1024
anime walmart
she's holding you guys 
💯
How is this pose called
ultra-chilling
hey, can someone do me a big favor? i haven't used sd in some time and everything is broken, can someone do 28 expressions with inpainting of a char? it would really help as i need them for a expression pack for my LLM char.
ok thats not what i had in mind but .. ok
how come you guys don't show the prompt in images?
i dont see metadata in your images
well good for you, I don't know why it doesn't copy meta data
anime manga style , princess Cat-Woman in (princess dragon dress:1.12), metal
Negative prompt: lowres, bad anatomy, (bad hands:1.31), text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, artist name, kkw-extreme-neg
Steps: 20, Sampler: Euler, CFG scale: 7, Seed: 2084150433, Size: 576x512, Model hash: 8bf3cc2354, Model: animesh_PrunedV21, Denoising strength: 0.53, Clip skip: 2, Mirror Mode: 2, Mirror Style: 6, Mirroring Max Step Fraction: 0.25, ADetailer model: face_yolov8s.pt, ADetailer confidence: 0.46, ADetailer dilate/erode: 4, ADetailer mask blur: 4, ADetailer denoising strength: 0.4, ADetailer inpaint only masked: True, ADetailer inpaint padding: 32, ADetailer model 2nd: hand_yolov8n.pt, ADetailer confidence 2nd: 0.3, ADetailer dilate/erode 2nd: 4, ADetailer mask blur 2nd: 4, ADetailer denoising strength 2nd: 0.4, ADetailer inpaint only masked 2nd: True, ADetailer inpaint padding 2nd: 32, ADetailer version: 23.7.9, Hires resize: 960x768, Hires steps: 10, Hires upscaler: R-ESRGAN 4x+ Anime6B, TI hashes: "kkw-Extreme-Neg: 43eba3ed8ad5", Pad conds: True, Color Enhance: 0.46, Postprocess upscale by: 2, Postprocess upscaler: R-ESRGAN AnimeVideo, Version: v1.5.1
just changed the last of seed number to 69
do you get control over image variation by manually changing seed #?
or if that is possible?
seed number as you cans see can have an effect as well as other thing
right, it does
but im thinking of changing a digit and if that gives control over rendering
atlho i have no idea how those number conventions work
we doin' superheroes again?
I think this image uses this one:
https://civitai.com/models/111408/niji-geometricshapes
that's it, thank you
Does anyone know how to use img2img? try to do this anime version for me, I can't
smol sus is at it again, creating some cute poses
imma be real here, it's rng

too unique to call it rng 🙂
then it's probably due to some lora 
may be, but its the projection of your ideas i mean
i gotta stop rendering for a bit til i can figure out how to free some space ><
i could uninstall genshin impact but ...
i might wanna polay it even though i havent touched it in weeks
if 2-3 or more images look very similar to each other pick one and delete the rest
i do gotta go through my outputs and delete all the unused stuff. i have so many grids and runs that i could trash
actually maybe i just have too many models
ok i guess that means i can render
model?
it's based66 using this lora https://civitai.com/models/106989/niji-tshirtdesign
Anyone know of a model or Lora that can get close to this style?

pastelmixstyli might work
my current webui folder size 🤕
i can feel you. i have to regularly trim the older models
yea i remember when 1.5 was hot everyone was releasing 1.5 models of different artists in ckpt format,now they all gone 🗑️

no but the early ones where ppl were making a single 4gb or 7gb model for an specific artist instead of a lora,those are dead and gone
good riddance
I recently cleared out ~500GB of shit from here and this isn't even counting the other 3 WebUIs I have installed or all of the LLM stuff I have
I haven't actively rendered anything for a month or two now, then 2 weeks before then for another 4 months
I need to go through all the images I've curated and saved along the way and get rid of the junk
my standards have changed from the early days 😄
Like, do I really need to keep this?
my new waifu 🥵
do i need sleep? yes, do i sleep? no, ill rather start with SD again randomly at night for no reason
thats like my whole virtual drive lol
what do yall use to find/sort/organize png's or otherwise similar files
infran thumbnails works kinda for me but its time consuming
damn wrong year
i dont organize anything my whole drive is a mess 😵💫
speakin of sizes this is the size of the anime folder 🙇♂️
hey guys, i need your help. can someone tell me which model/lora can make faces as good like this with that exact style?
this one? https://civitai.com/models/24149
You can use this model on SeaArt here . Mistoon_Anime is my blend of SD models that tries to achieve a more "cartoony" anime style with thick borde...
could be, ill try it out, thx. does that style fit to any lora you know?(only the face)
yea its good with loras , revanimated,counterfit,hasaku are also good with loras
thats not what i asked lol
i meant: "does the style of the image fit to any lora you know?(only the face)"
oh sorry idk that style looks kinda normal so any model can make it
ah yes... inpainting... i love it
😬
PLEASE SOMEONE TELL ME HOW TF INPAINTING WORKS
alriiight since you asked
masked only means the masked area will be regenerated at the target resolution, if the height or width is equal to the original generation there will be very little change in quality, but is the masked area is smaller then quality will be enhanced
0.1-0.3 denoise is target for small changes and increase in detail, 0.3-0.6 is target for altering things, 0.7+ is for deleting and replacing the masked area
certain things inpaint easier than others ~ (what is more common as a close-up in the model memory)
question, what are these: Denoise start, Image CFG Scale, Guidance Rescale? i didnt use SD in months idk if you even remember me xd
im running vladsomething
idk what it was called again
oh yea thats another webui srry not familiar with that one
its silimar tho, its a fork that was made when auto1111 was inactive for a month
💀 i cant do this anymore
reduce the noise
i finally did it.. somehow.. a bit
why tf is unipc so good with inpainting?
Kinda looks like her dress is peeing lmao
i now know why i stopped using SD, its frustrating as f
same lol
not if you were in my place, nothing works, everything is broken(as always), its soooo warm, i cant breathe because of the weather(idk how to translate it from german, but you know what i mean), despite it being night + raining, i feel like a wet noodle because i am sweating so much +nothing works = not goog on my psyche
+ive been trying to get this working for 3 days
humidity
hmm I think miku is also melting in that last one
idk you just cant breathe normally, its really frustrating, i just want 1 char for my ERP with expressions man
I believe me I know. I live onth e east coast of USA , very humid, slap you in the face humid. so yeah I get it
yea, but all that stuff combined is frustrating af
e-erp?!
f it, can someone do it for me?
yes
thats why I work from home with ac running and fan directing ac to me
there are no ac's in germany, and you are not allowed to install one here(if you can even find one)
Are you using auto1111?
no
yes, thats why i hate germany
no swamp coolers?
i dont even know what that is
I think it's an evaporative based cooler
just look up DIY swamp cooler
not in an apartmennt
I guess if you're in an apartment you don't have much for options
unless they got a window heat pump
what even is that?
it pumps heat in or out of your building
swamp cooler is literally a container for ice, fan on top blowing cold air
no we dont
I heard they were popular in Europe/Germany
it's good because it can heat in winter and cool in summer
FREE HER
can someone do it for me please? i am on the edge of smashing my head against the wall
do what?
inpaint a face of a char 21 or so times with given expressions
ok now she's next to a swamp? cooler? sigh
you want 21 expressions o.O
no
😂
its literally one char, i cant inpaint, i am frustrated
i dont wanna learn it man, i just want my char
how did i miss an expression lora?!
i am a failure
T^T
thats not what i said, i want to inpaint a existing image with specific expressions
well I tried to help witht he inpainting , the heat well. I tried with that too. so.. shrugs
Can you run the prompt overnight? When it's dark and coolest?
That's the only suggestion I can think of
it is 04:10 in am in germany
its even raining
And it's still too hot? Yikes 😬
yes, now get why im so frustrated
some nice expressions in there
make the swamp cooler
she's melting
great now sd ui is not reacting for the 4th time restarting, nah im gonna give up. Now i know why i dint use sd in so long. Gn👋
Stole ya girl lotta hehehe
I think she overheated in the german heat
"I seem to be having a issue"
I hate inpainting hands
huh. very odd effect
nice
melting
ok sorry they are called Evaporative Coolers
found some on the german amazon 😄
had to bust out Photoshop
before and after
maybe she lost too much nose cuteness
maybe a bit better
savin that into the "hands for fixin" folder
Hey guys, I have been on the server for a long time, but I wasn't able to use SD as my PC didn't have a dedicated graphics card. Now that I have a new PC (3070Ti), can anyone help me with the setup? I have already installed SD, but I have no clue about models and everything. Are there any YouTube videos I can refer to? Your help will be appreciated. Thanks.
https://civitai.com/ here you can find a ton of models, you can follow automatics features guide to know where to place it but...it's "models/stable-diffusion" 😛 oh and uh here's my model https://civitai.com/models/6209/dark-gemini-v3
@crimson osprey
https://www.sdcompendium.com/doku.php?id=start here you can find a lot of guides on the menu to the left
that should be all you need to get started mate
All I need is an artist that can draw well and is willing to put in the time as a passion project without pay until I begin to make money on it. AI imaging etc won't do it 😮
but what exactly do you mean by web and ai development anyway? that's pretty unclear @zenith heath
what?
THANK YOU!
my model is a natural language model on the prompt, it understands sentences to a degree. Not if you make em lengthy though.
it's also very versatile
if you want to try something, you can check the preview images on there too
and a few other examples I recently made with it
@crimson osprey
a few more gritty ones
also working on a new gemini model, or zodiac model..maybe I'll name it after another Zodiac. Think I'll skip cancer though, had enough of cancer
here's a few sneak previews on it for everyone that liked my dark gemini 😉
@pale monolith
and some city pics! Dystopic
and some ghost angels
haunting a forest
Add some victorian ladies to your cyberpunk city
like these? I mean they're all vampires and one of them is Lilith
but still
heya
hey
if you check the images I posted around your tag
whatdo you think? New model I'm working on
I meant just out of place Victorian ladies walking down the street lol
finally getting to it
Not that out of place, they just need some combo fashion and a few augments
cool! are those all straight prompts no lora/embed results?
yes
hmmm makes me curious, im too much of a lora addict to use anything without them heh
I mostly bquit doing Lora/embeds. Almost everything I have lately is either straight SDXL or SDXL+img2img with another model
from sneakpeak down to the ghost angels
haunted forest ones
This one is intended to be more specialized instead of diverse
you do have a more euro facedefault i think which is cool, don't see that usually
euro facedefault?
I prefer 2.5d and slightly more realistic proportions and faces than anime. But still maintaining that vibe
the facial features are not so asian, usually it feels like if you don't ask for a specific face/nationality models default to asian and have difficulty without some embeds etc
but the model can do it
(Beautiful young woman with vibrant medium long pink hair and intricately detailed cobalt blue eyes, she wears a Mizuiro inspired cyberpunk outfit and has visible augmentations on her hands, face and chest:1.20) solo, [Futurepunk::0.10], [Cyberpunk::0.40], [Neonpunk::0.45], [Dieselpunk::0.55], [Neo-Noir::0.65], (highly detailed beautiful face with smooth contours:1.18), stands in a neon lit cyberpunk city at night, ({Cyberpunk:Gearpunk:0.80}art style:1.20), (digital art:1.20), (2.5d:1.15), (splash art:1.10), cinematic lighting, cinematic angle
Negative prompt: low quality, no quality, bad quality, low resolution, lowres, normal resolution, no detail, low detail, deformation, text, title, Description, logo, watermark, image a rtifacts, cropped, cropped face, cropped upper body, cropped lower body, spaghetti body, siamese twins, conjoined twins, conjoined body parts, big breast, massive breast, major breast, big tits, massive tits, major tits, **censored**, **censored**, **censored**, **censored**, **censored**, **censored**, **censored**, **censored**, **censored**, **censored**, slutty, nude, nudity, naked, bad_hands, neg_hands, pastel, water color
Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 2871854571, Size: 768x768, Model: OtherGemini-v0.2, Clip skip: 2
what model?
hmmm interesting i have neon , cyber, and diesel punk loras 
these are just tags
omg my internet so slow tonight
Mixing This model is a continuation of my old Buttonmix, mixed using the scientific process of "Inebriated Mixing". I don't remember what exactly w...
there we go
yup i got it downloaded
enjoy spce man
awesome
an assassin appears behind you to slay you!
robutt
mmm if i simplify the prompt a bit
yeah DG3 deals in natural language really well, most other models are closer to booru tags
hot
I used to be Harrison Ford in Bladerunner, but then I took a fire bomb to my body and now I'm standing on tiny pylons doing nothing
haunted house, exorcist girl go
why on earth would she be homeless if she can do that?
jesse we need to cook
Part of a cures cus why not
oversized impractical JRPG weapons the ULTIMATE WEAPON
(((digital keypad))), 1girl, Misuiro, cyberpunk outfit, ((mechanical hands)), ((mechanical arms)), cyberpunk theme, [dieselpunk:0.55], intricate, neon, (night), rain, wet, pink hair, [[from side]], depth of field, blurry foreground, looking at viewer,
+lighting lora
hand looks crisp
now do it on sdxl
cool, why use ((())) and [[]] ?
Use (wordhere:1.#other number here) insted
abit overweight but crisp af
habit, sometimes i feel its easier to just add by 10%s with brackets than specifying
yes but theyre considered tokens and part of "words"
so it might become real weird
I've got to add a few color based merge blocks
idk i only have 49 tokens anyways so doesn't matter much xD
btw, nude models tend to have the best anatomy if you're block merging
so if you can get in the fingers and arms and legs from there, leave out the rest
You should have fewer third limb syndrome
nice, still your model?
yeah
any lora, lycoris, embed, upscale, multidiffusion, adetail?
so i mixed my model to maintain cute freckles and then did some fine small merging later to maximize lora compatability
this is currently
<lora:[LoConLoRA] pseudo-daylight偽日光 Concept_style:1> lora:ada_style:1 lora:UnlimitedBladeWorks1.6:0.3 lora:GoodHands-beta2:1 lora:Bloodborne-style:0.7
aurora and khfb negative embedds
no upscale/multi/adetail tho
damn lots of loras
imma add a few more :3
I'm gonna reupload dark gemini 2.1, seems a few people miss the strong gothic nature of it
that's one too many zeros
xD

ah shit a 202 token render
got it down to112
also using BREAK between language prompt and design commands
hmmm math isnt mathing
This happened @pale monolith - but these are from dark gemini 2.1 that I'm going to re-upload
well shit Im finding other models I've been working on
curious
gonna see what it does on exact same prompt
mmm yeah it's a ton better
primary versions. Have you updated any of downloaded from elsewhere?
"I aint touching that bridge"
hmmmmmm probably like 100 or so
Also idk if he's including SDXL loras, Horde can't use them yet. We only allocate 10-15gb for loras for each worker by default, so dunno what they're gonna do with SDXL loras
suppose i also keep older versions to compare too
too
these aren't dark gemini of any kind btw
they're of another private model I call HyperExalt
err sorry that should be Hypervariant
@pale monolith should I complete Hypervariant and release it?
bah I said solo...I guess it raun 2xg512 and decided to have fun
there we go, a bit atmospheric like taken with a grainy camera
Is there a way to merge models with loras?
Use SuperMerger extension
there's also this script if you'd rather do it in cli https://github.com/kohya-ss/sd-scripts/blob/main/networks/merge_lora.py
or comfy can do it with the save checkpoint node.
how is this pose called
Is it possible to go from an image of a character with a white background to a sketch?
Meow~
that's not his flesh being gone btw, he just spilled cranberry juice over himself and since he's so ripped, it shows
two different art styles of same hero
which lora you guys recommend for firearms?
my webui is consuming wayy to much memory wtf
its not normal for it to be using 9500 mb of memory right?
Considering that I gen just fine in kilobytes... yeah that seems too high lol
what is kilobytes lol
1mb = 1000kb
oh damn cant belive i forgot that
😂
do you have any idea of what could have happened?
Man I am loving this girl she just seems so badass
Not a clue... sorry!!!
np ty
vram or ram
for RAM, mine is like 1~2.5GB, depends on what task its doing
VRAM definitely always near the max
9.5 sounds like a lot, but did you change models recently?
I think it will sometimes hold the models in RAM for a while
oh damn how do i change that[
idk, i just let OS manage it
it's a bonus for me because then I can swap models quickly
Here:
I think they should both default to 0 though. I set my checkpoints to 2 so that models I'm merging stay in vram while I'm testing. Works very well with SuperMerger's runtime merging.
it was in 2 for me, i used supermerger one day and i stayed like that, ty
It's worth noting that if that RAM isn't needed for something else then it isn't hurting performance by letting WebUI use it to store models.
best discord for working on gifs in anime style?
Like what ya mean by that?
i want to make gifs that do animations
i'm interested in communities that are paying attention to what techniques can be used for this
(like improving coherence between shots, for example)
Like got an example?
I just need to know exactly what ya got in mind so i know if i can help or not
#1 This time I added TemporalNet support to the ControlNet Img2Img sequence script #stablediffusion #AIイラスト #ControlNet #temporalNet
things like this
The only one I really know of is deforum. Discord link in the readme. https://github.com/deforum-art/sd-webui-deforum
There's also AnimateDiff, but thats a very different process.
Made this based off another gif then with SD
If this is what you are looking for kinda thing i kinda know how to
is there a way to set custom directory for sd models
woah i havent used 111 in a few days is t his new or did i do this
oh wait i remember installing it now vaguely
lol
i like that
is that look due to a model
cool
that user has so many good loras
i have their flat illustration one
this one looks really cool
pic unrelated
inspo
cofee
I wanna make stuff like that eventually
Not sure if this is over the line, but I love how it came out so I'll shoot my shot.
borgar
kk its ez just use control net on a base gif then that should be enough or make the poses yourself but that takes more time

deku wit da counterfeit v3 model (he bout to go to prom fr)
that's pretty cool, yeah this is the type of thing i'm interested in
I'd be interested in the workflow to get consistent animation like this.
yay Tifa 😄
So, ya'll moved to SDXL workflows/models with Comfy now? Or are we still rockin' Auto & 1.5 models?
idk how to use a workflow i just drop any gif into something online that brakes the gif into single images and i send it into a file and feed it to stable diff and done ez as pie
this ai thing going crazy fr ong
where did you get it and how did you get it in english?
did u get it workin
i dont remember ven installing it ive been depressed lately
i tried to get my friend with an amd gpu to get into ai arts but she was having a problem getting it going or not interested or something
i wish it would like scrape the list from wildcard directory or something
cuz some of the suggestions or whatever in the negative embeddings suggests things i dont have installed
would i run into any problems if i moved my entier auto111 folder to another drive
are these files associated with auto111, i cant tell if theyre leftovers from sygilui
also should i move this to general or something
you might run into problems when it's time to update a1111. I would recommend doing a fresh git pull to the location you want it to be and then just move your embeddings, models and outputs folders over from the old install to avoid any complications.
a1111 doesn't use diffusers. you should be safe to delete that to free up space.
that sounds l.ike a good idea
ok thank you
switch
wabout the stuff in these folders is there any conda commands to clean these up
they look important but they might just be install leftovers
Can anyone tell me what check point and lora are like in this picture?
kinda looks like mistoon the one i was using the other night but im almost sure its not
i wanna figure out how to make more like this style i only ever got this one by accident
is that a render
that it is ai yes, it uses a fumo lora
Is there someone who sells those models on the net... I don't know if they are completely original or if they make variations of someone else's model... As these are all D&D concepts, look for that in Civitai and possibly you can find some merge is used to generate them..
D&D means?
Dungeons and Dragons.
oo thanks
Often also shortened to "dnd"
if you're using SDXL, then that's your reason
im not
seem trustworthy? This is her baby sister
I guess It's niji from midjourney
Correct me if I'm wrong
get enough of them and you could just make a lora
kawaii
what kind of things can u use the krita plugin for
im thinking about trying the blender addon to try to apply ai styles to like some 3d renders
Any idea what is that thing called? Mutant, cyborg, robot, and zombie doesnt work 
Flesh, guts,gore?
Organs
Spiderman stuff btw
maybe amalgamation
i do stufff like demonic amalgamation or aberration
that one maybe like infested or infected fleshy
eldritch tag too
Oh yeah I forgot about infested and eldritch
i wonder if maybe something like compound creature would create cool stuff
well i don't know myself so...
not sure if its new but i got mistoon the other day its got a cool look
i also want to find something new or interesting
i tried one of the 'isometric microworlds' type things but could not get it to do anything cool
you want to make more of this?
yeah? did you share any from that one?
Looks like mid journey
wow, how did you animate it? looks great
Huh what
My house literally has one


