Previous /sdg/ thread : >>102003178>Beginner UI local installEasyDiffusion: https://easydiffusion.github.ioFooocus: https://github.com/lllyasviel/fooocusMetastable: https://metastable.studio>Local installAutomatic1111: https://github.com/automatic1111/stable-diffusion-webuiComfyUI: https://github.com/comfyanonymous/ComfyUIForge: https://github.com/lllyasviel/stable-diffusion-webui-forgeSD.Next: https://github.com/vladmandic/automaticAMD GPU: https://rentry.org/sdg-link#amd-gpuIntel GPU: https://rentry.org/sdg-link#intel-gpu>Use a VAE if your images look washed outhttps://rentry.org/sdvae>Run cloud hosted instancehttps://rentry.org/sdg-link#run-cloud-hosted-instance>Try online without registrationflux-dev: https://huggingface.co/spaces/black-forest-labs/FLUX.1-devsd3: https://huggingface.co/spaces/stabilityai/stable-diffusion-3-mediumtxt2img: https://www.mage.spaceimg2img: https://huggingface.co/spaces/huggingface/diffuse-the-rest>Models, LoRAs & upscalinghttps://civitai.comhttps://huggingface.cohttps://aitracker.arthttps://openmodeldb.info>Black Forest Labs: Fluxhttps://huggingface.co/black-forest-labs/FLUX.1-schnellhttps://comfyanonymous.github.io/ComfyUI_examples/flux>Index of guides and other toolshttps://rentry.org/sdg-linkhttps://rentry.org/rentrysd>View and submit GPU performance datahttps://vladmandic.github.io/sd-extension-system-info/pages/benchmark.htmlhttps://docs.getgrist.com/3mjouqRSdkBY/sdperformance>Share image prompt info4chan removes prompt info from images, share them with the following guide/site...https://rentry.org/hdgcbhttps://catbox.moe>Related boards>>>/h/hdg>>>/e/edg>>>/c/kdg>>>/d/ddg>>>/b/degen>>>/vt/vtai>>>/aco/sdg>>>/trash/sdg
First for **** is the thread schizo
>mfw Resource news08/21/2024>IP-Adapter checkpoint for FLUX.1-dev modelhttps://huggingface.co/XLabs-AI/flux-ip-adapter>[kijai] ComfyUI Flux Trainerhttps://github.com/kijai/ComfyUI-FluxTrainer>Prompt-Guided Image-Adaptive Neural Implicit Lookup Tables for Interpretable Image Enhancementhttps://github.com/satoshi-kosugi/PG-IA-NILUT>MegaFusion: Extend Diffusion Models towards Higher-resolution Image Generation without Further Tuninghttps://haoningwu3639.github.io/MegaFusion/>Aesthetics Toolbox v1.0.0https://github.com/RBartho/Aesthetics-Toolbox>KeyCrafter: A lightweight keyframe editing toolhttps://github.com/null-hax/keycrafter>Luma AI Releases New Version of Dream Machine 1.5https://lumalabs.ai/dream-machine>Flux FP8 (true) Matmul Implementation with FastAPIhttps://github.com/aredden/flux-fp8-api>Microsoft releases Phi-3.5-visionhttps://huggingface.co/microsoft/Phi-3.5-vision-instruct>Magic Fixup: Streamlining Photo Editing by Watching Dynamic Videoshttps://github.com/adobe-research/MagicFixup08/20/2024>sam2-playground: Gradio based playground web UIhttps://github.com/jhj0517/sam2-playground>Easyanimate v4 comfyui release (t2v)https://github.com/aigc-apps/EasyAnimate>Deforum based on flux-dev by XLabs-AIhttps://github.com/XLabs-AI/deforum-x-flux>GGUF conversion of Google's T5 v1.1 XXL encoder modelhttps://huggingface.co/city96/t5-v1_1-xxl-encoder-gguf>hydit-live: Minimalist Hunyuan-DiT Gradio UIhttps://github.com/geoffitect/hydit-live>Anim-Director: A Large Multimodal Model Powered Agent for Controllable Animation Video Genhttps://github.com/HITsz-TMG/Anim-Director>TraDiffusion: Trajectory-Based Training-Free Image Generationhttps://github.com/och-mac/TraDiffusion>Procreate’s anti-AI pledge attracts praise from digital creativeshttps://www.theverge.com/2024/8/19/24223473/procreate-anti-generative-ai-pledge-digital-illustration-creatives>Comfyui_TTP_Toolsethttps://github.com/TTPlanetPig/Comfyui_TTP_Toolset
>mfw Research news08/21/2024>Transfusion: Predict the Next Token and Diffuse Images with One Multi-Modal Modelhttps://arxiv.org/abs/2408.11039>Large Point-to-Gaussian Model for Image-to-3D Generationhttps://arxiv.org/abs/2408.10935>A Grey-box Attack against Latent Diffusion Model-based Image Editing by Posterior Collapsehttps://arxiv.org/abs/2408.10901>Perception-guided Jailbreak against Text-to-Image Modelshttps://arxiv.org/abs/2408.10848>TDS-CLIP: Temporal Difference Side Network for Image-to-Video Transfer Learninghttps://arxiv.org/abs/2408.10688>MUSES: 3D-Controllable Image Generation via Multi-Modal Agent Collaborationhttps://arxiv.org/abs/2408.10605>MUSE: Mamba is Efficient Multi-scale Learner for Text-video Retrievalhttps://arxiv.org/abs/2408.10575>Prompt-Agnostic Adversarial Perturbation for Customized Diffusion Modelshttps://arxiv.org/abs/2408.10571>FAGStyle: Feature Augmentation on Geodesic Surface for Zero-shot Text-guided Diffusion Image Style Transferhttps://arxiv.org/abs/2408.10533>Adaptive Knowledge Distillation for Classification of Hand Images using Explainable Vision Transformershttps://arxiv.org/abs/2408.10503>Kubrick: Multimodal Agent Collaborations for Synthetic Video Generationhttps://arxiv.org/abs/2408.10453>The Brittleness of AI-Generated Image Watermarking Techniques: Examining Their Robustness Against Visual Paraphrasing Attackshttps://arxiv.org/abs/2408.10446>Understanding Generative AI Content with Embedding Modelshttps://arxiv.org/abs/2408.10437>CLIP-DPO: VLMs as a Source of Preference for Fixing Hallucinations in LVLMshttps://arxiv.org/abs/2408.10433>How Small is Big Enough? Open Labeled Datasets and the Development of Deep Learninghttps://arxiv.org/abs/2408.10359>AltCanvas: A Tile-Based Image Editor with Generative AI for Blind or Visually Impaired Peoplehttps://arxiv.org/abs/2408.10240>Comprehensive Survey on Diffusion Models and Their Applicationshttps://arxiv.org/abs/2408.10207
is forge good for flux now can i use that instead of trannyui?
>>102011008all i use is forge now
>FAGStyleThey're taking the piss at this point surely
>>102011036I was pretty amused by that lol
>>102011019wtf
>>102011036lel>>102011084
>>102011008It's pretty good, and as much as I like the slight speed boost Comfy gives me, i do prefer how much more used friendly Forge is.
>>102010968>>IP-Adapter checkpoint for FLUX.1-dev model>https://huggingface.co/XLabs-AI/flux-ip-adapterno one is talking about this? seems like a big deal
>>102011459Nobody reads his news is the thing
Are you ok anon?
>>102010825>>102009692https://discord.com/invite/Y4aH5KubP8Bepcord, join NOW! For your /ai/ needs
I'm tired
>>102011922no ;_;
>>102011922no, they trapped me in here
Morning anons
>>102012010good morning
>>102012010gm
>>102012010>>102012041gm
>>102012103Wholesome
>>102012127little do you know, he's not wearing pants
>>102012117Is she feeding birds to her dog?
>>102011983>>102012004Sorry to hear that anon. Here's a few gens. I hope they help.
>>102011922Yes :)
>>102012172now my day's even worse
>>102012172she look good gibsmedat
>>102012145He literally has suspenders on
>>102012272to throw you off
switched to pixelwave_flux1_dev_Q8_0_01.gguf (link on other thread)seems to work with forge with the lora i made on regular dev_q8_0 >>102012156maybe
Debo killed this general
>>102011961this discord doesn't have devs in it or the old guard. I doubt you guys have good advice
>>102012638It is a pedo grooming general, do NOT join
>>102012679too late, I'm going in
>>102012613>>102012613how did you get those scanlines/missprints? sampler error or pure intentional kino?
>>102012344does anyone else ever feel the desire to just shove FLUX up their asses due to how kino it is? AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAHH
>>102012706i've been trying to get rid of them forever lelit only happens if i have hires upscaler on, so i thought it was the upscaler but i've tried literally dozens and they all do it to some level or otheri've tried low/high denoise toomust be something in my prompt but i have a very long schizoid boomer prompt so hard to tell which term(s) would cause itbut overall, if you use "faded" or "vintage photo" it may occur
Testing
>>102012737you're truing to get rid of them, i'm trying to get them IN. give prompt
>>102012716that pic wasn't made with flux
>>102012764how would you know?
>>102012785I have inside information
>>102012757i'm telling you, it started after i used analog film vintage photoand/orfaded photoi'm also getting a lot more OOM errors with the pixelwave model
>>102012746welcome back :)
>>102012848Die already
>>102012853but if I'm dead, I can't gen. that seems counterproductive
>>102012885>I can't genAbsolute win and net positive for /sdg/
>>102012991this
lolwut
>>102013051>>102013116are you trying to make fever dreams?
>>102013171preciselyor hallucinations work too
>>102013215>>102013116how do you describe those chromatic aberrations that look like they're from alf tazos
At some point I really have to stop
>>102013293not sure how to answer that, but maybe this line> The air shimmers with an almost tangible energy, refracting light into fractured beams that dance across the scene.also the word "radiance" appears multiple times in different placesnew prompt now tho
>>102013371you didn't specify for the weird green things?
>>102013367>A cartoon drawing of the scout from Team Fortress 2 striking a sprinting pose and saying "Can't Stop, Won't Stop!"how does it not know what tf2 is? thats a 100 year old game
>>102013367have you made a character lora and if so how long did it take and pls post settings>>102013382nope
>>102013419I really miss that game, I should give it some time again.>>102013473Dreambooth unfortunately, I keep procrastinating when it comes to testing Flux.
>>102013717>Dreamboothyour patience is larger than mine
https://voca.ro/14X8IEUlbhWzback, and ready for a.i. technology discussion.
>>102013780weclome back. if you haven't deleted your last discord acc then maybe check it
console screenshots droppedhttps://huggingface.co/datasets/bigdata-pw/ConsoleScreenshots>>102013419they just don't have the data
>>102013809i have deleted the one before my current discord, on the current one i see nothing, which is good, because i am not interested in whatever is supposed to be there, ty.
based koff
>>102013981a mistake was made in checking your uploaded image. up to you if you decide to return, no stringent check required.
>>102014048hold you breath for my return>>102014059there really is no reason for posting (non a.i.) pics of handsome men itt
>>102014059not koff but you're an asshole anon
>>102014133>hold you breath for my returnfair enough. the server will always be there.
>>102014161is that his ex kek
>>102014161is this ai or are you bullying someone?
/sdg/ is healing
>>102014185he genned too young and paid the price
>>102014161you're a fucking psycho schizo anon
A bit much 'innit bit uncalled for
>>102014161who the hell is that
All me btw
@102014254nig*bo
we're so back bois
Haven't been here in a while - what is Flux/Forge?
>-
>>102014161lmao is that koff?
wow/sdg/so comfortable
imglets SPERGING RIGHT NOW
.
not selling me on the discord i am sorry to say desu
What a shit show this thread has become
it's worse now frfr
>>102014254Why would debo post that img?
>>102014400you should see the other one
>>102014400coming from the zoomer that goons to his own mother, lmao
Holy nogen seethe, Batman
Wait so, are we back to the "training an entire model to include 1 concept" days with the FLUX IP adapter thing?
back to the primordial ocean i go with my waifu
>>102014506she's cute, but what's with the ugly human?
>>102014541>what's with the ugly humani've spent too much time on /s/'s "girls with unconventional looks you want to bang" threadsalso highish cfg is messing things up
>>102014400tell ldg to go back then
>>102013780great to have you back>>102014541I like his humans. they're very rich in character>>102014703do you prompt for this specific horn style or did it just happen on its own?
>>102014731why is she looking at me like that, what does she want from me?
>>102014744>I like his humans. they're very rich in characterwell you have incredibly shit taste, so im not surprised
>>102014744"ram horns" sometimes it comes out that way
>>102014744>they're very rich in characterappreciate that, i like to try and make the whole image be part of a weird fever dreamstory>>102014765to each their own, anoni'd like to see what you're into
>>102014778>"ram horns"when I've tried stuff like that in the past, like 'ram horns' or 'goat horns', it would just put that animal somewhere in the gen, lol>>102014801I've always liked your gens because they try to be more than just a pinup girl. there's often fun, weird, or surprising stuff happening which is visually engaging. the 1girlers just wouldn't understand
ah i guess with a flux lora a bit higher (+0.25 or +0.5 than normal) cfg could be better
>posts shitty 1boy in you're path
>>102014920>you're
>>102014909>when I've tried stuff like that in the past, like 'ram horns' or 'goat horns', it would just put that animal somewhere in the gen, lolPony realism is a lot better at that. Though adding 'bunny ears' or 'cat ears' can still get you.
>>102014506>>102014079how does she look so out of place, like the lighting is different on her than on the background, like she is standing in front of a painting. it makes it look less like ai and more like it's real. how do you keep making these?
>>102014995employed spammer
>>102015018what does that mean?
>>102014909>the 1girlers just wouldn't understandtruth, even tho i am a 1girl poster>>102014995flash photography>less like ai and more like it's real. how do you keep making these?i always felt that was the goal with ai gen, combining real and unreal and making it uncanny and a little uncomfortable
>>102015027nigbo
je souhaite un grand pouvoir et une longue vie
>>102015179Le cafard dans ma cuisine a mangé mon permis de conduire
>>102015179>>102015226i wish i had some tacos right about now
i miss the banned purple schizo avatarfag
>>102015236Ojalá tuviera algo de fuerza para combatir el crimen de los tacos ahora mismo
이게 왜 이상하게 행동하는 거지?나는 1girl이 아닌 1guy를 요청했습니다
>>102015309you said "crowded gym" didnt you>>102015279tacos, mofo, do you eat them
>>102015368No it's flat out refusing to do a one-guyits korean I thought we where having a language thing
i remember being surprised mexicans have their own type of sandwich, 'tortas'
Ok, I achieved 1guy 'success'going back to 1girlI'll spare you the results, it was very creative
>>102015450>I'll spare you the resultslel you gotta post it now
hear me outcandy tacos
>>102015476It was very sweaty and intenseand lots to the void forever..
>>102015416shit is cash thoof course i get the shits for days after>>102015537ice cream tacos are a thing you know
>>102015537compelling argument
>>102015540do not be afraid of the gifts from the voidreject realityembrace them
>>102015594these are neat, i dont suppose the model you are using could be ran by a gtx 1080, 8 gb...
>>102015621>gtx 1080, 8 gbthere's different versions i believe, check the model pagehttps://civitai.com/models/141592maybe one of them will work for you
>>102015537they're just crepes and you turned up the sidescan't pull the wool over my eyes sunshine
>>102015770>"X is just Y"as is mexican tradition. or texmex at least. a burrito is just a 360 degree taco. fajita is just an unassembled burrito.
and they called me crazy lmao
Midjourney has a free trial again on their website (~50 images)
>>102015594>do not be afraid of the gifts from the void>reject reality>embrace themhttps://suno.com/song/259633d9-0200-48a6-97d6-acae900831b9
>>102016275>>https://suno.com/song/259633d9-0200-48a6-97d6-acae900831b9not enough 1girl
>>102016275Does it hurt?
>>102016318you mean when I fell from heaven?
well it almost worked lel
>>102016274ty>>102016275pretty cool, should spend my 5 rolls againsmt with wolfgirls but what I'm not sure yet
>>102016274That doesn't really look very good. Not criticizing you, but MJ is that's their current model. It handled multiple concepts well, but it's kind of sloppy
>>102016459I think they're under a lot pressure from flux and Ideogram 2.0 released today too
>>102016342No losing your tranny friend
https://suno.com/song/70cb5d1e-7561-41f0-a6e6-79f874dd24e5
>>102016520I used style reference from this pic and a lazy promptseems like their web interface is still crippled compared to their discord
>>102016522will def google that and good, pressure is healthy at this stage I'd say
>>102016595you will NEVER catch schizo anon debo
>>102016102Explain what's going on like I'm a retard living under a rock
>>102016579lel way too much going on there>>102016510coooooooooom>>102016595>BEPCORDwhomst?
>>102006737It is a Heavy Metal lora. And I did think she looked like Major Boobage from that South Park episode. But that was a parody of Heavy Metal, so I guess it all kind of works out.
Since when was this thread receptive to furry art?
https://suno.com/song/f186cb62-cf67-4487-a3c6-b80d16807322
>>102016672https://old.reddit.com/r/StableDiffusion/comments/1exubs0/ideogram_20_launches_beats_flux_pro_and_dalle3_in/>They claim to beat Flux Pro and Dall-E-3 in benchmarks(x) doubt
>>102016763since nogens replaced them as the bottom of the totem pole
Is debo mad at stells now?
>>102016763anything to keep this fag off-topic hugbox alive for them, they have little, if any, self-awareness
>>102016777quite the tall order and this feels pretty expensive
>>102016775awesome. I've been wanting to do some dnb. the drum pattern and syncopation on this came out really good
unlike your rushed shitgen nig*bo
>>102016793stells is unironically shit
>>102016834Wrong lol
after another failed attempt to use uncomfyui, back to the usual
what are stells
>>102016853hormone tablets for troons or cryptoshit i dunno
>>102016851Is this your blog or something?
>>102016865um, yeah?
>>102016814ty :)I did a slew of metal/punk gens and your previous made me think of something more electro-y, it turned out listenably
>>102016865what's wrong with treating this place as a blog and personal hugbox?
>>102016834thank you for interacting, it helps in the algorithm
>>102016853you are in /sdg/ so, of course, "stells" refers to one of the slop genning avatarfags newfriend
>>102016913"a woman visiting Paris"
>>102016969thanks stels
how can I make the models folder from my automatic1111 folder be shared with Fooocus?Used the config file to change to change path, even added double \\
>>102016853It's way to late to get into all that
>>102016969lel
This can't be comfortable
>>102017048a true warrior can endure it
>>102010825>"IN NO EVENT SHALL BLACK FOREST LABS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH USE OF THIS MODEL."https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.mdhttps://civitai.com/models/618692?modelVersionId=752058I think they're mostly doing this because they're covering their asses now that elections are very soon and this model can create nigh perepect mimics of real people. Trump is already embracing using AI to mock Kamala Harris (they'll clearly a shittier model compared to flux). I wonder how the internet will change once normies truly realize Machine Learning's true capabilities.
>>102017072err no one asked
guess i'll have another go at training a lora
up to 12m steam screenshots desugn
>>102017240gn
https://suno.com/song/bd1b50a5-496a-4200-a28d-7c6ffff01f50smt entirely different, vaporwave with a beat
>>102017278suno really can struggle with not putting a beat on everything. idk how accurate this is generally, but I've always thought of outrun as 'vaporwave with a beat'.
>>102017335I really don't know music terms but 'outrun' is the term for this kind of thing?
https://suno.com/song/27585803-2686-4feb-a60c-1bf467455cffI am of the opinion that I have created pretty relaxing funk
>>102016102>and they called me crazy lmaoI can't imagine why they would think that
>>102017356synthwave, retrowave, futurewave, vaporwave and outrun are all names for pretty much the same sound, if not the exact same sound in some cases. idk how much distance there is between any of these terms as its not really my wheelhouse.to make matters more confusing, they double as music genres and art styles. vaporwave is that miami sunset aesthetic while outrun is a darker variation
>>102017441>>102017457thats nice
>outrun is a music genre and artstyle nig*bo's brainrot is complete, blind and deaf
>>102017468Thank you! Put some effort in to make flux make consistent nice anime that doesn't change each gen
>>102016775She looks pretty serious if you ask me.
>>102017491makes sense considering the state of his DJ "career"
>>102017500hunting's a fun, but no joking matter, anongotta take the shot as it presents itself, gotta be ready
Coomers guide to flux where?
>>102017616download coom loras from civitai, use em, prompt, done .. be warned tho doesnt work for 2D for now, only for 3D
>>102017461mmh well the darker tone thenthat and daft punk gens when I'm full again
>>102017685I tried but it just won't work well
>>102017461Nice
Not cool making fun of debos career guys
https://youtu.be/-gGA7tXSC5M?si=Ng5G_avBwbnNGYo2messin with lcm sampler and lcm lora againe
>>102017877fire
>>102018179Thanks. That's really sharp, nice details.
>>102018301>>102017877Is this flux? It looks amazing
>>102018160Damn thats almost uncanny
>>102018356Belle Delphine lora + 80s movie and 80s outfit wildcards
skins shot on this one but i liked the thorns>>102018301thanks, youre nailing the retro film camera feel>>102018394really cool style
>>102018431Nice.
>>102018337Yeah, it's Flux with this lora. The images keep coming out kind of soft (blurry) thoughhttps://civitai.com/models/671094/flux-lora-dark-fantasy-80s-aesthetics-popular-on-tiktok?modelVersionId=751285
>>102018301Very nice. I'd watch that.
>>102018394Best I can do =D
>>102018456>made from sdxl gens>made from previous model sampleswhy are they all like this
>>102018456Cool, is using LoRAs on FLUX any different than using LoRAs on 1.5/XL?
>>102018559Nice
>>102018431I tried switching samplers, but still not as sharp as the example images for some reason. These are cool.>>102018570I didn't notice it said that. I don't know why people keep training on 1.5 and XL gens>>102018566No, I'm using Forge and it works exactly the same
>>102018633>No, I'm using Forge and it works exactly the sameno, i meant the loras. they're all made with ai gens from their own previous models, so it's already low quality, a copy of a copy of a copyand like 90+% of current flux loras say something similar>used my previous model "best images" to create>used ## of images generated from various modelsetc etc
>>102018597Thanks, but doesn't hold a candle to yours in likeliness
>>102018450you as well, i really like the color grading in that.
>>102018688I like the lips but, not a fan of self harm :/
>>102018671There are some good ones trained on non-Ai datasets. I posted this one the other night, but it says it's a merge of a 1.5 model and it really looks like 1.5https://civitai.com/models/656083/copycat-flux-testfp8fp16
>>102018688Nice>I do also adjust tone/contrast/levels all the bells and whistles,, initial gen was done with https://files.catbox.moe/v0trts.zip uses 3 samplers adding details
>>102018772best process imo, ive found its moot to try and get the model to add noise past a slight blur, and if I actually want to post it or finish it I just add it in post on photoshop + minor color fixes>>102018752dw its just an animation clipping midway thru
>>102018892heh :D
>Who's there?
>>102018892Cool.
>tuff times ahead
>>102019085I want a slurpee
Guis, I miss pdubs :<
>>102019158Funny thing is, that was supposed to be a smoothie. I only just now realize the funky colors. Shame the container rarely comes out right. Either way, it didn't stop me from genning a bunch of these
>>102019280Can't tell either way because it looks like shit
>>102019253same
>>102019322Love the buff shoulders and the clothing detail
>>102019372>>102019372>>102019372
>>102018633Think you mixed up the replies but I'm glad the LoRAs are actually used the same way.