walking on grassy mountains EditionPreviously on /sdg/: >>101053907>Beginner UI local installFooocus: https://github.com/lllyasviel/fooocusEasyDiffusion: https://easydiffusion.github.io >Local installAutomatic1111: https://github.com/automatic1111/stable-diffusion-webuiComfyUI (Node-based): https://rentry.org/comfyuiAMD 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 >Auto1111 forksForge: https://github.com/lllyasviel/stable-diffusion-webui-forgeAnapnoe UX: https://github.com/anapnoe/stable-diffusion-webui-uxVladmandic: https://github.com/vladmandic/automatic >Run cloud hosted instancehttps://rentry.org/sdg-link#run-cloud-hosted-instance >Try online without registrationtxt2img: https://www.mage.spaceimg2img: https://huggingface.co/spaces/huggingface/diffuse-the-restInpainting: https://huggingface.co/spaces/fffiloni/stable-diffusion-inpaintingpixart: https://huggingface.co/spaces/PixArt-alpha/PixArt-Sigma >Models, LoRAs & embeddingshttps://civitai.comhttps://huggingface.cohttps://rentry.org/embeddings >Animationhttps://rentry.org/AnimAnonhttps://rentry.org/AnimAnon-AnimDiffhttps://rentry.org/AnimAnon-Deforum >SDXL info & downloadhttps://rentry.org/sdg-link#sdxl >Index of guides and other toolshttps://codeberg.org/tekakutli/neuralnomiconhttps://rentry.org/sdg-linkhttps://rentry.org/rentrysd >View and submit GPU performance datahttps://docs.getgrist.com/3mjouqRSdkBY/sdperformancehttps://vladmandic.github.io/sd-extension-system-info/pages/benchmark.html >Share image prompt info4chan removes prompt info from images, share them with the following guide/site...https://rentry.org/hdgcbhttps://catbox.moe>Discordr2yJJDkm >Related boards>>>/h/hdg>>>/e/edg>>>/d/ddg>>>/b/degen>>>/vt/vtai>>>/aco/sdg>>>/trash/sdg
>mfw Resource news0619/2024>TroL: Traversal of Layers for Large Language and Vision Modelshttps://github.com/ByungKwanLee/TroL>diff-sampler: open-source toolbox for fast sampling of diffusion modelshttps://github.com/zju-pi/diff-sampler>Diffusers Adds≈ SD3 ControlNet and Multi-ControlNet Supporthttps://github.com/huggingface/diffusers/pull/8566>Better & Faster Large Language Models via Multi-token Predictionhttps://huggingface.co/facebook/multi-token-prediction06/18/2024>Release of training script of PCM-LoRA with Stable Diffusion 3.https://github.com/G-U-N/Phased-Consistency-Model#news>Autoregressive Model Beats Diffusion: Llama for Scalable Image Generationhttps://peizesun.github.io/llamagen/>Flash Diffusion: FlashSD3https://huggingface.co/jasperai/flash-sd3>SD3-Controlnet-Tile https://huggingface.co/InstantX/SD3-Controlnet-Tile>Sharing new research, models, and datasets from Meta FAIRhttps://ai.meta.com/blog/meta-fair-research-new-releases/>The next chapter for ComfyUIhttps://blog.comfyui.ca/comfyui/update/2024/06/18/Next-Chapter.html>DeepFuze: Deep Learning Tool Integrating Lipsync, Face Swap, VidGen and morehttps://github.com/SamKhoze/ComfyUI-DeepFuze>Glyph-ByT5-v2: A Strong Aesthetic Baseline for Accurate Multilingual Visual Text Rendering https://huggingface.co/GlyphByT5/Glyph-SDXL-v2>MegaScenes: Scene-Level View Synthesis at Scalehttps://megascenes.github.io/>microsoft/Florence-2-large - New 0.23B | 0.77B Modell for image captioninghttps://huggingface.co/microsoft/Florence-2-large06/17/2024>CivitAI: Temporary Stable Diffusion 3 Banhttps://civitai.com/articles/5732>VEGAIcon: Learning Interleaved Image-Text Comprehension in Vision-Language Large Modelshttps://github.com/zhourax/VEGA>ComfyUI-LuminaWrapperhttps://github.com/kijai/ComfyUI-LuminaWrapper>Generating audio for videohttps://deepmind.google/discover/blog/generating-audio-for-video/
>mfw Research news06/19/2024>Synergizing Foundation Models and Federated Learning: A Surveyhttps://arxiv.org/abs/2406.12844>LayerMerge: Neural Network Depth Compression through Layer Pruning and Merginghttps://arxiv.org/abs/2406.12837>VIA: A Spatiotemporal Video Adaptation Framework for Global and Local Video Editinghttps://arxiv.org/abs/2406.12831>Neural Approximate Mirror Maps for Constrained Diffusion Modelshttps://arxiv.org/abs/2406.12816>AITTI: Learning Adaptive Inclusive Token for Text-to-Image Generationhttps://itsmag11.github.io/AITTI/>Extracting Training Data from Unconditional Diffusion Modelshttps://arxiv.org/abs/2406.12752>AGLA: Mitigating Object Hallucinations in Large Vision-Language Models with Assembly of Global and Local Attentionhttps://arxiv.org/abs/2406.12718>SUPER: Selfie Undistortion and Head Pose Editing with Identity Preservationhttps://arxiv.org/abs/2406.12700>Do More Details Always Introduce More Hallucinations in LVLM-based Captioning?https://arxiv.org/abs/2406.12663>Mixing Natural and Synthetic Images for Robust Self-Supervised Representationshttps://arxiv.org/abs/2406.12368>Immiscible Diffusion: Accelerating Diffusion Training with Noise Assignmenthttps://arxiv.org/abs/2406.12303>COT Flow: Learning Optimal-Transport Image Sampling and Editing by Contrastive Pairshttps://arxiv.org/abs/2406.12140>ARTIST: Improving the Generation of Text-rich Images by Disentanglementhttps://arxiv.org/abs/2406.12044>Not All Prompts Are Made Equal: Prompt-based Pruning of Text-to-Image Diffusion Modelshttps://arxiv.org/abs/2406.12042>Decomposed evaluations of geographic disparities in text-to-image modelshttps://arxiv.org/abs/2406.11988>MAC: Benchmark for Multiple Attributes Compositional Zero-Shot Learninghttps://arxiv.org/abs/2406.12757>Disturbing Image Detection Using LMM-Elicited Emotion Embeddingshttps://arxiv.org/abs/2406.12668>What makes two models think alike?https://arxiv.org/abs/2406.12620
Everyone posting after me is a faggot.
>>101059885thx baker>>101059657>(Cohesium:1.5)i guess it works>>101059612
>>101059717All top notch!
>>101059926
SHANAANAKEE>>101059920very nice>>101059910
>>101059926>>101059931nice xayah. i haven't played league in years but that looks cool
>>101059976thanks, I don't play league much anymore either but some of my favorite character designs outside anime/eastshit
imagine the smell>>101059957
>>101059915is this the muslim cube
Holy quokkamole.
>>101059999
Why there is /sdg/ and /ldg/ nowadays?
>>101060047i still dont know whybut ldg posts sound more... reddit? to me>>101060028
>>101060047Ran and other redditors wanted their own thread and made /ldg/.
If you don't know at this point there's no reason to help you.
>>101060000Checked, Satan/Saturn Blackrock..
>>101060047autism
>>101060065
https://www.youtube.com/watch?v=BSkOzlWEmoI
>>101060131
>>101059910>(Cohesium:1.5)whether or not anon was joking about this it actually worked for my purposes lelso thank you>>101060159
https://arxiv.org/html/2406.11831v1I wonder if it's actually better than Dall-E, the benchmarks don't show anything to do with people doing stuff or styles, just composition(no one cares about this).
>>101060195
>>101060300
>>101060144Hajj
SD3 tile CN working in comfy yet?https://huggingface.co/InstantX/SD3-Controlnet-Tile
captcha: /g/ vag>>101060346
>>101060393>Hajj
>>101060421>*praying intensifies*
>>101060418
Hello bros! I am having trouble to generate a image. everything i try just makes takes it further away from the desired goal! Can somebody help me? :(I can't seem to get it anything close to my prompt!My prompt is>tall Japanese woman. Street alley at night . beautiful Japanese face covered in scars. two big scars on each side of the face. small scars on the nose. Small scars on the forehead. Small scars on the chin. redish-orange hair . messy ponytail with two loose bangs that hang on either side. eye retina is a pure sunlike yellow color. the white part of the eye is deep black in color. The surrounding area of the eyes is black . A petite body that is covered in huge and small scars . Smalls scars on the arms , huge scars on the arms , small scars on the thighs , huge scars on the tights. stomach is well toned and has a 6-pack. pointy teeth. turtleneck . lab coat . jeans. She is holding a knife covered in red liquid. The image is shot on a camera. Anime style.
>Do not direct your focus outward, go within.
Any good SD3 samples?
>>101060703no
>>101060437
>>101060703
>>101060703ok bet
Based Ella Freya enjoyer.
>>101060703fuck you 4chan
>>101060825
>>101060837
>>101060841
>>101060863
>>101060874
>>101060885there's a million more amazing imagesif someone says SD3 is bad, don't take them seriously. It has it's issues but it is NOT bad.I zipped the last time I ran this experiment and posted it. I'll probably do that again. These gens are all T5 only. no clip g or l
>>101060963Neat orb
>>101060963
>>101060974Ty, SD3 with +5 on modelsamplingsd3
>>101060980Nice.
>>101060951
>>101060755
>>101061001thx
>>101061112Sun in the cube!!soo good.
me on the right>>101061052
>>101059899>>101059890you ever going to accept my friend request on discord?at this point there are far more people in the discord than not.
>>101061166
>>101061231deep fried as fuck
>>101061233first time?
>>101061233>>101061241it's actually not lelcmyk colors>>101061231
>>101061254oh it absolutely is
>>101061271you won't get anywhere with this
>>101061271feel free to think what you wish>>101061231
fucking kek
>>101061282
>>101061231>>101061282you see how the floor is getting goosebump like textures?You see the knees?the bad chromatic aberration?the latex skin texturing?All classic signs of a deep fried gen, but feel free to be an oblivious moron
>>101061314you're assuming a lot of things>>101061303
>>101061329>points out specific examples>"you're assuming"da fuqhow retarded can you be
>>101061335nice>>101061329
>>101061352you pointed out things that would suit what you wish to see. in no way does your list of specific examples require me to agree with what you wish. that's all. as i said, feel free to think what you wish.>>101061353
Do any of you know of some software that can crop images really quickly? I'm thinking like a fixed resolution box that moves with the cursor and when I click it saves the new cropped picture. I have a lot of cropping I want to do for loras.
bring back the notables, this shit sucks
>>101061382who is a notable?
>>101061380FastStone Photo Resizer
>>101061394Ran
>>101061373this isn't about agreeing or disagreeing, deep fried gens have specific qualities which is why they're immediately identifiable as being deep fried. It's not a subjective term, it has a fixed meaning. You're just unaware of what you're trying to refute and sounding retarded in the process.
>>101061413lolyeah real notable
>>101061413I still have no idea who ran is
>>101061373
>>101061425Quiet debo>>101061421
>>101061438>dogshit gens>sucking ran's cockgoes hand in hand
>>101061437
>>101061404Saving is a little slow but this is good, thanks.
>>101061456what lora are you going to make?
>>101059915cube outta nowhere!>>101061449
>>101059885If this image was made by a real person, I would marvel at it. "Wow," I'd go, "I don't even like Halo, but it must have taken hundreds of hours of cumulative practice and knowledge to be able to make the lighting just perfectly so, to make the sheen of the armour look just that good."Instead, this was made by some 20 iq retard who wrote "Grassy hills background, Videogame in-game screenshot happy Adorable Quokka wearing halo spartan armor" into a text prompt.Nothing was gained.Nothing was lost, except some poor computer's clock cycles.Nobody worked for anything in this picture, except for the poor likely underpaid researchers who engineers who unwittingly opened this pandora's box. No craft was perfected. No skills were honed. You didn't learn anything about yourself, life, and much less of all art by "making" this picture. All you did was take one small step towards the new future of humanity, one so utterly deprived of aspiration, ambition, talent, and meaning that they comfortably slob around in their chairs while the machines that they built for them feed them and endless mindless cycle of soulless entertainment.Good luck, I say to you. You will find me in my cave, safe from the horrors of the world.
>>101061492>t. kaczynski
>>101061492researchers AND engineers*my apologies
>>101060992Are you using comfy??
First try SD3 outputMuch better than expected based on the hate
>>101061483FFT loras. I will be cropping faces from images like this and for another lora I'll be cropping the upper half or portraits of fanart to supplement style loras. I've also made some datasets for Yoshida's FF3 art, FFTA, a couple other things, and I got someone to send me raw video files from the FMVs. I think taking a bunch of screenshots of those would make a cool lora.I made one already if you want to try it. It's just using most of the full body job art.https://files.catbox.moe/r6e5fo.safetensors
>>101061492If this post was made by a real person, I would marvel at it. "Wow," I'd go, "I don't even like img gen, but it must have taken hundreds of hours of cumulative practice and knowledge to be able to make the shitposting just perfectly so, to make the sheen of the nogen look just that good."Instead, this was made by some 20 iq bot who wrote "Shit on prompt of grassy hills background, Videogame in-game screenshot happy Adorable Quokka wearing halo spartan armor" into a text prompt.Nothing was gained.Nothing was lost, except some poor computer's clock cycles.Nobody worked for anything in this shitpost, except for the poor likely underpaid anons who's bots opened this pandora's box. No craft was perfected. No skills were honed. You didn't learn anything about yourself, life, and much less of all art by "making" this shitpost. All you did was take one small step towards the new future of humanity, one so utterly deprived of aspiration, ambition, talent, and meaning that they comfortably slob around in their chairs while the machines that they built for them feed them and endless mindless cycle of soulless entertainment.Good luck, I say to you. You will find me in my cave, safe from the horrors of the world.
>>101061508Yes
>>101061492>>101061504I don't like the armor that much, i did try to make it look like halo but the model kept making armor that looked like Doom
>>101061485
Dead threadDead general
bred
>>101061590
maybe it does know
>>101061813Is that that guy from the Pattinson Batman movie.
>>101061536Neurons activated.
>>101061757
>>101061915
time to gogn all>>101061954
Florence-2 seems to be finally a good NSFW captioner. The short ones are actually the best I'm finding. some funny (accurate) samples:>A 3D rendering of a woman being fucked by two men.>An anime illustration of a woman being fucked by a werewolf in a barn.>A 3D rendering of a woman with long black hair and red lipstick being fucked by a black man.Short caption mode exclusively uses "fuck", "fucking" and "fucked" to describe the action for some reason lol
I think I assembled my 1girl wrong>>101062118gn>>101062159thats kinda funny since its a microsoft model
>>101057220>I'll use inpainting to fix it for him>inpaint: let me just get that out of there for you
>>101062159very interesting and somewhat unexpected
>>101061929>>101061929>>101061929Move to /ldg/, with Stability on its deathbed /sdg/ will be a thing of the past
I'm not moving to a containment thread
Good night fellas
>>101062322gn
>>101062316Yes you are
>>101062312You dont go to a funeral when someone is on their deathbed. You go after they die
>>101062338only if alig comes back
>>101062159i dont think its good enough and the other more descriptive modes seem to hallucinateperhaps as a supplement to another tagger or maybe if someone wanted to dropout the more descriptive tags to be able to have good results with short prompts when generating
>please you have to come to our dead thread and post to no one
no, stay here but i know you love shitting up the place so that wont happen
SAI will make an anouncment about the SD3 fiasco kek, was about fucking time
>stable diffusion is dead>posts image made with stable diffusionokay then
>>101062349Anon will just double bake an /sdg/ regardless
>>101062418people won't stick with SDXL forever, alternatives will come and move the imagegen model quality forward, like Pixart, HunyuanDiT or this shit maybe? https://arxiv.org/abs/2406.11831>Benefiting from the inherent ability of the LLMs and our innovative designs, the prompt understanding performance of LI-DiT easily surpasses state-of-the-art open-source models as well as mainstream closed-source commercial models including Stable Diffusion 3, DALL-E 3, and Midjourney V6. The powerful LI-DiT-10B will be available after further optimization and security checks.
>>101062415>SAI will make an anouncmentI hope it'll be more epic than this onehttps://www.youtube.com/watch?v=fueRUi5AWWQ
>>101062415>announcement: we're releasing 8b on our api and giving everyone a free month subscription as an apology!
>>101062234>>101057220figsed
>>101062436this is the 2nd paper to call sd3 closed source kek
>>101062239It seems the "FT" versions are the sanitized ones like in terms of what they will and won't say. The "Base" version of Large is what I've been using to get good NSFW results.
>>101058111Like so:https://civitai.com/articles/5799
>>101062344It will be a while before SAI actually dies, but better to move on now while you can
>>101062370which version were you using? Large non-FT is the best, I'm finding.I agree about using it conjunction with something else though, what I'm doing is outputting Florence first, and then running wd-vit-tagger-v3 and concatenating the list of Booru tags to come immediately after the Florence output.
>>101062476If you look at the prompts people post on there, it seems a large number use the on site image generator, which costs buzz. Hosting all those models has to be expensive though
>>101062415>HopefulyWhy the fuck SAI posts are always so cryptic and vague, if you don't know then shut the fuck up?
These really aren't nsfw despite being posted on /aco/ so I'm reposting them here.Gwen is better when she is masked.
>>101062510ty for the trans representation during pride month
>>101062491also using base large on hf
do you have to use either SD 1.5 OR an alternate model like AOM3 and not both? I was trying to somehow use 1.5 with AOM3/WaifuDiff and a VAE but couldn't figure out how to get all 3 to work together..
>>101062535Gwen isn't trans
I have a confession to make, I only fap to my slop.
>>101062556What are you trying to achieve in base 1.5 that you can't in aom3? You can always do a second pass, that way you gen on either base1.5/aom3 then pass that latent to another ksampler for the other model (and repeat for a third pass if you want)What are you trying to do ultimately is what I'm getting at? An anon might be able to put you in the right direction
>>101062556ty for the trans representation during pride month
>>101062535Thank you anon, but don't sell yourself so short, You are representation enough ;)
https://stability.ai/news/stable-diffusion-3-medium2b is all we gonna get>Future Plans>We plan to continuously improve Stable Diffusion 3 Medium based on user feedback, expand its features, and enhance its performance. Our goal is to set a new standard for creativity in AI-generated art and make Stable Diffusion 3 Medium a vital tool for professionals and hobbyists alike.
>>101062612>based on user feedbackMy fucking ass
>12 Jun sd3 is good and I don't give a shit what you guys say
>>101062612>Announcing the Open Release of Stable Diffusion 3 Medium, Our Most Sophisticated Image Generation Model to Date
>>101062567She died in The Amazing Spider-Man #121–122 and it fucked Peter Parker up so bad he started dressing like her and being Spider-Gwen. So, yes.
>>101062645REEEEEEEEEEEEEEEEEEEEEEEEE spoilers
>>101062574Does aom3 include 1.5 or something? I thought they were completely separate methods or whatever and I was trying to have them influence each other. I may just not have a good understanding of what models actually are... I tried merging them and now I can't seem to switch back to aom3.. it's stuck on v1-5.pruned despite that not even being in my model folderJust trying to create good weebshit stuff, and definitely not imitate the work of any artists I enjoy.
>>101062631>sd3 is good
>>101062645Good one lol
>>101062645so now everyone that crossdress is trans? I like how the woke reinforce stereotypes: "If you wear a dress you're a girl"
>>101062654At least put in some effort into highlighting some of the problems>hint it's still a great model >>101062653>Does aom3 include 1.5 or somethingIn short, yes it does - Choose it as your checkpoint/model. The only thing that relies on the checkpoint/model (yes there are other things) but the main thing would be LoRas which are tiny little models which influence your main checkpoint/model. tl;dr set Aom3 as your checkpoint/model and have fun.
>>101062693>hint it's still a great modelthe SA researchers literally called it a "failed experiment" but you still wanna pretend it's a good model? >exhibit 101 of the Stockholm syndrom
>>101062715I'm not coping, I see the results. It can yield some incredible gens and you're delusional if you think otherwise (including all of the faults).
>>101062725So you know better than the SAI researchers that created the model and went to the conclusion it was a failure? All right, you're entilted with your opinion I guess?
>>101062731We can go back and forth on this but the bottom line, for me, is the results. It yields what I want. Yes the training dataset was pruned, yes it there are a plethora of safety precautions used and at the end of the day it doesn't matter because I like what it gens. It's truly that simple.
>>101062725>I'm not copingthat's someone who's coping would say
>>101062715I don't see why they would push to put it out then. Releasing a model that people think isn't good isn't going to make anyone want to buy the 8b model.
>>101062679>I like how the woke reinforce stereotypeshe actually typed this >>101062715>made by people who left a while agodoes that mean that their attempts at training models were so bad that they had to dig through the old researchers' discard bin in the hopes that i'd be marketable?
>>101062747>he actually typed thiswhere's the lie, that mf literally said that because Spiderman decided to use Gwen's clothes, that means he's trans, since when "wearing dresses = trans"? That's a fucking stereotype
>>101062747or>let's give them this one we can't profit off of for free
>>101062725I have stock in the other companies so I need it to be bad. Please say it's bad.
>>101062745>I don't see why they would push to put it out then.So that they can say "See? We fulfilled our promise of releasing SD3... it's just that it was only sd3 medium, but that's sd3 nonetheless"
>>101062759But again, by calling it SD3, that makes the 4b and 8b "look" bad
>official Stability AI thread
>>101062776Even the SAI reddit is a dumbsterfire, it became so bad it's not a place to make Luma dream machine ads kekhttps://reddit.com/r/StableDiffusion/comments/1djhtzt/im_testing_to_animate_cats_with_dream_machine_and/
I don't think we're getting a pre-release drop tonight
>>101062887guess I can go to sleep gn sdg
>>101062947gn
>>101062725post a catbox of something good made with sd3
why didn't anyone gen justin timberlake drunkenly driving into children or something? seems like it coulda been a fun genre for the day
>>101063127>>101063127>>101063127
I guess I'll fill this
END
>>101062478>but better to move on now while you canAnd why is that?
>>101061492holy based
>>101062453where's the other one
>>101064312https://github.com/Tencent/HunyuanDiT?tab=readme-ov-file#-comparisonsthey actually call it open source in the paper itself but on github comparison they have it as closed
>>101059885It's so strange i knew this was halo from the landscape in the thumbnail alone
>>101061149Any way to modify these so they look lile they were shot in the late 90s? On tape
