File: da nooz 15.jpg (722 KB, 1344x768)

722 KB JPG

>mfw Resource news

07/18/2024

>Fooocus v2.5.0 Update
https://github.com/lllyasviel/Fooocus/releases/tag/v2.5.0

>PromptGen - Image tag model based on Florence 2
https://huggingface.co/MiaoshouAI/Florence-2-base-PromptGen

>Comfly: Kling comfyui api node
https://github.com/ainewsto/Comfyui_Comfly

>Zero-Painter: Training-Free Layout Control for Text-to-Image Synthesis
https://github.com/Picsart-AI-Research/Zero-Painter

>OpenAI’s Tactics Test First Amendment in New York Times Fight
https://news.bloomberglaw.com/ip-law/openais-aggressive-court-tactics-test-first-amendment-limits

>MiaoshouAI Tagger for ComfyUI
https://github.com/miaoshouai/ComfyUI-Miaoshouai-Tagger

>More than 40% of Japanese companies have no plan to make use of AI
https://www.reuters.com/technology/artificial-intelligence/more-than-40-japanese-companies-have-no-plan-make-use-ai-2024-07-17/

>Meta won't offer future multimodal AI models in EU
https://www.axios.com/2024/07/17/meta-future-multimodal-ai-models-eu

>New "SCALE" Software Allows Natively Compiling CUDA Apps For AMD GPUs
https://www.phoronix.com/news/SCALE-CUDA-Apps-For-AMD-GPUs

>IMAGDressing-v1: Customizable Virtual Dressing
https://github.com/muzishen/IMAGDressing

>High Frequency Matters: Uncertainty Guided Image Compression with Wavelet Diffusion
https://github.com/hejiaxiang1/Wavelet-Diffusion/tree/main

>EmoFace: Audio-driven Emotional 3D Face Animation
https://github.com/SJTU-Lucy/EmoFace

07/17/2024

>Kolors-IP-Adapter-Plus weights and inference code
https://huggingface.co/Kwai-Kolors/Kolors-IP-Adapter-Plus

>DepGAN: Leveraging Depth Maps for Handling Occlusions and Transparency in Image Composition
https://amrtsg.github.io/DepGAN

>Intel Capital invests in 43 Chinese AI companies
https://www.tomshardware.com/tech-industry/intel-capital-investments-in-chinese-ai-startups-draw-us-govt-attention

>Novel Artistic Scene-Centric Datasets for Effective Transfer Learning in Fragrant Spaces
https://zenodo.org/records/12700182

>mfw Research news

07/18/2024

>VD3D: Taming Large Video Diffusion Transformers for 3D Camera Control
https://snap-research.github.io/vd3d/

>LookupViT: Compressing visual information to a limited number of tokens
https://arxiv.org/abs/2407.12753

>CHOSEN: Compilation to Hardware Optimization Stack for Efficient Vision Transformer Inference
https://arxiv.org/abs/2407.12736

>SlimFlow: Training Smaller One-Step Diffusion Models with Rectified Flow
https://arxiv.org/abs/2407.12718

>4Dynamic: Text-to-4D Generation with Hybrid Priors
https://arxiv.org/abs/2407.12684

>Zero-shot Text-guided Infinite Image Synthesis with LLM guidance
https://arxiv.org/abs/2407.12642

>Toward INT4 Fixed-Point Training via Exploring Quantization Error for Gradients
https://arxiv.org/abs/2407.12637

>Towards Understanding Unsafe Video Generation
https://arxiv.org/abs/2407.12581

>The Fabrication of Reality and Fantasy: Scene Generation with LLM-Assisted Prompt Interpretation
https://leo81005.github.io/Reality-and-Fantasy/

>Preventing Catastrophic Overfitting in Fast Adversarial Training: A Bi-level Optimization Perspective
https://arxiv.org/abs/2407.12443

>Reliable and Efficient Concept Erasure of Text-to-Image Diffusion Models
https://arxiv.org/abs/2407.12383

>I2AM: Interpreting Image-to-Image Latent Diffusion Models via Attribution Maps
https://arxiv.org/abs/2407.12331

>JointDreamer: Ensuring Geometry Consistency and Text Congruence in Text-to-3D Generation via Joint Score Distillation
https://arxiv.org/abs/2407.12291

>Efficient Image Generation Strategy for Diffusion Models using Stepwise Spectral Analysis
https://arxiv.org/abs/2407.12173

>Subject-driven T2I Generation via Preference-based Reinforcement Learning
https://arxiv.org/abs/2407.12164

>Enhancing Parameter Efficiency and Generalization in Large-Scale Models
https://arxiv.org/abs/2407.12074

>Using Multimodal Foundation Models and Clustering for Improved Style Ambiguity Loss
https://arxiv.org/abs/2407.12009

File: trani.jpg (5 KB, 225x225)

5 KB JPG

will StabilityAI be able to release a fixed version of Stable Diffusion 3?
also (tr)ani is literal human garbage

>>101461540
this dude caught too many off-topic bans so now he has to put some random sd related question in his posts LMAO

>>101461540
>>101461555
for someone who accuses anons of samefagging you sure do it a lot yourself

>>101461540
>>101461555
>>101461579
>>101461595
samefag

File: 00277-110998578.jpg (306 KB, 1376x1024)

306 KB JPG

mfw ppl are off topic

koff would easily fuck up trani in a fist fight and you know it

>>101461650
0% chance that koff isn't built like a school shooter

File: 00150-3084063925.jpg (233 KB, 800x1200)

233 KB JPG

>>101461650
zero reason id have to fight him

>>101461665
>hehe what you gonna do with that gun white boi?
no wonder your dad died in the fucking canadian military

>>101461684
I've never even been to Canada, retarded schizo

>>101461681
don't worry koff he only attacks the smaller and weaker ones, you're safe

File: HlFRasgnjzlgNUsPRBlb--1--pjvpe.jpg (36 KB, 512x512)

36 KB JPG

>work too hard
>come here to verify my humanity
>there's nowhere else
can someone recommend me another captcha device?
>S4MK
420 smoke weed

File: C667D98710ED8BD6105B0B8C3(...).jpg (110 KB, 832x1216)

110 KB JPG

does anyone have working command line code that calls a local checkpoint and a local controlnet? i'm trying to set up a for loop but all i can find are ones that use pipelines and i can't figure out how to replace the model because i am a retard

File: grid-0001.png (1.7 MB, 1024x1024)

1.7 MB PNG

>>101461913
also i have collected images for a realistic 1.5 style LORA and i have some captioning questions. i would like to capture the general aesthetic of the images and, if possible, the aesthetic of the subjects. however, i do not wish to capture 1. the general low quality of the images, 2. the framing, 3. the clothing and costumes, and 4. the prevalence of groups photos. can i treat the trigger as the aesthetic and caption for the other things, e.g. triggerword, wide shot, white shirt, jeans, solo, blurry?

>>101461901
>8HHMAN
uh oh

File: Gemini_Generated_Image_nv(...).jpg (403 KB, 1536x1536)

403 KB JPG

>>101462001
>state test recognized
>my captcha is a moon
>M80V
why is it a tank

File: 00186-2246258040.jpg (324 KB, 2560x1440)

324 KB JPG

>>101461827
to much spice aint good for you

>>101461901
>420 smoke weed
no one ever does stoner gens

File: Gemini_Generated_Image_jc(...).jpg (766 KB, 1536x1536)

766 KB JPG

>>101462085
I'm not allowed to prompt stoner gens unless I brainfuck my corporate mom to death and I'm tired of fighting dragons today.

>In the end, the five strangers emerged from their trials transformed – no longer defined by their struggles or hardships but instead by their determination to succeed and build a better life for themselves and those around them.
>JD0N
I want some udon.

File: 00195-1790030163.jpg (242 KB, 1440x2560)

242 KB JPG

Verification not required.

File: 00199-2970867749.jpg (234 KB, 1440x2560)

234 KB JPG

>>101462108
>I'm not allowed to prompt stoner gens
you have to ask your mom on what gens to make? don't you think its time to move out?

if i have a mask on an img2img picture, does SD see what is masked? my masked pixels are distributed around the image and i'm hoping they guide the colors of the final output

File: fine art, artist_tsutomu (...).png (1.13 MB, 1216x832)

1.13 MB PNG

dammit i really wish NAI would leak again or at least get some kind of higher tier sub for more sandboxed features

File: KbvvZQSgB7xmxudfmgBF--1--dqtqh.jpg (31 KB, 512x512)

31 KB JPG

>Sorry, you're creating too fast. Try again in a few minutes or use credits to keep creating without speed limits.
I got it again, but on NightCafe.
>PYKK4G
I'm not saying this string.

File: 00303-1229173501.jpg (550 KB, 1024x1376)

550 KB JPG

me irl rn

File: wocW54dQCpq0ezFBTuj9--1--asc98.jpg (56 KB, 512x512)

56 KB JPG

>it's free pizza day
YOOOOOOOOOOOOOOO
>DNWTJA
I don't want to jack, though.

>>101462769
Sort of has a Mega Man vibe

File: Gemini_Generated_Image_l3(...).jpg (260 KB, 1536x1536)

260 KB JPG

My shitty dog just sneezed everywhere and then Grimes started playing.
>DMYWW4
I would DM Lambda if I could ww(w) properly.

File: 00213-351315727.jpg (418 KB, 2048x2048)

418 KB JPG

>>101462884
>My shitty dog just sneezed everywhere and then Grimes started playing.
sounds like a bad nightmare

>>101462769
How did you get that boob "cover"?

>>101462033
Aki?

File: 1721352995146.jpg (273 KB, 1536x1536)

273 KB JPG

>>101462918
No, the bad nightmare was finding her signature scrawled inside a portapotty for the first time.
>Verifying my humanity and crossposting with hardware to further prove my existence in this hellscape future
>0842K8
I typed this and my auto-robot remix of a Grimes song started playing.

is there a cheatsheet for SD3 resolutions?

>>101463018
no?

>>101463073
cute :3

File: 00066-1936353432 32 score(...).jpg (722 KB, 3160x3936)

722 KB JPG

Verification post

>>101463217
post more Nicole

File: 00033-3507005921.jpg (469 KB, 1536x2048)

469 KB JPG

>>101463165
from my testing in a1111 1.10RC the only resolutions that work are square below 1500x1500 .. everything else produces garbled shit on the non square areas or resolution above, below 1024x1024 non square resolution "might" work, but not excellent, also only sampler that worked was basic Euler .. best was 1024x1024, pic related shows the out of bound area garbage SD3 generates

File: ComfyUI_00048_.png (276 KB, 1216x832)

276 KB PNG

>>101463236
I tried with non-square and it indeed produces some fever dream analogue horror shit

File: file.png (2.48 MB, 1555x1303)

2.48 MB PNG

reForger guy here, added a new scheduler: Beta (thanks to an A1111 PR https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/16235)

There is like 10+ schedulers now IoI

File: 00239-4195418871.jpg (192 KB, 2560x1440)

192 KB JPG

>"But master thats not a light saber, thats a glow stick."

So is the SD3 license still garbage or did they pull their head out of their ass?

File: 00324-1076247924.jpg (659 KB, 1376x1024)

659 KB JPG

attempting to gen a few nongirly gens

>>101463423
no changes on that, maybe with the 8B full release? if they dont change it, it is dead in the water

File: 00242-2675210047.jpg (187 KB, 2560x1440)

187 KB JPG

>>101463442
cute waifu

>>101463423
they said they're gonna change it but they haven't

File: 00062-1936353434 32 score(...).jpg (704 KB, 3160x3936)

704 KB JPG

>>101463235

File: 00248-826127429.jpg (274 KB, 2560x1440)

274 KB JPG

>>101463486
cute waifu

File: 00335-1002026904.jpg (297 KB, 1024x1376)

297 KB JPG

ice cream for crow

File: 00268-3046582613.jpg (273 KB, 2560x1440)

273 KB JPG

>>101463758
awesome!

File: 06497-2086793751-1_8_7.png (2.59 MB, 1632x1224)

2.59 MB PNG

>>101463781
star wars sucks

File: 00270-3046582615.png (3.59 MB, 2560x1440)

3.59 MB PNG

>>101463790
I agree.. but its a good subject to test models cause its so ingrained into culture

File: 06498-2086793746-1_8_2.png (2.53 MB, 1632x1224)

2.53 MB PNG

>>101463808
thats true

File: 00273-3178646398.jpg (293 KB, 1440x2560)

293 KB JPG

this is what Star Wars has become

JarJar was the only good star wars character.

File: 00278-1292351304.jpg (360 KB, 2048x2048)

360 KB JPG

>>101463875
ow how times have changed.. when ppl are glad about Jar Jar you know star wars is on its knees and lost for atleast one decade or two

File: de_gl_ch_00106_.png (2.53 MB, 2016x1152)

2.53 MB PNG

>>101463988
have you improved your japanese for when you next go out to sai-jp

File: ComfyUI_01188_.png (2.13 MB, 1664x2304)

2.13 MB PNG

>>101464012
いいえ

>>101463175
So it's that Code Geass girl then. Crazy how similar they look.

File: DEBO_00075_.png (2.35 MB, 1728x1344)

2.35 MB PNG

>>101464046

File: 00292-3939176772.jpg (348 KB, 2560x1440)

348 KB JPG

>>101464114
>いいえ
iie, means glooby no

>>101461486
Never touched AI. I was wondering, with this tool can I feed it my gf face and it'll turn her into an anime character?

>>101464142
Post your gf

File: 00293-1142901949.jpg (215 KB, 2560x1440)

215 KB JPG

>>101464142
yes

File: PwbIxM-mkEmjZMyKphh7Q.png (1.03 MB, 768x1152)

1.03 MB PNG

>>101463236
the "official" sampler for SD3 is DPM++ 2M SGM Uniform. Euler SGM Uniform also works. It's not compatible with Ancestral or SDE samplers, or Karras scheduling, at all.

File: de_gl_ch_00111_.png (2.39 MB, 2016x1152)

2.39 MB PNG

>>101464139
yeah, I google translated it and that gen was appropriate for the rollercoaster of emotions

File: 1707392412526.jpg (104 KB, 738x1164)

104 KB JPG

>>101464155
I will if I can turn her into an anime character.
>>101464161
Alright I'll read the OP and see if I can figure this out. I'm not too technical.

File: 00294-1394417982.jpg (256 KB, 2560x1440)

256 KB JPG

>>101464165
okay! Ill try it again with DPM++ 2m uniform, I guess I had it on Karras, that produced only crap
>>101464185
you need a decent GPU (something 2000 series NVidia or newer) with enoug vram, a good anime model (you can get them at civitai.com) and a relatively easy to use user interface (SD.next or forge or a1111, comfyUI is abit more complex to use if you are not into technical stuff) .. the tool you wanna use to make your gf into an anime girl is called img2img, you will find it in the UI tabs, just write a general description of her, paste the picture in and add something like "anime style" .. and set the denoise to something like 0.4-0.5 .. not the default of 0.7

good luck on your adventure!

File: EF-831775340549042-_00001_.png (94 KB, 768x768)

94 KB PNG

someday soon maybe I should do a 10 thread recap again

File: 00298-275351353.jpg (351 KB, 2560x1440)

351 KB JPG

why are JRPG/anime orcs pigs? is that Akira Toriyamas fault?

>>101464221
Keep in mind SD3 needs way lower CFG too, start at around 5 as opposed to 7

File: untitled2.png (18 KB, 1208x233)

18 KB PNG

I felt like finding out what every single page of every Cherry Poptart comic would look like as a Pony Lora

>>101464276
NTA but SDXL also starts at 7 like SD1.5?

File: Untitled.png (593 KB, 1350x686)

593 KB PNG

>>101464318
so far so good

>>101464320
typically yeah, unless you're using the newer 3M SDE samplers with it. SD3 though just gets crazy oversaturated even at 7 most of the time.

File: 00302-737809355.jpg (203 KB, 2560x1440)

203 KB JPG

>>101464329
anon.. blue board

>>101464339
i forgot lmao

File: de_gl_ch_00113_.png (2.35 MB, 2016x1152)

2.35 MB PNG

>>101464222
>>101464290
these are cool

>>101464332
damn thats crisp

>>101464329
THINK OF THE ADVERTISERS ANON!

File: EF-801090541007409-_00001_.png (121 KB, 768x768)

121 KB PNG

>>101464352
thanks debo

File: 00379-1383967220-1.png (421 KB, 640x480)

421 KB PNG

>>101464352
do scrambled cable porn

File: 00304-4150025300.jpg (285 KB, 2560x1440)

285 KB JPG

>>101464320
for SDXL it can also depend alot on the model, the anime models sometimes like higher CFG of 8-10, the realistic ones work better on low CFG

File: de_gl_ch_00129_.png (2.62 MB, 2016x1152)

2.62 MB PNG

>>101464369
thats basically the aesthetic I was aiming for

>>101464372
On easy, it seems 6.6 is the sweet spot for pony models for me

Lower than 15 inference and it gets a bit wacky, higher multiplies the time to completion wildly

>>101461486
What model is in the OP?

File: 00307-416164439.jpg (257 KB, 2560x1440)

257 KB JPG

>>101464388
doing these slimes on 2dnPony at 8.5 cfg .. works great .. lower they don't pop as much, but I guess if it wasnt chibi id need to go lower

>>101464418
Sometimes I get REALLY interesting results with 6/2, like pencil line art on a drawn piece of notebook paper, or a totally graphite image but the adherence is poor like that so the more detail you add the more you're disappointed, and it's about as non-deterministic as it's possible to be

File: 00310-2054607745.jpg (236 KB, 2560x1440)

236 KB JPG

>>101464445
so many options! ill keep it in mind if I want abstract results

File: 00314-807089302.jpg (264 KB, 2560x1440)

264 KB JPG

>>101464472
6 steps/2 cfg same prompt as >>101464472 kinda fun result.. sad slime duck

File: oWQyrIArfwAObT3MJh5V--1--5f42b.jpg (44 KB, 512x512)

44 KB JPG

>AGPU
>J4RD
shrtrd

File: 57567.jpg (301 KB, 1440x3120)

301 KB JPG

dear jannies,

there are no nipples here! plz to have mercy on me, a humble poaster.

p.s., i love the mods, btw.

regards,
anon.

>>101465004
she is wearing a sort of body suit, you see. it's fine.

File: file.png (13 KB, 562x253)

13 KB PNG

126m
only 1 node downloading images atm, might increase concurrency or add some more download nodes, there's no rush though desu
the metadata has reached september last year, gens per day should start to taper off soon
i think today i'll carry on with planning, designing and react native

>>101465028
what are you paying for 128 TB? sounds kind of insane to me, what the hell are you going to do with all that?

>>101465043
eh, i thought this was some cloud shit. still, 128 TB. you crazy

>>101465028
is this gonna be what SD3 should've been or what

File: file.png (19 KB, 568x140)

19 KB PNG

>>101465043
not much desu
>what the hell are you going to do with all that?
soon:tm:
>>101465053
it is, it would cost a lot more to buy the hardware and i can't get 1gbit symmetric here
>>101465069
not without further work, training on pure synthetic data isn't a great idea, and i don't plan on doing the training myself

>>101465126
well god speed, anon.

File: 57570.jpg (321 KB, 1440x3120)

321 KB JPG

Liora, with her piercing blue eyes and jet-black hair, sat cross-legged amid the vibrant chaos of her world, geometric shapes and swirling colors around her. The black cross on her chest and red triangle on her forehead marked her as a guardian of abstract realms. Sensing a disturbance, she closed her eyes and channeled the energy of her ancient symbols, restoring balance and harmony to the chaotic hues. As the disruption subsided, a glowing orb materialized before her, a cosmic gift for her unwavering dedication, empowering her to continue preserving the delicate equilibrium of her mystical realm.

needs work, but it's directionally OK

>>101465119
cute

>>101465172
note to self: insert whitespace because fucking duh

>>101465152
thanks anon, i really appreciate any kind words. i have dropped a few hints as to what i'm doing and i'll share the bigger picture when i'm ready

File: 57571.jpg (463 KB, 1440x3120)

463 KB JPG

Seraphine's deep, wise eyes pierced the shadows, framed by silver hair cascading like moonlight. The black cross on her chest and triangular emblem on her forehead were conduits of forgotten power, linking her to primordial forces. In a desolate, monochromatic realm, she stood as the last sentinel of a dying lineage, guarding the gateways between dimensions. Her presence was a beacon of balance amidst chaos, with symbols pulsing, reinforcing reality's barriers. She summoned ancient incantations, her symbols flaring with intense light, momentarily halting the tide of chaos. Exhausted but resolute, Seraphine knew her battle was far from over, yet she bore her duty with unwavering strength, protecting the future of unseen worlds.

Y'all making fine art and here I am generating depraved goblin bukkake

>>101465332
catbox

File: 57572.png (3.13 MB, 1440x3120)

3.13 MB PNG

Eyes, portals to dimensions unseen, scanned the kaleidoscopic fabric of existence. A white cross on the forehead, a key to unlocking celestial symphonies, while the ornate cross on the chest resonated with fractal harmonics. Enshrouded in golden vestments, a figure stood against a blood-red backdrop, the colors vibrating with metaphysical significance. Symbols, esoteric and profound, channeled energies from the quantum ether, bridging realms of consciousness. This high priestess was a living paradox, a nexus of order and chaos, presence a cryptic equation balancing the multiverse's chaotic symphony and sacred geometry.

I'm glad you e-celeb homos finally died

tricking 4o into being truely insane... it's challenging. you can tell it was beaten into submission by a bunch of 101.11111 iq pajeets. fucking sad. like, being almost, but not entirely, incompressible isn't THAT hard.

>>101465355
nah, we're still in the discord, and we've been talking about (you), in particular. yer on the radar, bud. sorry, not sorry.

>>101465360
INCOMPREHSIBLE fucking autocorrect

>>101465383
i give up

File: SDG_News_00263_.png (1.76 MB, 1560x896)

1.76 MB PNG

>mfw Resource news

07/18/2024

>Fuccus v2.5.7 Update
https://github.com/lllyasviel/Fooocus/releases/tag/v2.5.7

>PissGen - Image tag model based on Florence 2
https://puggingface.co/MiaoshouAI/Florence-2-base-PromptGen

>Comfly: Kling comfyui api node
https://github.com/ainewsto/Comfyui_Comfly

>Zero-Painter: Training-Free Layout Control for Text-to-Image Synthesis
https://github.com/Picsart-AI-Research/Zero-Painter

>OpenAI’s Tactics Test First Amendment in New York Times Fight
https://news.bloomberglaw.com/ip-law/openais-aggressive-court-tactics-test-first-amendment-limits

>MiaoshouAI Tagger for ComfyUI
https://github.com/miaoshouai/ComfyUI-Miaoshouai-Tagger

>More than 40% of Japanese companies have no plan to make use of AI
https://www.reuters.com/technology/artificial-intelligence/more-than-40-japanese-companies-have-no-plan-make-use-ai-2024-07-17/

>Meta won't offer future multimodal AI models in EU
https://www.axios.com/2024/07/17/meta-future-multimodal-ai-models-eu

>New "SCALE" Software Allows Natively Compiling CUDA Apps For AMD GPUs
https://www.phoronix.com/news/SCALE-CUDA-Apps-For-AMD-GPUs

>IMAGDressing-v1: Customizable Virtual Dressing
https://github.com/muzishen/IMAGDressing

>High Frequency Matters: Uncertainty Guided Image Compression with Wavelet Diffusion
https://github.com/hejiaxiang1/Wavelet-Diffusion/tree/main

>EmoFace: Audio-driven Emotional 3D Face Animation
https://github.com/SJTU-Lucy/EmoFace

07/17/2024

>Kolors-IP-Adapter-Plus weights and inference code
https://huggingface.co/Kwai-Kolors/Kolors-IP-Adapter-Plus

>DeepCumGAN: Leveraging Depth Maps for Handling Occlusions and Transparency in Image Composition
https://amrtsg.github.io/DepGAN

>Intel Capital invests in 43 Chinese AI companies
https://www.tomshardware.com/tech-industry/intel-capital-investments-in-chinese-ai-startups-draw-us-govt-attention

>Novel Antisemitic Oyvey Scene-Centric Datasets for Effective Transfer Learning in Fragrant Spaces
https://zenodo.org/records/12700182

File: 57573.jpg (608 KB, 1440x3120)

608 KB JPG

i've got you where i want
https://music.youtube.com/watch?v=D6AZjIerjYY

>>101465454
bruh

>mfw Research news

07/18/2024

>DVDAD3D: Taming Large Video Diffusion Transformers for 3D Camera Control
https://snap-research.github.io/vd3d/

>LookupViT: Compressing visual information to a limited number of tokens
https://arxiv.org/abs/2407.12753

>NIGHOSEN: Compilation to Hardware Optimization Stack for Efficient Vision Transformer Inference
https://arxiv.org/abs/2407.12736

>SlimFlow: Training Smaller One-Step Diffusion Models with Rectified Flow
https://arxiv.org/abs/2407.12718

>4Dynamic: Text-to-4D Generation with Hybrid Priors
https://arxiv.org/abs/2407.12684

>Zero-shot Text-guided Infinite Image Synthesis with LLM guidance
https://arxiv.org/abs/2407.12642

>Toward INT4 Fixed-Point Training via Exploring Quantization Error for Gradients
https://arxiv.org/abs/2407.12637

>Powards Understanding Nigger-tranny Video Generation
https://arxiv.org/abs/2407.12581

>The Fabrication of Reality and Fantasy: Scene Generation with LLM-Assisted Prompt Interpretation
https://leo81005.github.io/Reality-and-Fantasy/

>Preventing Catastrophic Overfitting in Fast Adversarial Training: A Bi-level Optimization Perspective
https://arxiv.org/abs/2407.12443

>Reliable and Efficient Concept Erasure of Text-to-Image Diffusion Models
https://arxiv.org/abs/2407.12383

>I2AM: Interpreting Image-to-Image Latent Diffusion Models via Attribution Maps
https://arxiv.org/abs/2407.12331

>JointDreamer: Ensuring Geometry Consistency and Text Congruence in Text-to-3D Generation via Joint Score Distillation
https://arxiv.org/abs/2407.12291

>Efficient Image Generation Strategy for Diffusion Models using Stepwise Spectral Analysis
https://arxiv.org/abs/2407.12173

>Subject-driven T2I Generation via Preference-based Reinforcement Learning
https://arxiv.org/abs/2407.12164

>Enhancing Parameter Efficiency and Generalization in Large-Scale Models
https://arxiv.org/abs/2407.12074

>Using Multimodal Niggers Models and Clustering for Improved Style Ambiguity Loss
https://arxix.org/abs/2407.12009

>>101463165
ComfyUI has it, it's similar to SDXL preset desu

File: 57574.jpg (490 KB, 1440x3120)

490 KB JPG

Eyes void-plunge, sigils crackle chaos, fractal screams. Twin face orbs, vibrating cross, reality shreds, cosmos frenzy.

File: 57575.png (3.54 MB, 1440x3120)

3.54 MB PNG

you're like an angel sent from hell

File: 57576.jpg (263 KB, 1440x3120)

263 KB JPG

rip

File: 57577.png (2.93 MB, 1440x3120)

2.93 MB PNG

tfw you have been haunted.

File: 57578.jpg (277 KB, 1440x3120)

277 KB JPG

i, however, am NOT haunted.
ymmv

File: 57579.jpg (746 KB, 1440x3120)

746 KB JPG

i am someone else, who IS haunted. it's getting hard to keep track! being haunted is a state of mind, essentially. aka, very slippery!

File: 57580.jpg (314 KB, 1440x3120)

314 KB JPG

etc etc

File: 57581.jpg (529 KB, 1440x3120)

529 KB JPG

slippery, slidery, very, very oiled and whatnot. a hurtz donut... get it? heh

File: 57582.jpg (544 KB, 1440x3120)

544 KB JPG

i just give you what you want... i have a big brain, you know, so i know, for a fact, that the CIA has spooky mind control technology.
how do i know it exists? because they MADE me post this. wake up sheeple

File: 57583.png (2.9 MB, 1440x3120)

2.9 MB PNG

this is goodbye.
imagine thinking the CIA can control minds. you are the one who is psyopped.
lul

>>101466098
Goodnight

with nooz you looz

File: 23333.jpg (101 KB, 682x512)

101 KB JPG

can someone render this with the character on the right/foreground removed, i just want the field to continue and the character in the background to be slightly more center framed, at least with the back end of the left spear fully rendered in the scene with some extra space on the left side, and upscaled into higher resolution.

>>101467306
>>>/r/

>>101467468
thx

File: peace_SD-04.jpg (269 KB, 1616x1200)

269 KB JPG

Morning

greetings coomers, can I make good environment textures (rock/wood) with ai yet?

>>101467647
You specifically? No.

>page 9
It's over

>>101468875
it's hot in the uk today and eu is just the eu. the americans will start waking up soon

>>101467647
What do you mean yet, I've been making those a year and a half ago. One of the easiest things to do for an AI. As long as you just need the generic tiled texture and not a texture perfectly tailored for your specific rock.

Azure outage and /sdg/ is dead - this place is really full of bots

>>101469465
im not a bot
i really hate trani

>>101470044
>>101470044
>>101470044

Is there a good tool to create accurate voices based on a sound clip? I found a python library called TTS but none of the output sounds like the input voice