#sd3

[etlu-yume]

"Weekend" they said. "Free time"

And my brain proceeded to know absolutely NO chill this weekend.

(Okay technically most of "Can't hold me down 'cause you know I'm a fighter" was done last weekend, but it was still Outstanding™ and had to be finished. The rest were all done in the time since Friday afternoon.)

[ancientwhispersofbabylon]

[somethingcooltolookat]

[seriously-mike]

I just found something funny regarding Stable Diffusion 3 and its ridiculously fucked output.

The base data model is built to avoid generating NSFW content, but... if you put "NSFW" or any other NSFW tags into the negative prompt, the coherence improves for some strange and suspicious reason.

What the fuck are they trying to hide that causes this behavior? This, tangentially, reminds me of the situation where I tried to mash four celebrities, but not a fifth one, into a Western movie character using a SDXL model, and once I put the name of the fifth guy into the negative prompt, I suddenly got an entirely unrelated and even more generic character instead. There's no relation that you would expect, as opposed to me monkeying with the #Photobooth From Hell and getting a more masculine appearance from a goth/metalhead character when I explicitly added "woman" to the negative prompt in addition to asking for a man in heavy makeup.

This is like they intentionally selectively poisoned the model to fail on all attempts of generating smut and didn't tell anyone about it. Not like the negative prompts aren't a minefield anyway - aside from the issue with replicating anime, I also encountered a problem generating patterned clothing in SDXL with a typical "default" negative prompt. But then, I also mentioned that Stable Diffusion is ornery and full of garbage anyway.

[iks-iivovii]

Stable Diffusion 3 + Leonardo's refiner and universal upscaler are simply perfect.

[ailockup]

Stable Diffusion 3 Is FINALLY Here | Try SD3 Models With This Free AI Im...

[secourses]

Zero to Hero Stable Diffusion 3 Tutorial with Amazing SwarmUI SD Web UI that Utilizes ComfyUI

Zero to Hero Stable Diffusion 3 Tutorial with Amazing SwarmUI SD Web UI that Utilizes ComfyUI : https://youtu.be/HKX8_F1Er_w

Do not skip any part of this tutorial to master how to use Stable Diffusion 3 (SD3) with the most advanced generative AI open source APP SwarmUI. Automatic1111 SD Web UI or Fooocus are not supporting the #SD3 yet. Therefore, I am starting to make tutorials for SwarmUI as well. #StableSwarmUI is officially developed by the StabilityAI and your mind will be blown after you watch this tutorial and learn its amazing features. StableSwarmUI uses #ComfyUI as the back end thus it has all the good features of ComfyUI and it brings you easy to use features of Automatic1111 #StableDiffusion Web UI with them. I really liked SwarmUI and planning to do more tutorials for it.

🔗 The Public Post (no login or account required) Shown In The Video With The Links ➡️ https://www.patreon.com/posts/stableswarmui-3-106135985

0:00 Introduction to the Stable Diffusion 3 (SD3) and SwarmUI and what is in the tutorial 4:12 Architecture and features of SD3 5:05 What each different model files of Stable Diffusion 3 means 6:26 How to download and install SwarmUI on Windows for SD3 and all other Stable Diffusion models 8:42 What kind of folder path you should use when installing SwarmUI 10:28 If you get installation error how to notice and fix it 11:49 Installation has been completed and now how to start using SwarmUI 12:29 Which settings I change before start using SwarmUI and how to change your theme like dark, white, gray 12:56 How to make SwarmUI save generated images as PNG 13:08 How to find description of each settings and configuration 13:28 How to download SD3 model and start using on Windows 13:38 How to use model downloader utility of SwarmUI 14:17 How to set models folder paths and link your existing models folders in SwarmUI 14:35 Explanation of Root folder path in SwarmUI 14:52 VAE of SD3 do we need to download? 15:25 Generate and model section of the SwarmUI to generate images and how to select your base model 16:02 Setting up parameters and what they do to generate images 17:06 Which sampling method is best for SD3 17:22 Information about SD3 text encoders and their comparison 18:14 First time generating an image with SD3 19:36 How to regenerate same image 20:17 How to see image generation speed and step speed and more information 20:29 Stable Diffusion 3 it per second speed on RTX 3090 TI 20:39 How to see VRAM usage on Windows 10 22:08 And testing and comparing different text encoders for SD3 22:36 How to use FP16 version of T5 XXL text encoder instead of default FP8 version 25:27 The image generation speed when using best config for SD3 26:37 Why VAE of the SD3 is many times better than previous Stable Diffusion models, 4 vs 8 vs 16 vs 32 channels VAE 27:40 How to and where to download best AI upscaler models 29:10 How to use refiner and upscaler models to improve and upscale generated images 29:21 How to restart and start SwarmUI 32:01 The folders where the generated images are saved 32:13 Image history feature of SwarmUI 33:10 Upscaled image comparison 34:01 How to download all upscaler models at once 34:34 Presets feature in depth 36:55 How to generate forever / infinite times

37:13 Non-tiled upscale caused issues 38:36 How to compare tiled vs non-tiled upscale and decide best 39:05 275 SwarmUI presets (cloned from Fooocus) I prepared and the scripts I coded to prepare them and how to import those presets 42:10 Model browser feature 43:25 How to generate TensorRT engine for huge speed up 43:47 How to update SwarmUI 44:27 Prompt syntax and advanced features 45:35 How to use Wildcards (random prompts) feature 46:47 How to see full details / metadata of generated images 47:13 Full guide for extremely powerful grid image generation (like X/Y/Z plot) 47:35 How to put all downloaded upscalers from zip file 51:37 How to see what is happening at the server logs 53:04 How to continue grid generation process after interruption 54:32 How to open grid generation after it has been completed and how to use it 56:13 Example of tiled upscaling seaming problem

1:00:30 Full guide for image history 1:02:22 How to directly delete images and star them 1:03:20 How to use SD 1.5 and SDXL models and LoRAs 1:06:24 Which sampler method is best 1:06:43 How to use image to image 1:08:43 How to use edit image / inpainting 1:10:38 How to use amazing segmentation feature to automatically inpaint any part of images 1:15:55 How to use segmentation on existing images for inpainting and get perfect results with different seeds 1:18:19 More detailed information regarding upscaling and tiling and SD3 1:20:08 Seams perfect explanation and example and how to fix it 1:21:09 How to use queue system 1:21:23 How to use multiple GPUs with adding more backends 1:24:38 Loading model in low VRAM mode 1:25:10 How to fix colors over saturation 1:27:00 Best image generation configuration for SD3 1:27:44 How to apply upscale to your older generated images quickly via preset 1:28:39 Other amazing features of SwarmUI 1:28:49 Clip tokenization and rare token OHWX

[carsthatnevermadeitetc]

Rover 216 Vitesse EFi, 1985. The SD3 200 series was a badge engineered version of the 2nd generation Honda Ballade though quite a lot of development work was done in the UK for the Rover versions which were built in Longbridge. Smaller engined versions used Honda's 1.3 litre EV2 4 cylinder engine but the 1.6 litre was BL's own S-series engine fitted with electronic fuel injection for the "high performance" Vitesse version.

[chacusha]

Trials of Mana protagonists - for the prompt "Upside Down" (part 1)

[etlu-yume]

Magic Rope July 2024

I knew I should've stocked up at the village. Too bad Magic Ropes don't work in the middle of a boss battle.

[ancientwhispersofbabylon]

The bus picks me up in an empty parking lot. And is taking me to the Central Canyon desert hybrid city that is located on the island of my dream. 

Oh! Who goes? It chirps.

Hooray! 

[robotsafari]

yeah the new aai2 localization names are giving me actual psychic damage

[somethingcooltolookat]

[seriously-mike]

So Stable Diffusion 3 is available to download...

*cue soft clonking sounds of the algorithm hitting the glass ceiling*

...and it still fucks up hands like nothing changed.

People are upset that it doesn't improve that much from SDXL, even with just one core model built by StabilityAI, who - you'd expect - have a wee bit more manpower than just some sweaty nerds building data models in their basements.

And then you have chuds crying they can't make porn with it.

There goes the Great Artist Replacement Theory.

Not to even mention it requires four different files that total 10 GB to be uploaded into your video card's VRAM. Great Scott, Marty! Ten jiggabytes! And a completely new prompt with three fields, like two-field prompt for SDXL wasn't enough. People will be playing a guessing game with those three for a long time.

[holmoris]

even as a big proponent of it it absolutely is the mediocracy machine and anyone who doesn't approve of that name is not to be trusted

[glitch-e-stardust]

Barely put that Miyoo down since I got my hands on it tbh, been checking all the games and finding some Absolute Gems

Also it appears to have DOSbox preloaded, so theoretically? I could set dungeon keeper up on there. This would be a very bad not good idea since no mouse or keyboard but it would also be hilarious to see that running on a handheld

Mostly i want to throw a translation patched bahamuts lagoon over there because I mean. Gotta right? Gotta.