[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / r / s / t / u / v / vg / vm / vmg / vr / vrpg / vst / w / wg] [i / ic] [r9k / s4s / vip / qa] [cm / hm / lgbt / y] [3 / aco / adv / an / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / pw / qst / sci / soc / sp / tg / toy / trv / tv / vp / vt / wsg / wsr / x / xs] [Settings] [Search] [Mobile] [Home]
Board
Settings Mobile Home
/g/ - Technology

Name
Options
Comment
Verification
4chan Pass users can bypass this verification. [Learn More] [Login]
File
  • Please read the Rules and FAQ before posting.
  • You may highlight syntax and preserve whitespace by using [code] tags.

08/21/20New boards added: /vrpg/, /vmg/, /vst/ and /vm/
05/04/17New trial board added: /bant/ - International/Random
10/04/16New board for 4chan Pass users: /vip/ - Very Important Posts
[Hide] [Show All]


[Advertise on 4chan]


File: x_02927_.jpg (475 KB, 1536x2048)
475 KB
475 KB JPG
Previous /sdg/ thread : >>101915164

>Beginner UI local install
EasyDiffusion: https://easydiffusion.github.io
Fooocus: https://github.com/lllyasviel/fooocus
Metastable: https://metastable.studio

>Local install
Automatic1111: https://github.com/automatic1111/stable-diffusion-webui
ComfyUI: https://github.com/comfyanonymous/ComfyUI
Forge: https://github.com/lllyasviel/stable-diffusion-webui-forge
SD.Next: https://github.com/vladmandic/automatic
AMD GPU: https://rentry.org/sdg-link#amd-gpu
Intel GPU: https://rentry.org/sdg-link#intel-gpu

>Use a VAE if your images look washed out
https://rentry.org/sdvae

>Run cloud hosted instance
https://rentry.org/sdg-link#run-cloud-hosted-instance

>Try online without registration
flux-dev: https://huggingface.co/spaces/black-forest-labs/FLUX.1-dev
sd3: https://huggingface.co/spaces/stabilityai/stable-diffusion-3-medium
txt2img: https://www.mage.space
img2img: https://huggingface.co/spaces/huggingface/diffuse-the-rest

>Models, LoRAs & upscaling
https://civitai.com
https://huggingface.co
https://aitracker.art
https://openmodeldb.info

>Black Forest Labs: Flux
https://huggingface.co/black-forest-labs/FLUX.1-schnell
https://comfyanonymous.github.io/ComfyUI_examples/flux

>Index of guides and other tools
https://rentry.org/sdg-link
https://rentry.org/rentrysd

>View and submit GPU performance data
https://vladmandic.github.io/sd-extension-system-info/pages/benchmark.html
https://docs.getgrist.com/3mjouqRSdkBY/sdperformance

>Share image prompt info
4chan removes prompt info from images, share them with the following guide/site...
https://rentry.org/hdgcb
https://catbox.moe

>Discord
6wUwtcJsr2

>Related boards
>>>/h/hdg
>>>/e/edg
>>>/d/ddg
>>>/b/degen
>>>/vt/vtai
>>>/aco/sdg
>>>/trash/sdg
>>
put a plum in my bum and a bottle of rum
>>
>>101915209
fastflux.ai
>>
nick fe moment
>>
>>101917476
Depending on age, she's probably on the d carrousel, hitting the wall, marrying a beta or divorcing him. Less probable, but she could also be a single mom.
>>
Hello. What is "PW"?
>>
File: PW_59802.png (1.32 MB, 864x1152)
1.32 MB
1.32 MB PNG
>>101921519
the village rapist
>>
>>101921519
password
>>
>>
File: pirate cove.webm (2.57 MB, 1536x768)
2.57 MB
2.57 MB WEBM
oh man
>>101921007
That would be awesome. I especially like older artists but most of their work that is trained into the 1.5 and XL models gave overly sepia toned outputs.
I think I will start by looking for decent resolution images of paintings from artists from the Hudson River School and see if a lora with 20-200 images similar to PWs gives promising results.
>>
File: PW_82640_.png (788 KB, 1024x768)
788 KB
788 KB PNG
>>101921548
That seems pretty fair!
>>101921519
Hello anon, I'm PW hahaha
>>
>Hello anon
>I'm <name>fag
>:]
>>
>>101921655
Are you a cyst?
>>
File: PW_82538_.png (818 KB, 1024x768)
818 KB
818 KB PNG
>>101921674
huh? LOL
>>
>>
>>101921692
I may have been given bad information. Sorry.
>>
File: PW_82537_.png (801 KB, 1024x768)
801 KB
801 KB PNG
>>101921711
It's ok hahaha, it happens! :]
Especially here, don't take anyone speaking negatively seriously
Especially if they don't post images haha
>>
now you see why people avoid him haha :]
>>
File: 000000_16342_.png (2.53 MB, 1365x1365)
2.53 MB
2.53 MB PNG
G'mornin Anons,
>>
File: PW_82518_.png (791 KB, 1024x768)
791 KB
791 KB PNG
>>101921803
Good morning, anon! I hope you've slept well :]
>>
File: xformers_flux_dev.png (583 KB, 1536x768)
583 KB
583 KB PNG
>>101920697
I just ran a batch of 10 with dev and schnell with the prompt
>a simple logo with the text "XFORMERS"
and got the right letters 7/10 times with dev and only 3/10 times with schnell.
Most of the mistakes said "XFORFORMERS"
>>101921803
Morning
>>
>>101921961
Can you try flux neger? In my experience its a very good model
>>
File: file.png (160 KB, 1166x487)
160 KB
160 KB PNG
small dino dataset thanks to the natural history museum
maybe the drawings could be used with controlnet to make more images
>>
>>101921961
>XFORFORMERS

ah so dev is quite a bit better than schnell in that regard
>>
>>101921867
Good mornin PW, no, it was horrible and now off to work kek..
>>101921961
Wish I could gen more, l8rs
>>
>>101921984
flux what ? :)
>>
File: PW_82517_.png (765 KB, 1024x768)
765 KB
765 KB PNG
>>101922016
LOL oh no! I hope works goes well hahaha, or quickly at least!
Ughh I gotta work today too haha but it's a short shift!
>>
File: xformers_flux_schnell.png (1.62 MB, 1536x768)
1.62 MB
1.62 MB PNG
>>101922004
it's hard to say because the sample size is small and the seeds are set to increment. A different batch of 10 seeds might favor the other model in other instances.
For example, I just tried it with a more complicated prompt and flux schnell got it 6/10 times and flux dev didn't get it a single time. in fact it didn't even try to make any letters, just scenery and objects but no text.
>>
>>101922065
TY,Same, I'm working only half a day, slow times.. l8rs
>>
File: PW_82506_.png (746 KB, 1024x768)
746 KB
746 KB PNG
>>101922341
Goodbye, anon!! I hope to see you later on! :]
>>
>>101921711
Look at the way it acts like a harmless little girl, anon, this is indeed a cyst.
>>
File: jungle beach 2.webm (3.08 MB, 1536x768)
3.08 MB
3.08 MB WEBM
>>101922016
Hope work goes by quickly for you.
>>
>>101922021
At first i thought its one of those civit schizo merges but to my surprise its actually good and can generate good looking booba
>>
File: file.jpg (323 KB, 1792x1024)
323 KB
323 KB JPG
dino drop
https://huggingface.co/datasets/bigdata-pw/Dinosaurs
>>
>>101922484
based
what are these datasets for anyway
>>
howdy
i broked it again
>>
File: file.jpg (332 KB, 1792x1024)
332 KB
332 KB JPG
>>101922769
lots of stuff. that one has drawings of dinosaurs, i think they could be used with controlnet to generate a bigger synthetic dinosaur dataset, then that could be used for training a dinosaur lora or something
maybe with frames from documentaries like Walking with Dinosaurs, and a lora trained on that, then controlnet to get the same style cgi reconstructions for all the dinosaurs
>>
>>
>>101923048
oh i figured it was for that sort of thing, i meant what are you doing with them? or you're just the collector?
>>
File: file.jpg (357 KB, 1792x1024)
357 KB
357 KB JPG
>>101923146
i've trained a lyrics llm with a (unreleased) dataset of lyrics
ill do other stuff myself too, i'd rather just do the data acquisition but if you want something done you have to do it yourself desu
>>
>>101923266
>but if you want something done you have to do it yourself desu
very true
>>
>>
File: Riojasaurus.png (781 KB, 800x768)
781 KB
781 KB PNG
>>101922484
Thanks for this. They're fun to use as a starting point for Img2img. Would definitely need a controlnet setup for accurate details.
>>101923124
Crazy gen, nice.
>>
File: flux_00011_.png (1.06 MB, 1024x1024)
1.06 MB
1.06 MB PNG
>>101923124
May I ask you for a workflow catbox? I only used basic setup from initial comfy example. Haven't tried Q8 and larger scales images. Also prompt seems to be something interesting.
>>
>>101923548
i'm using forge so no workflow
also a boomer/schizo prompt so good luck
i think the key is
>hyperrealistic detail
somewhere in your prompt
and this line is fun
>it's an impossible snapshot of reality, where photorealism collides with surreal dreamscapes and the line between observer and observed blurs.
>>101923414
>Crazy gen, nice.
riding that edge with highish cfg
>>
File: Chaoyangsaurus.png (491 KB, 800x592)
491 KB
491 KB PNG
the original has one of the weirdest dinosaur face I've seen.
>>101923600
>riding that edge with highish cfg
https://youtu.be/NFeUko-lQHg
>>
>>101921637
what are you using for this anon? the lack of movement besides the few little features gives this an incredibly cool vibe, hard to describe, like a retro PC game menu screen or something
>>
File: file.jpg (217 KB, 1792x1024)
217 KB
217 KB JPG
started on inaturalist. seems like a great source of user contributed animal, bird and plant images. has over 200m observations. there are scientific names and other tags. just need to make a few changes so can distribute it to multiple nodes
>>101923414
yeah the controlnet thing would be a cool project. i'm sure the natural history museum would really appreciate it if they got photorealistic images of all the dinosaurs based on the drawings. i think collaboration with mission-driven organizations like museums could be beneficial to the future of ai
>>
>>101923874
strange days have found us
>>
>>
>schizo thread
>>
>>
>forge fever
ok
>>
h-hi
>>
>>101924328
nice
>>
>>
>>
File: jungle beach 3.webm (2.62 MB, 1920x640)
2.62 MB
2.62 MB WEBM
>>101923947
Thanks, those old games were definitely an inspiration.
For these I've been doing the base gen with flux schnell, and the animation is inpainting with animatediff.
>>101924001
I will give it a try later with a controlnet and some compositing to see if I can get something halfway decent. That's a good idea with the museums and ai outreach.
>>
>>
>>
>AMD GPU

>https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Install-and-Run-on-AMD-GPUs
>https://community.amd.com/t5/ai/updated-how-to-running-optimized-automatic1111-stable-diffusion/ba-p/630252
>https://github.com/lshqqytiger/stable-diffusion-webui-directml
>https://github.com/nod-ai/SHARK
>https://rentry.org/sd-nativeisekaitoo
>https://rentry.org/sdamd
>https://rentry.org/ubuntuamdsd
>https://rentry.org/ayymd-stable-diffustion-v1_4-guide
Newfag here, are these guides up to date for ayymd and rocm?
>>
>>101924750
Tech discussions go in /ldg/, you should ask there

>>101923884
>>
File: taylornpepe3.png (3.02 MB, 1536x1536)
3.02 MB
3.02 MB PNG
>>
File: CUI_iced_00029_.jpg (855 KB, 1440x2560)
855 KB
855 KB JPG
gm anons

>>101924238
bird girl and background monster guys back woo
>>
File: jungle beach sunset.webm (1.89 MB, 1920x640)
1.89 MB
1.89 MB WEBM
>>
>>101925043
Woah
>>
>>101925009
the test is conclusive
you ARE the father!
>>
File: CUI_iced_00030_.jpg (1.1 MB, 1440x2560)
1.1 MB
1.1 MB JPG
>>101925043
awesome aspect ratio, what did you use to animate?
>>
>>
File: ComfyUI_00031_.png (1.3 MB, 1024x1024)
1.3 MB
1.3 MB PNG
Can somebody give me some prompts they want made in flux? i'm out of ideas and keep just making various themed fantasy characters or 70s pinup models
>>
>>
>>101925595
>Alright, buddy, imagine you've got a really big pile of your favorite toys, but sometimes it gets so big and messy that you can't find your favorite ball anymore. The Technical Decrescence Initiative is like a plan to help clean up that big toy pile, so you can always find your favorite ball when you want it.
>But instead of toys, we're talking about all the technology and gadgets that people use every day. Sometimes, there are just too many gadgets, and things get confusing and hard to use. So, this plan is about making things simpler and getting rid of the extra stuff we don't really need, just like how we'd put away the toys we don't play with as much. This way, life stays fun and easy, just like when you have your favorite ball ready to go!

or just use song lyrics
>>
>>101925595
80s music videos
>>
>>101925595
Maybe you can get something better. I'm back to SD1.5 because Flux can't do proper skin.
How much vram you have?
photo-realistic portrait of blonde Aimee Teegarden, no makeup, natural skin, wearing an intricate medieval costume adorned with lace and gold trims. She is standing on a pebble bridge crossing a small river next to an immense castle, she is adjusting her tiara on her head. Her skin texture is natural allowing us to see skin pores and imperfections.
>>
File: jungle beach night 3.webm (1.4 MB, 1920x536)
1.4 MB
1.4 MB WEBM
>>101925067
Glad you like it anon.
>>101925127
Thanks, I'm a fan of panoramas, like your vertical gen is great. I'm animating in comfyui with animatediff v3, by drawing masks and using them to control what areas of the image should be animated and to what degree should they be animated.
That last one and this one have two separate masks, one for the ocean at full strength, and another one for the sky, which is blended in at a lower strength. The rest of the image is not masked and doesn't get animated. The pirate girl with the hair had 3 separate masks.

It's late for me anons, I'll see you later.
>>
File: ComfyUI_00191_.png (1.42 MB, 1024x1024)
1.42 MB
1.42 MB PNG
>>101925690
Is Aimee Teegarden your gf anon? Did she say its okay that I prompt her? If so please look at this pic.

I have no idea what she's supposed to look like really, but here's my first one
>>
File: ComfyUI_00192_.png (1.42 MB, 1024x1024)
1.42 MB
1.42 MB PNG
>>101925825
>>101925690
I haven't tried any prompting of actual people in Flux yet until now desu. I was using reactor to faceswap before in SD.
>>
>>101925825
>>101925857
Seems like its just making a generic face
>>
this is the opposite of what i prompted lel
>>
>>101925825
>Aimee Teegarden your gf anon?
waifu anon, waifu... one of many
I don't need to faceswap with sd1.5
>>
File: tmpoumhjgj5.png (1.48 MB, 960x1440)
1.48 MB
1.48 MB PNG
post movie posters
>>
File: FLUX_00025_.png (1.48 MB, 896x1152)
1.48 MB
1.48 MB PNG
>>101925927
>>
>>
File: ComfyUI_00181_.png (1.89 MB, 1024x1024)
1.89 MB
1.89 MB PNG
>>101925956
>>101925927
>>
File: CUI_iced_00044_.png (2.39 MB, 1440x2560)
2.39 MB
2.39 MB PNG
>>101925127
appreciate the info, only ever used animatediff on a1111, never tried on comfy, gn anon
>>
File: 856769.jpg (505 KB, 3000x2248)
505 KB
505 KB JPG
>>101925857
I fired up SD with reactor off one of her pics and it kinda looks like Egg from arrested development
>>
>>101925690
attached your prompt to mine on flux1-dev-q8
>>
File: grid-0000.png (1.49 MB, 1024x1024)
1.49 MB
1.49 MB PNG
>>101926017
>>
File: FLUX_00056_.png (1.05 MB, 896x1152)
1.05 MB
1.05 MB PNG



[Advertise on 4chan]

Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.