File: 1771296245461_fe109a26-48(...).png (1.71 MB, 1024x1024)

1.71 MB PNG

It finally did what I wanted but the background is still normal, it looks like a quokka statue
Also yay new thread!

File: 00000058-946431900782398-(...).jpg (1.8 MB, 2048x3072)

1.8 MB JPG

me on the left

File: 00000059-391268894630324-(...).jpg (1.81 MB, 2048x3072)

1.81 MB JPG

>>108167515
looks like a glossy magazine ad

File: 1771298121836_db946c24-bb(...).png (1.23 MB, 768x1152)

1.23 MB PNG

>>108167528
Quokkuoritos
For the cute!

File: nbp-2026-02-17_00002_.jpg (2.74 MB, 1792x2400)

2.74 MB JPG

thanks baker

File: 1771303126894_894302c2-d8(...).png (650 KB, 768x1152)

650 KB PNG

Not beer
but Mead!

i miss schizo anon

File: Z-00120-Flat colors style(...).png (710 KB, 1152x768)

710 KB PNG

Last one from me
good night anons

>>108167467
this is so cool, anon! How'd you get this japanese painting style? is it a LORA?

>>108168900
I'm the anon who made that gen and there is no LoRA involved.
I used "a traditional chinese ink painting of [...]" for the prompt and that is the result

>>108169637
nice, I'll give it a shot, thanks! Which model was that?

>>108170605
Z-Image Turbo

File: 00004-comfyui-cfg=1.0-ste(...).jpg (323 KB, 1200x1800)

323 KB JPG

gm

File: 00029-comfyui-cfg=1.0-ste(...).jpg (258 KB, 1200x1800)

258 KB JPG

happy Chinese New Year to our eastern friends.

File: 00000069-136479451547438-(...).jpg (1.73 MB, 2048x3072)

1.73 MB JPG

>>108170915
is it that time already?

File: 00060-comfyui-cfg=1.0-ste(...).jpg (281 KB, 1200x1800)

281 KB JPG

>>108170952
Yeah, right after the superbowl and after groundhogs day.

File: 00000115-453546258156140-(...).jpg (1.84 MB, 3072x2048)

1.84 MB JPG

klein 9b
base (cfg 4 steps 32) >>108171879
<- distilled (cfg 1 steps 12)

File: Z-000121-Painterly-Steamp(...).png (1.29 MB, 1152x768)

1.29 MB PNG

Morning anons
>>108172058
Very nice

File: 00000001-1030669335028544(...).jpg (2.53 MB, 3072x2048)

2.53 MB JPG

also latest comfy update added NAG support on all flux models (klein, flux2, flux, chroma)
surprisingly nothing broke for me with the update lol

>>108172140
nice flying quok

File: 00000008-993481315947031-(...).jpg (2.56 MB, 2048x3072)

2.56 MB JPG

>gens taking less than a minute
i cant work with this, i need time to shitpost and browse while waiting for a gen

File: 1771349656340_01e70f7d-ac(...).png (1.18 MB, 1024x1024)

1.18 MB PNG

>>108172433
Kek

been awol from local sd genning for past year or two, seems like theres some decent prompt controlled image edit models now? last time I played with it was when flux.1 kontext dropped. any recs for some fun ones to play with? recently upgraded to a 5090 from my old 3090 and figured it would be fun to revisit this stuff

File: 00000015-933479918504678-(...).jpg (1.6 MB, 3072x2048)

1.6 MB JPG

>>108172770
klein for edit stuff
klein-9b should work find on 5090
https://blog.comfy.org/p/flux2-klein-4b-fast-local-image-editing

>>108172786
thanks anon, gonna give it a try

File: Flux2-Klein_00017_.jpg (295 KB, 3240x2160)

295 KB JPG

>>108172799
not bad, a lot better than kontext for sure.

File: Flux2-Klein_00065_mix.jpg (3.17 MB, 2160x2160)

3.17 MB JPG

>>108173067
very fast model too (using distill atm)

>>108173375
gorgeous style anon, are these all from base klein and prompting?

File: 00000029-1029187241185512(...).jpg (1.89 MB, 3072x2048)

1.89 MB JPG

>>108173417
yah it's super fast

>>108173435
yah just klein 9b (distilled) and prompt magic
might help to use sampler gauss_legendre tho, which is super slow but at 12 steps it takes less than 90sec to gen at 1536x1024

>>108173621
try this:
>Lens Effect & Perspective: The use of an extreme fisheye lens distorts perspective, making the building appear curved inward and the foreground subject appear enormous.

File: 00000031-900093476748624-(...).jpg (1.68 MB, 2048x3072)

1.68 MB JPG

>>108173637
the funny thing is that prompt didnt mention fish-eyed lense or a circular image

File: 00000035-1125325233669859(...).jpg (2.68 MB, 3072x2048)

2.68 MB JPG

also i think too much else is going in the prompt style-wise for the lense effect to work. i'll have to clean up most of that out

File: image (37).png (2.39 MB, 1024x1536)

2.39 MB PNG

>>108173729
what are the odds lol

im running into a LOT of 6 or 4 finger gens. Seems like we are going back in time a bit, anyone else run into this on klein? might be the distill.

File: 00000040-1045826441738826(...).jpg (2.03 MB, 2048x3072)

2.03 MB JPG

>>108173948
how many steps? try 10-15 at cfg 1
also depends on what you prompt
"upside down" is still hard for most (if not all) models

File: Flux2-Klein_00155_.png (1.08 MB, 832x1248)

1.08 MB PNG

>>108173958
was only on 4 steps, set to 10 and fixed nearly instantly. thanks

File: Flux2-Klein_00170_.png (1.02 MB, 832x1248)

1.02 MB PNG

what upscaler is solid these days? still ESRGAN/remacri? Tried this one called SeedVR2 but it seems quite badly coded, at least the comfyui node

File: 00000002-1004844574654764(...).jpg (1.91 MB, 2048x3072)

1.91 MB JPG

>>108174091
seedvr2 is top tier, it's all i use nowadays

File: out.jpg (3.9 MB, 4096x3072)

3.9 MB JPG

i dont think comfy's NAG implementation works, or maybe just not on klein 9b distilled
without/with

File: 00090-comfyui-cfg=1.0-ste(...).jpg (211 KB, 1200x1800)

211 KB JPG

>>108174342
You made me cross my eyes. I kinda works but very subtle.

File: 00000003-239853966929290-(...).jpg (2.18 MB, 3072x2048)

2.18 MB JPG

>>108174411
lel
i know at one point someone was making 3d type images like that with a slight variation with noise injection or something on the second image

>>108174456
Gandalf the bot - kek

File: 00000005-586026787241765-(...).jpg (1.37 MB, 2048x3072)

1.37 MB JPG

>>108174466
R2D-chthul-andalf

>>108174475
lovely.

File: 00000007-718333749827586-(...).jpg (1.82 MB, 3072x2048)

1.82 MB JPG

>>108174487
new lora seems more flexible with styles even with photo only sources
likeness is there 80+% of the time, otherwise klein just defaults to "asian"
also likes to make her young looking (even if i specify an age over 24yo)

File: 00000008-327362562719629-(...).jpg (1.69 MB, 3072x2048)

1.69 MB JPG

>>108174517

File: 00000004-633977835865034-(...).jpg (1.96 MB, 3072x2048)

1.96 MB JPG

et tu, snakedog?

https://files.catbox.moe/iibpzw.mp4

File: 00000005-981216154868657-(...).jpg (1.47 MB, 2048x3072)

1.47 MB JPG

>>108174806
this is getting silly

File: 00000008-613146366556811-(...).jpg (3.14 MB, 3072x2048)

3.14 MB JPG

>pointilism
er.. i guess it works

File: 00000013-499544749655170-(...).jpg (3.06 MB, 2048x3072)

3.06 MB JPG

>>108174935
>just got my period
lel nice

File: 00000036-929797941643019-(...).jpg (1.87 MB, 3072x2048)

1.87 MB JPG

fish people everywhere

File: 1771375604879_72327e53-26(...).png (1.33 MB, 768x1152)

1.33 MB PNG

>>108175672
>"He looks just like... Me"
https://youtu.be/3tTHn2tHhcI?si=dcSfDFRJoANHVkw6

>>108175070
Yeah, she's not pregnant yet. Husbando might be sterile.

File: 00000048-1070647480880652(...).jpg (1.64 MB, 3072x2048)

1.64 MB JPG

>>108176145
dungens and drag-ins

>>108175844
Very cute!

i miss schizo anon

i've got amd card and installed comfyui throught that ai bundle they have from their drivers
it works, but after trying to install the comfy manager, it tells me comfyui is outdated and it won't work
do you know if i should just update it normally by git pulling or do they have a special way to do it with those drivers?

Whatever happened to actual stable diffusion?
Like the company,
last I heard was that Emad left and they were going to try something else

>>108179977
Stability AI? They have a new boss and it has been quiet since afaik.
I believe they might be working on some government stuff or at least this is what I would do, working with the spooks and training their surveillance models.

File: 2026-02-18-09-20-58_00001_.png (2.12 MB, 1536x864)

2.12 MB PNG

>>108180375

File: 00000064-1030949108351950(...).jpg (2.03 MB, 3072x2048)

2.03 MB JPG

>>108180594
soul-less

>>108180815
Ya. Z image base has potential because I get the occasional gem, but I have no idea what the magic words are yet.

File: 00000069-224679231191331-(...).jpg (1.82 MB, 3072x2048)

1.82 MB JPG

>>108181053
boomer prompting is the way for all current models

File: Z-00122-1991 old tv VHS A(...).png (1019 KB, 1152x768)

1019 KB PNG

Morning anons
>>108181053
>>108181404
I use this one, it's for Z Turbo but they should work on Z Base as well
https://simple-static-content.s3.ap-southeast-2.amazonaws.com/style_presets.py

File: o_00570_.png (1.37 MB, 1152x896)

1.37 MB PNG

>>108181423
morning

>>108183199
>>108183199
>>108183199