ah I just thought flux would know what emotions facial expressions and impliment them onto your face
ah I just thought flux would know what emotions facial expressions and impliment them onto your face
![Dazzastrous [4090]](https://cdn.discordapp.com/avatars/871389071686135838/c4a5f4d81d9eeb18d856295d692e2c07.webp?size=16)
Fstill can do but include to get perfect
![Dazzastrous [4090]](https://cdn.discordapp.com/avatars/871389071686135838/c4a5f4d81d9eeb18d856295d692e2c07.webp?size=40)
Dunderstood thanks Doc
Dapologies to your wife for keeping you hard at work she must be very understanding lol
Fi work from at home so it is fine :d
Fwe really should include all kind of poses 
F
F
Dsuch a fungi
Dquite the adventurer Doc, one min your mountaineering the next swimming with sharks hehe
Dcan you do a clay model version ?
Daction figure?
Fit fails at non realistic stuff atm
Freally overfits model
Ffor realism
Fhaha this one good
F
Fgenerating examples to show at the intro part of video
IThese are amazing
ICan we train in kaggle
Fnope
Fbf16 card mandatory
Fbut you can use in kaggle after training
JFlux is fantastic
Fyep
Fhaha this one cool
F
S@Dr. Furkan Gözükara Do you also have side by side comparisons for unet only + now that we have clip_l training aswell
Fyes i have
Fbut i didnt perfected clip l yet
Flet me update patreon post for early experiment
SSounds good
FPatreon
Get more from SECourses: Tutorials, Guides, Resources, Training, FLUX, MidJourney, Voice Clone, TTS, ChatGPT, GPT, LLM, Scripts by Furkan Gözü on Patreon
SAwesome will have a look
DJust out of interest Doc when generating are the majority of the images holding the same likeness in quality? do u get many images do you get that have some deformity or have no fidelity?
Falmost all perfect
Fwith fux
F
Fi am mass testing prompts atm not like generating a lot of same prompt
Dhehe mars
Fbut looks like we really should add different angles
Flike looking down
Fup
Fleft right
Ffor such prompts
Fit works really good
Dsmiling , angry, confused
UHey! Do you know if there is a way to specify image color palette for Stable Diffusion or Flux, without using Controlnet extensions?
Finclude in train dataset and should work perfect
Fnot that i know
Fi am testing torch preview nightly
