Software Engineering Courses (SECourses)•3y ago

what about Filewords for images?

CCoqueTornadoMusic what about Filewords for images?

Furkan Gözükara SECourses•1/30/23, 10:06 PM

in TI i didnt use any

Furkan Gözükara SECourses•1/30/23, 10:06 PM

you define what you dont want to teach

Furkan Gözükara SECourses•1/30/23, 10:06 PM

since i have only my face in training

Furkan Gözükara SECourses•1/30/23, 10:06 PM

i dont need

Furkan Gözükara SECourses•1/30/23, 10:06 PM

TI sd 2.1 512

CoqueTornadoMusicOP•1/30/23, 10:06 PM

some people uses this:

"a photo of a man as [name]"

CoqueTornadoMusicOP•1/30/23, 10:07 PM

maybe the solution is to use SD 2.1 512 instead of protogen

CoqueTornadoMusicOP•1/30/23, 10:07 PM

seriously, I have been thinking about it today

CCoqueTornadoMusic some people uses this: "a photo of a man as [name]"

Furkan Gözükara SECourses•1/30/23, 10:07 PM

you really should watch entire video

Furkan Gözükara SECourses•1/30/23, 10:07 PM

that is my most informative video

CoqueTornadoMusicOP•1/30/23, 10:07 PM

yeah, done already twice or three times

CoqueTornadoMusicOP•1/30/23, 10:07 PM

hahah

CoqueTornadoMusicOP•1/30/23, 10:07 PM

is great!

CoqueTornadoMusicOP•1/30/23, 10:08 PM

but I love Protogen

aidiffuser•1/30/23, 10:08 PM

train on 1.5 then merge with protogen = 1.5yourface-1.5

Aaidiffuser train on 1.5 then merge with protogen = 1.5yourface-1.5

CoqueTornadoMusicOP•1/30/23, 10:09 PM

you mean dreambooth? I am trying to avoid it...

aidiffuser•1/30/23, 10:09 PM

CoqueTornadoMusicOP•1/30/23, 10:09 PM

she is a friend, Fátima, at least the 70% of her

CoqueTornadoMusicOP•1/30/23, 10:10 PM

TI is hard to achieve. Also I tried to train in protogen infinity but with batch 3 because of lack of VRAM in google colab as long as it has 8Gb of size... I will try with 3.4 + batch 4... maybe that...

CoqueTornadoMusicOP•1/30/23, 10:10 PM

I have been 10 days testing...

Furkan Gözükara SECourses•1/30/23, 10:10 PM

CoqueTornadoMusicOP•1/30/23, 10:11 PM

yeah but is not protogen hahahah, is it?

Furkan Gözükara SECourses•1/30/23, 10:11 PM

CoqueTornadoMusicOP•1/30/23, 10:11 PM

maybe there is a cool model in SD 2.1 in civit.ai

Furkan Gözükara SECourses•1/30/23, 10:11 PM

this is TI

CoqueTornadoMusicOP•1/30/23, 10:11 PM

using 2.1, right?

CCoqueTornadoMusic using 2.1, right?

Furkan Gözükara SECourses•1/30/23, 10:12 PM

yes

Furkan Gözükara SECourses•1/30/23, 10:12 PM

512

CoqueTornadoMusicOP•1/30/23, 10:12 PM

protogen + my friend Emily

Furkan Gözükara SECourses•1/30/23, 10:12 PM

which style should i generate for this video cover

CoqueTornadoMusicOP•1/30/23, 10:13 PM

I mean, 2.1 maybe is powerful... I should try it... in december I used it... ok... maybe it makes it... but I am still in the protogen mood hahahah...

Furkan Gözükara SECourses•1/30/23, 10:13 PM

protogen really powerful

CoqueTornadoMusicOP•1/30/23, 10:13 PM

(she is Emily at 70% aswell)

Furkan Gözükara SECourses•1/30/23, 10:13 PM

when you inject

CoqueTornadoMusicOP•1/30/23, 10:14 PM

you mean dreambooth?

Furkan Gözükara SECourses•1/30/23, 10:14 PM

yes

CoqueTornadoMusicOP•1/30/23, 10:14 PM

should I drop the idea of make Emily at 98% with textual inversion?