O wow. Can we set this parameter ?
O wow. Can we set this parameter ?
Fnot yet
Fnot sure :d
Fthe prompt has dramatic effect on sd 2.1
Cthat is the thing... dreambooth make magic
HThese S/R tests are important then
BWhat's the best way to go about getting a similar style to clipart images, I've collected a small set and tried one DB model but I think I overtrained it quite a bit. Was a few weeks ago. Was going to give it another shot. DB again, maybe an embedding? Here's my source
Cnah, been trying with Fátima and Émily with S/R tests and not luck... also prompting can help... it is quite a hard work... but not really worked
Cmaybe using inpainting xD...
HYeah. That one or dreamlike photoreal was my favorite in 1.5
Cprobably the TI thing we are talking about it was made for things like this...?
Cor even better, just try the newest extension custom-diffusion-webui and let us know the results...
HYeah probly 2.1 TI style training
HYou used the same process as TI video? Or new findings
BSorry I missed that discussion, I'll try and find it above. I haven't messed with TI stuff yet so I guess down that rabbit hole I go
Cgood question
Fsame
Fi am working on prompts now
Ffor style TI is said to be better
Fyou tried?
Cthe TI thing was made for styles they said in the papers. So.. But it is possible to achieve the 90% of similarity in humans
BI haven't tried anything with TI yet, been using DB for everything. I guess today's the day though :e_kekw:
BYouTubeSoftware Engineering Courses - SE Courses
In this video, I am explaining almost every aspect of Stable Diffusion Textual Inversion (TI) / Text Embeddings. I am demonstrating a live example of how to train a person face with all of the best settings including technical details.
If I have been of assistance to you and you would like to show your support for my work, please consider bec...
If I have been of assistance to you and you would like to show your support for my work, please consider bec...
BThis still the way to go about it?
Fthis is interesting
F
F
Cnegative prompts
Fhow 75 goes to 150
Fweird
Cpart of a token?
C
Cmayb a bug
Fby the way
Fnegative prompt
Fseriously removes your face
F:d
Ffrom results
Cyep
Cin TI I used to avoid too many negative prompts
C(if that is the context you are talking about now heheh)
Fyep
Cyes, prompts like "ugly" can make the face another one
Cmaybe it is because TI is about tokens, prompts
Cso one can change the whole thing
CI dunno really
FTI affected more than dreambooth that is for sure
Fi am gonna test just like this now
Cok
Cit is quite delicated the thing of TI
Ci have found that if you train it in the vanilla, it works better in the vanilla. But not too much in protogen. And if you train it in protogen, it works better in protogen. But not really amazingly. Just an 60%/70% ... it is quite hard... I need like 10 or 20 more tests... trial and error. Maybe you, the community can help