yes is it is "photo of man" in class prompt, all black random guys.
yes is it is "photo of man" in class prompt, all black random guys.
Fyour class photos like that?
Fthat is unsual
Fi have publicly shared dataset
F
W
Wnice thank you
Finteresting
Wmay be I set something wrong. I will check
WCan I delete a model created?
Fsure
WJust deleting the C:\stable-diffusion-webui-master\models\dreambooth\model folder?
Fyes but if you have checkpoints you also need to delete them
Fto save space
Wok thanks
SUmm, site is working, but with page errors and big time spend, did you have, like, boosty or smth?
Smaybe, i can transfer this money to you by some another way, what is not blocking in Russia
Fyes i can also receive bitcoin or bnb crypto
WAny advise to avoid overtraining? I only got 2 or 3 samples correctly, but the 3 o 4 is over-training, this means, the Tomer Hanuka sample is not Tomer Hanuka stile anymore stile anymore. The problem happen before the chekpoint is saved, so I'm not able to get not over-training. Checkpoint, and it is very difficult to make stiles with the first checkpoint.
Fhow many per epoch do you save?
Fyou can reduce number of epochs
Fthen use x/y/z to compare
Fpost processing style teaching video
WI'm using images, no video yet
Fno i mean i am post processing new tutorial video
Fi recorded it
Wha ok, thanks. I will check the epochs
WTraining steps per image 150, save model frequency 25, save preview frequency 5
F25 could be big
Fi saw 5 epoch made overtraining in my recent tutorial
Fstill processing it not available yet on youtube
Whaha, sure, I don't know where I took that number haha
Dquick question, if I'd like to maintain a look of photo I trained I'm getting good results, for example it's a certain type of "ohwx City", should I train another model on top of a previous checkpoint or should I return to stable diffusion 1.5? For example if I'd like to train a certain type of "elos Bus" to show up in the photos of the "ohwx City" should I continue from the ohwx city checkpoint or train it on the s.d. 1.5 checkpoint?
Fit depends on purpose
Ffor example in upcoming video
Fi trained myself then trained style on top of that
WOk, you create two different models right?
Fit works like this
F1st train model and you get model check point
Fthen create a new model from that checkpoint
Fand train it
Wmmmm interesting
Fwait new video explains a lot
HVery cool 
Fperfect hands but misisng head :/ genereating some images to put in video while talking
Fthe weakest part is hands
Fdecent one can fix face easily
FFor two weeks I have tried all the variables that I found in tutorials for dreambooth without good results.
And I don't see many realistic Lora in civitai either, the only one I found that has some good ones is this https://civitai.com/user/dogu_cat
I don't know what I'm missing but it seems to be quite complicated to achieve reliable results
Your examples are all great but I can't replicate that similarity level
And I don't see many realistic Lora in civitai either, the only one I found that has some good ones is this https://civitai.com/user/dogu_cat
I don't know what I'm missing but it seems to be quite complicated to achieve reliable results
Your examples are all great but I can't replicate that similarity level
Average Rating: 4.9 (149), Models Uploaded: 31, Followers: 453, Total Likes Received: 4.9k, Total Downloads Received: 41k.
Fwhat you want to achieve?
Freplicate a face in some realistic way
I always end up with results that look like the person but not exactly
I always end up with results that look like the person but not exactly
FI caption the image set, and I have tried using regularization images too