Home Creators Posts Import Register Favorites Logout

Content

Hey everyone! I'm still Alive! 🙌 Sorry for being really quiet these last few weeks, been taking some time off BUT ALSO WORKING HARD ON MY NEXT VIDEO which is....drumroll, please.... LORA STYLE TRAINING!  That's right, and I gotta tell you it's really really hard work but also super fun! I've been testing and testing and testing again to bring you the BEST style training video possible. The video is still not done yet, it will come out soon but you can see some small samples of some style training I did recently down below, with the showcase of the walking dead video game style + walkers & spider-verse style featuring nicolas cage and margot robbie, it's really fun to generate those images and I can't wait to teach you to do that yourself.🤪  And if you are interested in those future models AND MUCH MUCH MORE, they will soon be available right here for you, my dear patreon supporters!   Meanwhile, wait patiently, the video is coming soon, I promise! 😘

Files

Comments

patchzy

Will this also work with creating envinroments in that style? and is it maybe possible to show how to take an existing image but style transfer using the created lora? that would be wonderfull thanks <3

Aitrepreneur

yes it's a style, as long you have some training images featuring environment you can 'reverse-engineer' it for a different landscape ;)

Hugo Acuna

Fantastic! Can you do objects after this? Regardless of what people vote, there are fantastic use cases for each one!

Khaz

Good stuff :)

nellz

Can you make a video about training embeddings for SDXL?

Aitrepreneur

LORA already works like an embedding, just better, so there is no need for embeddings for SDXL

nellz

not for my purposes - embeddings are always better than lora for me, but i figured it out SDXL TI training is working on kohya now

MtdXRavage

man, your videos are so good it's the first time I decided to become a patron. Keep up the good work!

Aitrepreneur

Thanks really appreciate it, more coming soon ;)

MtdXRavage

have you thought of making a video about swarmui? Just realised I have a too weak graphic card to run sdxl 1.0 locally, even with xformers and --medvram. Trying to install swarmui with Google Collab now

ProudChinaLover

can you make videos for Dreambooth SDXL? I wanna train it locally on my computer

Aitrepreneur

Yes I might do a video about dreambooth training, altough you will probably need a 3090 or 4090 min to train sdxl with dreambooth

Flonix

what about training a lora and mixing it into sdxl? same thing right?

Aitrepreneur

not exactly, dreambooth should be more powerful than lora but it uses more power. Altough I still don't know if dreambooth is really needed with sdxl over lora

ProudChinaLover

I do have a 4090 and an i9. Many of your subscribers are hardcore users. Please make one and guide us

HOOD GAMER

I'm waiting for the new video, it seems the other way has an issue with using the replaced person. I have to run it several times to get warpfusion to recognize the character. i wish i could only have the person i am training and not a replacement because the person I'm using to replace always shows up

Aitrepreneur

not sure how warpfusion works exactly so I can't really help there but if the training was done correctly you shouldn't ever see the original person in the generation

EricRollei

Did you make this training video yet? I'm wanting to train lora on my own photo artwork

Aitrepreneur

yes that video came out a few weeks ago already ;)

indievish

Is this released?