LoRA of Gina Carano
If you wonder why this LoRA was uploaded please check my new article and be amazed: https://civarchive.com/articles/1721
I'm creating Dreambooths, LyCORIS, and LORAs. Here is my base model Serenity: https://civarchive.com/models/110426
I upload 2-3 times a week, however, I write down suggestions all the time. If you want to see someone made into a model just leave a comment :)
Suggestions left on my donation page ( https://www.buymeacoffee.com/malcolmrey ) are taken with the highest priority and you will be able to enjoy them way before they get released here. (I also accept custom requests)
The reason is that I can't keep up with uploading and the models are being made pretty much all the time :-)
As mentioned, via the donation page I accept custom requests, so feel free to engage with me there, I will answer all your questions :)
Description
FAQ
Comments (19)
I think this one is a little off. I'm assuming this is aiming for her younger fighter days (you can tell from the eyes)? But she's just missing some size and bulk to her head - chin and jaw in particular.
in general she was very problematic to train, this was not my first attempt
as for the chin/jaw - would you mind preparing some image album of her? i could try rerunning her again :)
@malcolmrey Yeah, I know - she's tricky as hell to get right! I think you have to pick a specific era and stick with that (young fighter - movies from Haywire to Deadpool - Star Wars and beyond). I can try to put some images together for one of them :)
@joakimkunzdesign128 that would be perfect!! :)
i've seen haywire, shame that they dubbed her voice but overall cool movie :)
@malcolmrey Sorry I've been slow to respond. I finally got my s¤#% together and scoured the interwebs for some suitable images. I set the search to the period 2012-2016, to limit the range of her look. What types of images do you need for this (head, full body, etc?), and how many? Or should I just wrap up the whole lot and make a google drive link? I get the sense that there's no direct messaging here, so I'd probably delete the link once you have the images. Is a .rar file okay? And do you prefer .png or .jpeg?
@joakimkunzdesign128 pngs are fine, rar is fine, if you put more images - i will be able to pick better images for the datasets
both are fine - head and full body, i usually go with 70% of headshots and rest are half body shots (i use full body shots if the costume/clothing is important)
yeah., you can drop a link here and tag me and i will see it :)
@malcolmrey 128 is more than I have now - I'll have to keep looking around for a while. Let you know when it's put together :)
@joakimkunzdesign128 perfect, thnx :)
@malcolmrey How come you use so many images? I usually see people use between 20 and 30.
@joakimkunzdesign128 what do you mean? :) I use around 25 (i do download way more because some get removed due to not meeting standards, then i remove some after cropping, and then i select the best 22-25 from those that remain)
@malcolmrey Ha ha ha! Oh no! I didn't realize my own name had a number tagged at the back of it! XD :D I thought you wanted 128 images! XD I have enough for you, then - I'll pack them and send them when I come home from work today :P
@joakimkunzdesign128 yeah, no worries about the number :-)
however the link is not public, i clicked to request access however :)
@malcolmrey Goddamn - not making this easy, am I?! XD I've just shared it. :P
@joakimkunzdesign128 now it works, i'm downloading it, thanks! :-)
@malcolmrey Don't know if you've had a chance to try it yet, but I baked a checkpoint on a different site, and it seemed to work out fairly well - although some of her very light make-up tends to shine through in an annoying way.
@joakimkunzdesign128 hey Joakim! I have only downloaded the samples and looked at them (the samples seem very good) but I had no opportunity to handle them yet.
I will do that for the upcoming week, so we can compare :)
Mind sharing some of your outputs from your baked checkpoint? :)
And which site did you use?
Cheers!
@malcolmrey I made via GetImg.Ai - I don't know s¤%# about git or python, so making them on my own machine feels like flying blind, and it takes up all the resources on my computer for a long time.
I can upload the checkpoint to Drive, and you can take a look at it yourself. Maybe you can also help me make sense of the other files that come along with it. *Edit: Here's the checkpoint link. https://drive.google.com/file/d/1AdMB_HT4aHFH8boAvBgarx1W3uuJg8yf/view?usp=sharing
prompt is Gtmg Woman.
I uploaded an image here on the Gina model, but I'm not sure the mods have cleared it yet. So here's a link instead: https://ibb.co/C6nNb0D
@joakimkunzdesign128 the image looks really nice, but I'm fairy certain mods will not approve it since bikini clothing is not allowed (i mean, it is too revealing :P).
I've briefly checked your model and it is quite nice. If you tell me which RV version did you use as a base I can extract the lycoris for you :)
Btw, I've checked the page you linked, it has nice design. Seems like training a dreambooth costs you roughly 8.5 dollars or so. And they have a weird limits (2 dreambooths at a time in the basic plan?).
How do you like their service in general?
Cheers :)
@malcolmrey It's just the standard RV3. I don't think I need a Lyroris or Lora of this one right now - I'm gonna try my hand at ripping it myself with Kohya and see if I can't figure it out :) . But I was thinking you could try and make your own, and then compare the two. And you can use the one I send to predict errors and which images to omit.
I used that site when I first got into SD, but after getting it set up on my own machine, I haven't used it much - until now. They have a very simple setup for generating models, and it's fairly quick sometimes, but obviously it gets quite costly if you screw up a dataset, and have to try again. Fx, I tried setting up a 70's Pam Grier model, and that did NOT work, so it's back to the drawing board! XD :P
I talked to one of the mods on that other site, and he said they limited the models because they have to store them as well - and with thousands of users creating 7,6 gb models, it takes up a lot of space for them.