Finally a Multiple Breast Lora (that is if I had actually uploaded this a few days ago like I had intended. Go check out ChineseRoom's model as well).
This model is designed to give a character 4 breasts to furry/anthro characters rather than just 2, though might be able to preserve any extra breasts in img2img.
This model works best when using either V2 or V1 in parallel with V1.5 at a weight of 0.5-0.75 each. Also don't let the samples fool you, it isn't designed just for dragons and V2 isn't even trained on any, I just like dragons.
Description
V1.5 is trained on r34 images as well and is fairly similar to V1. Thought the art style isn't as overbearing, but it gives only 2 breasts as well as random extra breasts fairly often (samples are cherry pick from successful gens).
Best used at or around a weight of 0.75, or alongside V1 at a lower weight for better results.
FAQ
Comments (15)
works pretty good, Question though, can you make a multi arm lora to go along with this?
Ironically I had specifically tried to avoid multiple arms in the training set because it would just lead to an absolute mess of limbs coming out of everywhere. I can try to make it a separate lora, see if I can get it to look good at all, but that might be a project for someone much more skilled in dreambooth than I.
@lostdragon5 why use dreambooth? kohya is much easier, stable, and faster
@silverace71 Because I have to train on google colab since I don't own a graphics card with more than 4gb of VRAM. Sadly Kohya doesn't work on colab, at least I couldn't get it to work after hours of trying, and when I tried to use their cpu only option on my own computer it just gave me a different error each time.
Long story short, dreambooth is the only one I can actually use at this time.
@LostDragon5 I can help, I got a 3070 and 128GB of normal Ram
@silverace71 Then why not just train it yourself? I'm not necessarily skilled at training nor building datasets or anything.
@LostDragon5 I just got kohya_ss to work on my system so ima do that rn
what model have you used to make these sample photos. With your LORA, theyre incredible...
Your dragons on preview are so beautiful) can you tell me the Lora you used to create them? I want to try to fake the same, but casual ones with one pair of breasts XD
Having some trouble getting this to work. I'm using Automatic111111, and I dropped it in both my \models\Stable-diffusion folder as well as in my \models\Lora folder, both of which didn't do anything even when I tried directly copying the above trigger words.
In Automatic1111 there are 2 different ways to use a Lora.
1) You can put it in \models\Lora and then add it to the prompt like so:<lora:(filename):(multiplier)>
2) You can install the Additional Networks extension (which I personally recommend) by following the instructions on their github page: https://github.com/kohya-ss/sd-webui-additional-networks
where do I get monsterstaber and furrystaber_v4 from, I can't find them anywhere.
Monsterstaber was I custom mix that I no longer have nor remember which model I mixed with furrystaber_v4 to get it (most likely one with the word monster in the name). As for furrystaber, the person who made it took it down and replaced it with BunnerbroMix (https://civitai.com/models/9597), which is essentially a newer version of that model.
is it an additional resource model? cause I cant find it when I look it up
I can't use this model. I removed all my settings and requests, but I can't use this model. .
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.