This is a merge of bigASP and LUSTIFY, which, in my opinion, are the best non-Pony SDXL checkpoints for NSFW images. This merge produces results similar to bigASP but with improved image quality and aesthetics.
Links to checkpoints merged:
Description
This is almost the same merge but with lustify 4.0 merged instead of 1.0
Darker images
May or may not be better
FAQ
Comments (36)
Which version of LUSTIFY is merged in v1.5 of this model? Thanks!!
About this version
This is almost the same merge but with lustify 4.0 merged instead of 1.0
Darker images
May or may not be better.
this is written in the description
I tried using this in ForgeUI and all the outputs were visible in preview but turned completely black when complete. First checkpoint this has ever happened on.
Might need VAE enabled/disabled/changed
Incredible checkpoint. Works with a variety of LoRAs too (character and concept alike). Good work!
Will test this on my LoRA's soon and make some comparisons with V1. Currently I only have access to a slow outdated laptop for a few days.
Awesome checkpoint, a combination with Natvis would be crazy
This is a very good checkpoint, Anatomy is insanely good, skin textures are amazing.
Adherence is good as well, i found that using DMP++2m SDE with exponential yields the best results (at least for me).
I threw all types of loras and the model handled it well.
Good job and thanks.
What are the prompts to get it to do double penetration
Thank you! Great model, apart from eyes being a bit weird, but it's easy to fix in inpaint with lora.
4 step lightning version please
Comparison of v1.0 & v1.5 with my LoRA.
is this me or there's a lack of lgbtq content in the training dataset ?
Incredible model, thank you!
Why are people using pony score tags?
Because they are too lazy to delete them after grabbing a prompt. Just like they keep SD1.5 loras when generating SDXL images.
Because bigASP was trained on the same mechanic and it is merged into this checkpoint.
@Romlo聽But you don't need them. I don't use them with my bigLust gens.
@SubtleShader聽you can generate without them, i was just answering a question :)
@Romlo聽Was just indicating possibilities 馃榿. I like your images!
I love this model and I am super hyped about 1.5, thank you! Any suggestions for FaceDetailer in comfyUI? I am having trouble with the faces, the features almost always turn out blurrier after the fix pass
@waterdrinker It looks like the BigASP guy is in the process of training a new BigASP with 6 million images! I've found Big Lust to keep all of BigASP's advantages while making it more flexible. Is there any chance you will grace us with a Big Lust mixed with the next BigASP?
BigaspV2 is out and incredible. A new merge would be amazing!
Just merge it yourself, BigLust 1.0 is literally just 60% BigASP + 40% Lustify V1 with the Textencoder from BigASP
Can you do a merge with BigASP V2, Anteros, and Lustify? Each has their own strengths, and you are a proven master of the merge.
Would love to see Lustify v2 (outperforms v4) + BigASP v2.
i was training lora on this model but i cant seem to get the setting right, any advice? models trained from sdxl base don't seem to work that well, the faces are not accurate at all. Love the model btw.
train on lustify
@Romlo聽Got it, will try it out. I did train a model on BL yesterday, came pretty okay. But it was a style lora so hard to say.
@Ceylon_Ai聽just a tip cause it worked for me, if you train on lustify a char lora, make it 30 to 40 pic and remove the background, and use 1 tagging word for the dataset. worked surprisingly well.
@Romlo聽Thank You!! will try it out on the next one! also any chance you can share a json file with me.
this is the model btw, https://civitai.com/models/914740/slnude-sri-lankan-nude-v1-big-lust?modelVersionId=1023798
@Romlo聽Sorry I am pretty new to this and experimenting. What do you mean by removing background? Do you just paint over it with black paint in photoshop? ^_^
Also I trained one char lora before so far without doing this and it seems to work so far. What is the benefit of "removing background"? Thanks in advance for any response.
@wynnettetodd364聽you dont have to do it all the times, i only did it while training on lustify, lets say i was experimenting with multiple tries for character lora on lustify and got the best results when i removed the background. i used https://www.photoroom.com/tools/background-remover to remove the backgrounds from the images, it is completely free.
@Ceylon_Ai training was done on colab so no json file sadly. but i am glad to help with any information you want, just dm me and ask away :)
Don't forget to add a 'realistic' tag (not just photorealistic) so this awesome Big Lust model reaches even more people!