It's is a character made with 20-30 pictures mades with grok ai
SDXL Lora
Train under Ai toolkit with Big_LoveXL Photo4.5 checkpoint
All pictures presented here are made with this checkpoint, video with Wan 2.2
LTX2 Lora
The LTX2 was train with the same dataset included 6 videos of breast massages done with grok, With the LTX2 19b model (not distilled) breast may have some additional physics reactions
Music video done with S2V on ltx2 with AI created song https://www.mureka.ai/song-detail/Bcj9WG7sDrTqW4fBCgsAai?is_from_share=1
With this 2 models You can create A sdxl Pictures and use it with LTX2 on i2V using the ltx2 lora to keep consistency of the character, or in T2V
All LTX2 video are made with WAN2GP , only way for me to generate something with 8GB VRAM without breaking everything on my comfyui library lol
Description
FAQ
Comments (9)
Wow she is HOT! Love the wet pussy!!
You said Grok generated the ultra-spicy images with a wet hairy pussy like that? Can it still generate this level of spiciness or is it too censored now?
What was this actually trained with? You say "video with Wan 2.2" but then say the videos were made by groq. So what is it?
it's was train from grok pictures for the sdxl And ltxv2 lora, i just add some video from grok to complete the dataset
All pictures presented Here are from Biglove4.5 XL checkpoint (under the sdxl lora section)
Video always under sdxl section are from wan 2.2 (no sound easy to recognize)
@jaimybattesti958 Thanks for the additional info. :)
Is the face disfiguration when the character is not zoomed in something typical of LTX2, or is that related to this Lora?
it's seems more related to the low res and fps, if you can do 1080p and 50 fps the result will be better, ic detailer lora can help too, my local machine is a bit weak to run full HD
It's run better too if the model is next to the camera, face may be focused for lipsync
@JaimyB got it, will look into it then, thanks for the reply.