Big Love

Please note: The Gwen versions of Big Love are pay-only and special license conditions apply. See description!Gwen3 produces sharper images with better skin texture, better anatomy, more accurate sex scenes and improved txt2img. Editing still produces better quality than txt2img.The 8-bit versions of Gwen3 require 16+ GB VRAM. The 4-bit &amp; GGUF versions need only 12 GB VRAM.New int8 convrot version that runs 2x faster on most Nvdia cards (with Triton installed). It works with the native Load Diffusion Model node, but requires a up-to-date ComfyUI version.Much improved GGUF version with higher quality output than the GGUFs of Gwen2 &amp; Gwen1.To use the nf4 version with newer versions of ComfyUI, you have to launch Comfy with the --disable-dynamic-vram arg in the bat file. Hopefully the BNB node will be updated soon to make this unnecesary.To train a lora download the diffusers .zip file first and unzip it. Select Qwen Image (not Qwen Image Edit 2511) for training (best in OneTrainer as it trains 2-3x faster than AI-Toolkit). Then point it to the unzipped diffusers folder.

bigLove_gwen3.safetensors

bigLove_gwen3.gguf

bigLove_gwen3_trainingData.zip

Gwen3

KreaLife1

KreaEdit1

Klein4

Cat1

Gwen2

Klein3

Gwen1

Eryn1

Ultra5

Photo6

Zuna1

Klein2

Zia1

Hyper1

Klein1

Ultra4

Photo5

Ultra3

Photo4.5

Ultra2

Photo4

Ultra1

Lust1

Photo3

Insta1

Photo2

Photo1

Pony3

XL2.5

Pony2

Pony1

Files with SHA256: 259999e47523827925a52703d2f1ce5f2fefa0d0ee59231d6b332fd3e61d89de