ZBizz33.safetensors

81 MB · SafeTensor

Mitgliedschaft erforderlich

<izzy-ai-zit>

Izzy is my first LORA that I trained with CivitAI, built from 39 generated images (1024×1024). Izzy is a cute brunette with expressive eyes and a warm smile. I would love to see what you create and let me know if there is tweaking to be done, enjoy! PS: Izzy is a fictional character. Any resemblance to real persons, living or dead, is purely coincidental and unintentional. If you are interested in a LTX2.3 version check out which produces fantastic results! Klein 9b All of the sample images were made using the Big Love Klein checkpoint that I have been very pleased with. Also works well with the flux-2-9b-fp8 diffusion model. QWEN 2512 Trained from a curated dataset from Z-Image Base. I'm very impressed with the quality of generations and character consistency! Z-Image Base Dataset retrained on Z-Image Base, compatable with Base and Turbo. Increase LORA strength as needed on Turbo. Z-Image Turbo Only The dataset was adjusted to fix clarity issues with her eyes and to keep nipples under clothes. Also, Izzy is more expressive than ever before and has better prompt adherence. SDXL Performs much like Z-Image. I use it primarily with Big Love but it also works great with CyberRealisticXL. Big Love Hyper1 I left the SDXL version up but this Big Love Hyper1 trained LORA is head and shoulders above. I definitely recommend trying out Hyper1! WAN Made from the same dataset as SDXL. I used Smooth Mix for all of the T2V which forces large breasts, not that that's a bad thing!

SHA256

1498EEFCAE6A5096D1BB6DC24F9AD98785A7F293029DD2ED1883D1D0BB3233B2

ModelTrends.ai-Modell-ID

#20511

Sampler

Euler a

Schritte

40

Guidance

4.5

Seed

777309345780494

Größe

2048x2048