Sign In

h4_DreamXL

Updated: Jan 15, 2025

base model

Download

1 variant available

fp16 SafeTensor

H4_DreamXL_v2_00001_.safetensors

Half precision, best balance • 6.31 GB

Verified:

Type
Checkpoint Trained
Stats

113

Reviews
Published

Jan 15, 2025

Base Model

SDXL 1.0

Training
Steps: 5,000
Epochs: 10
Usage Tips
Clip Skip: 2
Hash
AutoV2
293B9BA0D3
default creator card background decoration
Followers - 44

44

Likes - 580

580

Generations - 3864

3.9K

h4_DreamV2_Images_00001_.png

This is one of my many attempts at a Merged...Merge? Essentially I want to make a model that can generate Anime and Realistic. This is a step in the right Direction. My DreamXL Model has tags or triggers if you will that when used will focus the model on that style. I kept it simple. Anime in the Prompt and Realistic in the Negative and swap for the opposite.

Adding higher and more quality and style tags does help. As you can see from the images, some still come out with the uncanny valley vibe, or shiny rendered skin, but we can still get some solid realistic images. It's in no way perfect yet, but I'm getting there.

Originally trained on a mix of Anime and Real stock images without watermarks.

Then put through the merging process.