This is a very old model of mine and does not conform to my new very high-quality standard established with my "Essenz" model series, so please do not use this! It will be replaced in the near future with a new and better model in the "Essenz" series!

See here for an extensive article about my entire training workflow and configuration if you want to train your own SDXL 1.0 LoRa's and achieve results similar as me!

Recommended weight: 1.0 - or lower when you combine it with other LoRa's

Recommended model and VAE: SDXL 1.0 with 0.9 vae

Recommended settings: 35 steps | DPM++ 2M Karras | CFG 7

Recommended negative prompt: text, watermark, signature, abstract, burned, instagram, low-quality, low-res, action figure, child drawing, jpeg, low contrast, glossy, puppet, cliché, stock photo, magazine cover

Description

A highly experimental LoRa for SDXL 1.0 for mid-transformation/morphing images of people transforming/morphing into animals.

Weights used for some of the sample images:

• Ghibli style sample image: Ghibli LoRa 0.9 | Morphing LoRa 0.7

• Semi-realistic concept digital art style sample image: Semi-realistic concept digital art LoRa 1.0 | Morphing LoRa 0.8

• Inkpunk style sample image: Inkpunk LoRa 0.9 | Morphing LoRa 0.6

Sample images are cherrypicked.

Trigger words

• "x morphing into y"

Known issues

1. Almost solely works with people morphing into animals, not into objects or people etc

2. Does not work well with all animals equally, some animals work much better than others (e.g. ducks do not work well at all - they seem to be highly overtrained in base SD - and neither do dragons since basic dragons are already too complicated for base SD to draw correctly half the time)

3. Does not work well with all styles equally, some styles work much better than others - digital artworks, realistic paintings and cartoon style images seem to work best, photos work worst

4. Some/many generations may fail to look good and/or coherent. Its just the nature of this concept. SDXL already has a hard time as is to portray normal people anatomically correct with the correct amount of limbs.

Changelog

• Initial release

Additional Info

Credit is not mandatory but I would still appreciate credit when posting images using these LoRa's on social media! Feel free to @ my Twitter and/or Instagram accounts too (@ai_characters)! I would love to know when it is used!

I share all of my model's and training knowledge for free, but donations to my Ko-Fi would be greatly appreciated! Testing so much and training so many model's costs a lot of money (and time).

Feel free to also try out my RVC2 AI voice models which you can find here!

Stats (v1.0)

Dataset:

• 58x artworks

Epoch chosen for the final model:

• 25

Learning rate:

• 8e-5

For the configuration used to train this model see my guide which I linked in the beginning of this post.