This Lora has been trained on ~750 carefully curated high quality images of glamour style photography with a strong emphasis on aesthetic with elegant setting and professional lighting which is the way I like it :)

While my other Lora araminta-amazing people is based on the concept of high quality, this one goes beyond better faces or skin texture as in encompass beautiful backgrounds (often simple studio backgrounds), professional lighting and gorgeous bodies and the dataset is almost 4x the size of my other Lora.

The end result is a versatile Lora aimed at "improving overall quality" of your images, pushing it more towards realistic uncluttered well lit results.

While it's optimized for photos, it can also be used on illustrations or other non-realistic images, potentially adding an interesting and unique effect to your work (see the examples posted here).

Flux Version

This Lora is intended to be used to bring a bit of realism to default Flux1 faces (which are still too stereotypical most of the time) when used with a strength of 0.1-0.3 or give a more glamour look up to 1.0 strength.

It is also a very useful tool IMHO to "modify just a bit" a Flux image when not everything is OK without completely modifying the image.

This Flux1 version is a very useful tool mostly for realist images but can also be used on illustrations with a sometimes very interesting result and sometimes a meh result :P

SDXL Version

Examples images are a comparison between the SAME image with (right or top) and without (left or bottom) the Lora applied. The strength is 1 if not specified otherwise.

All images use my base model The Araminta Experiment.

The strength can be pushed up to 2.5 or even be negative up to -1.0 for an even more pronounced (and often surprising) effect, especially when applied to images which are not realistic photos.

For normal usage on photographies, a strength between 0.2 and 1.0 is a good choice.

描述:

训练词语: