V3 has been released.

I have fundamentally reviewed the training method and tried to reduce the file size.

After 10 tests, I have finally made it usable for release.

Our focus has been on reducing file size, so if you are concerned about a decrease in quality, please use an older version.

I divided the concept of nosehook and had it learn.

1.nosehook

2.nostrilspullup

3.long nostrils

By dividing these, the reaction to the base model has improved.

If you exclude the trigger word [nosehook], you can see the nostrils without the hook.

Since I have not trained as intensively as before, the compatibility with other people's Lora may have improved.

I have not confirmed it myself, so please try it.

The current problem is that it is weak against the side of the face.

Please use Lora with less strength.

If you make it too strong, the shape of the nose and HOOK tends to collapse. You will understand if you try it for yourself.

The generated images occasionally gave me nosebleeds, but I couldn't stop them.Take control with negative prompts.

There is a discrepancy between the file name and the published version as usual, but please do not mind it.

V2 has been released.

Some changes have been made to the training file and the depiction of shading.

There was a problem with the previous trigger word, so it has been changed to [nosepullup]. It will still work if you leave it as [nosehook] as before.

The reason for changing the trigger word is that the IL base model seems to have nose hook trained in an incomplete state, which is hindering my Lora. Also, it sometimes causes nosebleeds.

There may be a discrepancy between the published version and the file version, but please do not worry as this is due to differences in file management.

The 3way version has been released. The trigger word is [3waynose].

It supports depictions from the front, diagonally, and side.

The accuracy decreases as the face gets farther away.

There are still some limitations to the depiction.

It comes with extra decorations, so cancel them out with negative prompts.

I made an Illustrious-XL version in response to requests.

Accuracy decreases as the subject becomes farther away.

If your face collapses in long distance shots, try using ADetailer. It often fixes it well.

I am uploading some success stories to my gallery.

The trigger word is [nosehook].

I found a prompt to change the color of the band.

You can change the color by entering [nosehook XXXXX band]. Enter any color for XXXXX. Just enter the color specification along with the trigger word.

It doesn't always work, but it's worth a try.

If you leave out the trigger word and use [nose nostrils], the chances of drawing without a hook will increase significantly. We recommend setting the strength to 1.

The training content is the same as the PONY version. Training images have been adjusted.

It mainly supports faces from the front to the side, but is weaker on the side.

Please be careful when generating copyrighted characters.

Please use in accordance with the copyright holder's guidelines.

Please do not worry about the version difference between the download file and the title here.

Search AI Models & Addons

very few results