Dettagli
Raccomandazioni
v1
Atla Fayon - The Rising of the Shield Hero | アトラ - 盾の勇者の成り上がり

Atla Fayon - The Rising of the Shield Hero | アトラ - 盾の勇者の成り上がり

3.7K
88
270
#อนิเมะ
#Character
#shield hero
#เสือสาว
#the rising of the shield hero

LoRA model for Atla from The Rising of the Shield Hero.

Trained using a mixture of anime screencaps, official artworks and fan artworks.

Recommended settings

  • LoRA network weight 0.6-0.9

  • Please generate using highres fix for best effect.

For LoRA Block Weights:

  • For her default appearance, dropping OUTS blocks (1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0) seems good.

  • For doing other outfits, or getting more freedom of poses/composition, using only OUTD blocks (1,0,0,0,0,0,0,0,1,1,1,1,0,0,0,0,0) gives more flexibility, but her age appearance will need to be prompted for

Suggested prompts

Regular Outfit (S3 anime) (Pre-vol 15)

1girl, closed eyes,
white hair, two-tone hair, long hair,
tiger ears, tiger tail,
white dress, multicolored dress, green dress, short sleeves, cross-laced, black skirt,
twin braids, low twin braids, hair ornament,
bracelet,
black footwear,

This LoRA was primarily trained on this outfit.

Known issues:

  • Her hair ornaments and bracelets sometimes double up. Sometimes this can be fixed on highres fix using Euler a.

  • The model will only sometimes put her tail below her skirt, and can often depict it too long. SD has trouble with tail placement in general, but sometimes it can help to delay the drawing of the tail by using [:tiger tail:0.2] or similar. Adding modifier tokens like "tiger tail under" or "tiger tail below" has some marginal benefit.

  • It's relatively common for the number of turns in her braids to increase from model, especially for tall images.

  • Generation of views from side and back of her clothing are less consistent, but can work (I specifically added training data to get it better at this but it's pretty limited still)

  • Her vol 15 outfit is not supported

Visualizza la traduzione

Valutazioni e recensioni

5.0 /5
0 valutazioni

Non ancora ricevute valutazioni o commenti sufficienti

avatar
Machi
31
986
Conversazione con il modello
Annuncio
2023-12-14
Pubblicare modello
2023-12-14
Aggiorna le informazioni del modello
Dettagli modello
Tipo
LORA
Data di pubblicazione
2023-12-14
Modello Base
SD 1.5
Parola Chiave
1girl, closed eyes, white hair, two-tone hair, tiger ears, tiger tail, white dress, multicolored dress, green dress, short sleeves, cross-laced, black skirt, twin braids, low twin braids, hair ornament, bracelet, black footwear
Copia
Introduzione alla versione

First public release. Trained on NAI.

Ambito di Licenza
Model Source: civitai

1. I diritti dei modelli ripubblicati appartengono ai creatori originali.

2. I creatori originali che desiderano reclamare il proprio modello devono contattare il personale di SeaArt AI tramite canali ufficiali. Clicca per rivendicare

Ambito di Licenza di Creazione
Immagini Online
Effettua una Fusione
Consenti Download
Licenza Commerciale
Le immagini generate possono essere vendute o utilizzate per scopi commerciali
Consenti la rivendita del modello o la vendita dopo la fusione
QR Code
Scarica l'app SeaArt
Continua il tuo viaggio di creazione AI su mobile