Flux Photoshoot - Showgirl

0
0
0
0
Portrait PhotographyGirlPhysical/Facial feature
Baru Diperbarui: Diterbitkan Pertama:
Portrait Photography,Girl,Physical/Facial feature,Checkpoint,FLUX.1Image info
Portrait Photography,Girl,Physical/Facial feature,Checkpoint,FLUX.1Image info
Portrait Photography,Girl,Physical/Facial feature,Checkpoint,FLUX.1Image info
Portrait Photography,Girl,Physical/Facial feature,Checkpoint,FLUX.1Image info
Portrait Photography,Girl,Physical/Facial feature,Checkpoint,FLUX.1Image info
Portrait Photography,Girl,Physical/Facial feature,Checkpoint,FLUX.1Image info
Portrait Photography,Girl,Physical/Facial feature,Checkpoint,FLUX.1Image info
Portrait Photography,Girl,Physical/Facial feature,Checkpoint,FLUX.1Image info
Portrait Photography,Girl,Physical/Facial feature,Checkpoint,FLUX.1Image info
Portrait Photography,Girl,Physical/Facial feature,Checkpoint,FLUX.1Image info
Portrait Photography,Girl,Physical/Facial feature,Checkpoint,FLUX.1Image info
Portrait Photography,Girl,Physical/Facial feature,Checkpoint,FLUX.1Image info
Portrait Photography,Girl,Physical/Facial feature,Checkpoint,FLUX.1Image info
Portrait Photography,Girl,Physical/Facial feature,Checkpoint,FLUX.1Image info
Portrait Photography,Girl,Physical/Facial feature,Checkpoint,FLUX.1Image info
Portrait Photography,Girl,Physical/Facial feature,Checkpoint,FLUX.1Image info
Portrait Photography,Girl,Physical/Facial feature,Checkpoint,FLUX.1Image info
Portrait Photography,Girl,Physical/Facial feature,Checkpoint,FLUX.1Image info
Portrait Photography,Girl,Physical/Facial feature,Checkpoint,FLUX.1Image info

Position reserved, will come back to edit later, and add some gacha machine links...

++++++++++++++++++++++++++++++++++++++

      Recommended sampler: dpmpp_2m

      Scheduler: sgm_uniform

      Steps: 30-35

      Resolution: 1080X1920 or 1920X1080

      500 materials, including Nuomeizi, Feiyue, Yue, Lizhi, **** **nyi, **** **nyao, Yuzijiang, Zhizhi, Yutu, Qili, Jiana, Biju, Kerr, etc., each of them has been edited with Ps, manually modified and labeled, ran 80,000 steps, over 60 hours. However, it still doesn't meet my expectations, the facial features are somewhat blurry, although the style of showing people has been learned well... For now, that's it. It seems that it's impossible to achieve the call of multiple character face types through the training method of LORA. If there's time in the future, I might use existing materials to train some face models with LORA.

Diskusi

Paling Populer
|
Terbaru
Kirim
Segera Hadir
Unduh
(0.00KB)
Deskripsi Fungsi
Tipe
Hitungan Hasil
0
Unduhan
0
Parameter yang Direkomendasikan
Sampler method
CFG
3.5
VAE
Tidak Ada

Galeri

Paling Populer
|
Terbaru