Gakki | Aragaki Yui | 新垣結衣

1.2k
10
Type
LoRA
Downloads
6,820
Last Update
Feb 19, 2023
Versions
2
Base Model
Other
Tags
Trigger Words
gakki

Please use this with ❤️ love

  • V1: dataset includes 21 high res images to train with dreambooth

  • V2: dataset includes 38 fixed and high res images to train with LoRA and dreambooth

    • best layout: portrait and close up shot

    • good layout: lay down

    • 0.6~0.9 is best range when using basil_mix related models

    • small size model : only 151M

    • base models could use with

Lora Model Usage

  1. place gakki-auto-height.safetensors file into stable-diffusion-webui/models/Lora

  2. place vae file into stable-diffusion-webui/models/VAE/

  3. prompt with gakki key word

CKPT Usage

  1. place ckpt file into stable-diffusion-webui/models/Stable-diffusion/

  2. place vae file into stable-diffusion-webui/models/VAE/

  3. prompt with gakki key word

VAE

Highly recommended for use with VAE

the sample images using

https://huggingface.co/stabilityai/sd-vae-ft-mse-original/blob/main/vae-ft-mse-840000-ema-pruned.ckpt

Legal & Risk

It is prohibited to use this model for commercial purposes and any scenarios of illegal acts and purposes.

Tips

prompts:

gakki,

negative prompts:

(((simple background))),monochrome ,lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, lowres, bad anatomy, bad hands, text, error, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, ugly,pregnant,vore,duplicate,morbid,mut ilated,tran nsexual, hermaphrodite,long neck,mutated hands,poorly drawn hands,poorly drawn face,mutation,deformed,blurry,bad anatomy,bad proportions,malformed limbs,extra limbs,cloned face,disfigured,gross proportions, (((missing arms))),(((missing legs))), (((extra arms))),(((extra legs))),pubic hair, plump,bad legs,error legs,username,blurry,bad feet

Next:

Needs more high resolution photo of Gakki

Please leave the photos url in the comments. Thanks!

Versions

Rating
(9)
Downloads
6,560
Uploaded
Feb 16, 2023
Base Model
Other
Steps
3,900
Epoch
1
Trigger Words
gakki
Hash
AutoV2
6D3266B78D
About this version
  • dataset includes 38 fixed and high res images to train with LoRA and dreambooth

    • best layout: portrait and close up shot

    • good layout: lay down

    • 0.6~0.9 is best range when using basil_mix related models

    • small size model : only 151M

    • base models could use with

00021-951204476.png
02499-3705481952-(masterpiece),(best quality_1.0), (ultra highres_1.0), (upper body shot_1.5), detailed,_1 young girl,_ (detailed eye_1.5), glass.png
00024-1384447449.png
00022.jpg
02537-2090126272-(masterpiece),(best quality_1.0), (ultra highres_1.0), (full body shot_1.2), detailed,_(riding bike_1.5),_1 young girl, big smil.png
02532-3940897690-(masterpiece),(best quality_1.0), (ultra highres_1.0), (full body shot_1.2), detailed,_(lay down in bed_1.5),_cozy bed,_1 young.png
02534-3940897690-(masterpiece),(best quality_1.0), (ultra highres_1.0), (full body shot_1.2), detailed,_(lay down in bed_1.5),_cozy bed,_1 young.png
Verified:
PickleTensor
Rating
(2)
Downloads
533
Uploaded
Feb 13, 2023
Base Model
SD 1.5
Steps
2,100
Epoch
1
Trigger Words
gakki
Hash
AutoV2
4FB72BB0E2
About this version

Train using Gakki's recent advertising and event photos from the past few years.

01075-3961320894.png
00936-1404798372-gakki, 1girl, long_hair, solo, upper_body, lips, hair_ornament, brown_hair, bangs, looking_at_viewer, realistic, brown_eyes, sle.png
00944-4027302630-gakki,{best quality}, {{masterpiece}}, {highres}, {an extremely delicate and beautiful}, extremely detailed face, extremely deta.png
01175-2429227575-gakki,{best quality}, {{masterpiece}}, {highres}, {an extremely delicate and beautiful}, extremely detailed face, extremely deta.png

Discussion

Btw, would u like to make an embedding or lora as they save storage and are easier to combine with other models? ^u^

("769" in topic :P)

Hardly ever use 768 model so met some bad outputs at first. Copying your prompts and alternate some words, here are some (most of output not ideal):

How to create full-body image? And is Hires.fix doing a disservice when using 768 model, so we can only create 512*768 at most? T_T

I would like to ask about training myself in Lora. Can you provide advice regarding the following:

1.Which res should I use to train?

2.What variations of poses should I include, such as close-ups, far angles, looking in all directions, or just portraits, considering that I want my generation to be diverse?

3.How many pictures should I aim to include, ranging from 10 to 100?

4.Are there any recommended parameters that I should be aware of?

Gakki is cooking dinner for me after work.What a good wife!

ただいまヾ(〃゚ω゚)ノタダィマ☆

i have tried to train a model in dreambooth for days without any luck, always aliens generated...//cry

what is the source model you trained this based on?

this is just amazing!

i want some advise on training real face!
Can you give me some help?

I wil be so grateful if you contact me.

june@learners.company

Discord : Zag#8373

it is restricted tbh, but i love gaki

weight 1.0 not bad