Sign In

ForgottenMix - Animation - 2D

1.3k
6.7k
199
Verified:
SafeTensor
Type
Checkpoint Merge
Stats
5,882
Reviews
Published
Jun 22, 2023
Base Model
SD 1.5
Hash
AutoV2
90C70627D3

To begin with, English is not my mother tongue, so I apologize in advance if there is any mistake.

I'm in love with this model but I think it struggle sometimes with hands and feet.

All the images I posted have been inpainted using adetailer extension.

Personnal configurations /recommendations (Feel free to experiment):

Image Size: 512x512 / 512x768 (768x512 can be unstable).

Sampling method: DPM++ 2M SDE Karras / DPM++ 2M Karras / Euler a

Sampling steps: 30

Upscaler: 4x-AnimeSharp ScuNET PSNR

Hires steps: 20

Denoising strengh: 0.51

Clip skip: 1 / 2 (I think 1 is better to stick to your prompt)

VAE: vae-ft-mse-840000-ema-pruned / clearvae_main

Positive prompt:

(detailed, perfect, beautiful, exotic, special, unique, charming, hypnotic, breathtaking, skindentation, masterpiece, high quality, high resolution, extremely detailed, sharpness, intricate, high quality_textures, high resolution_textures, absurdres, sharp details, award_winning, bokeh, rule_of_thirds, moody lighting, detailed shadows, light diffusion, white_balance, Realism, RAW_Image, realistic, photorealistic environment),

Warning: It can produce beautiful pictures but can ignore some of your prompt sometimes.

Negative prompt: (worst quality, low quality, normal quality), greyscale (if you want vivid colors)

You can try to play with Clip skip and "(worst quality, low quality, normal quality), strengh. It will alterate the output for a more or less anime style.

Please, don't use restore face.

Please, experiment and show me what my checkpoint is capable of !