modelli/GoodFit - v2.0

GoodFit - v2.0

shadowrui

10/4/2025

9:32:51 AM

| Discussion|

Parole Chiave e Tag Correlati

A blonde girl wearing a shiny black leather jacket stands on a rainy night street illuminated by street lights with wet reflections on the pavement.

Portrait of a girl with long black hair wearing a white shirt, black tie, and black suit skirt against a simple dark background with front lighting.

A woman with long black hair dressed in a black suit, white shirt, and black tie, standing on a forest path with tall trees and soft front lighting.

Prompt Consigliati

8k, highres, best quality, 1girl

Prompt Negativi Consigliati

sketches, (worst quality:2), (low quality:2), (normal quality:2), (lowres:2), bad_hand, broken_hand, extra_hand, extra_fingers, blurry, blurry background, blurry foreground, (panties), (bra), bikini, (underwear), underpants, (sex toy), (glans), ((text)), ((watermark)), (pubic hair), bag, handbag, holding, hold, hand bag, nsfw

Parametri Consigliati

samplers

Dpm++ 2m Karras, Dpm++ SDE Karras, DPM++ 2M Karras

steps

cfg

resolution

512x768

vae

vae-ft-mse-840000-ema-pruned.ckpt (c6a580b13a)

Parametri Consigliati per Alta Risoluzione

upscaler

ESRGAN_4x, 4x_RealisticRescaler_100000_G, realistic4x_RealisticRescaler_100000_G

upscale

steps

denoising strength

0.4

Suggerimenti

Don't use too many prompts about image quality or realism as they may harm results.

Less is better for prompting.

Using ADetailer can give better facial details and is highly recommended.

Punti Salienti della Versione

Adding more lighting, Better fitting to my Lora, eaiser to change the background and pose when using Lora.

Sponsor del Creatore

Credit goes to:

The point of this checkpoint is to get a high-resolution image with a realistic 3d face.

The checkpoint can give a clean, high-res image without huge tagging.

expecting to achieve 3d or animated characters in real-world lighting. Especially for the Lora I made, which mostly are 3d movies or game CG characters. (helps to recreate what they should be looking for in CG or movies.)

Adding high contrast for betting lighting, at night, or when it rains.

Hoping to get betting lighting overall for the 3d or animated characters.

v2------------

Adding more lighting, and higher contrast than v1.

Less blurry background, and a much higher chance to get a highly detailed background (with my Lora).

V2 is easier to move around the character Lora than v1, but sometimes time may cost less likeness, So might need a higher weight.

V1----------

Adding high contrast for betting lighting, at night, or when it rains.

Hoping to get betting lighting overall.

Tested on my own Loras, Getting back the 3d looks as they should be.

Hoping to have the GoodFit with all Loras.

Testing around lots of checkpoints, Merged with the best few I have used so far.

Hoping to show fewer broken hands, especially with doing squatting poses, or poses when the hand is too close to the feet. Hands are more likely to go behind or on the knee, other than mixing with feet.