NoobAI-XL Detailer - EPS1.1_v1.0
Palabras Clave y Etiquetas Relacionadas
Prompts Recomendados
1girl, (eversoul:0.9), achyue, momoko \momopoco\, masterpiece, best quality, (torino aqua:0.4), amazing quality, very aesthetic, absurdres, newest, very awa, crisp, sharp
Prompts Negativos Recomendados
worst quality, low quality, quality, lowres, worst aesthetic, jpeg artifacts, scan artifacts, signature, bad id, bad twitter id, english commentary, bad hands, username, mutated hands, mammal, logo, anthro, furry, scanlines, blur, blurry, comic, (long body, long neck:1.3), extra digit, missing digit, old, simple background, (censored:1.3), stomach bulge, cleft of venus, bar censor, half-closed eyes, twins, splatter, plump, harness, magical girl, wet, leotard, fake animal ears, maid headdress, yellow skin, red eyes, steam, extra ears, cum, blurry background, bokeh, forehead jewel, hair intakes
Parámetros Recomendados
samplers
steps
cfg
resolution
other models
Parámetros Recomendados de Alta Resolución
denoising strength
Consejos
If not explicitly stated, the default is 99% eps prediction version.
V-Pred V0.5 works best with weight 1.5 for better image details.
V-Pred V1.0 darkens images and produces very dark imagery especially with short prompts like '1girl', 'dark', 'night'. Use more descriptive prompts to avoid black on black silhouettes.
Recommended weight range for V0.5 is -6 to +6; suggested usage between 1-2.
Merging V1.0 with custom flat color lora at weight 1 slightly alters the image.
Aspectos Destacados de la Versión
Baked a custom undertrained symthetic flat color lora and merged at negative ratio with v0.5. Almost a week of experimentation.
Result is slightly changing the scene but in my tries only for good. Due to custom synthetic lora has little to no effect on artist styles. Intended to use at weight 1.
This is simple detailer trained on model NoobAI-XL.
V-Pred versions work better on v-pred model. If you don't know what it means - read checkpoint description. If not explicitly stated - 99% it is eps prediction version.
V-pred related specifics:
V0.5 - just slap it on with weight 1.5 makes image better.
V1.0 - slightly alters image giving more details. Weird interaction that I found - it darkens dark. V-pred NOOB-AI is first SDXL model that I see that can make really dark imagery. ESPECIALLY with short prompts, like 1girl, dark, night. If you are feeling that that is not what you want - just be more descriptive. describe face, vlothes and you won't get those black on black silhouettes anymore. Btw that applies to the NOOBvp1 itself too. See example here: https://civitai.com/images/47951726
On the upside it can darken stuff that base model has issues with darkening, like some poses.
V1.0
Merged with custom flat color lora, recommended weight is 1. Slightly alters image, especially with no prompts like 1girl.
V0.5
Weight range -6 - +6
I recommend using 1-2.
Colaborador
Detalles del Modelo
Tipo de modelo
Modelo base
Versión del modelo
Hash del modelo
Creador
Discusión
Por favor log in para dejar un comentario.