modèles/Detailifier - SD3.5/Flux/SDXL/Pony/(SD1.5) - Flux V1 (V5.0 Dataset)

Detailifier - SD3.5/Flux/SDXL/Pony/(SD1.5) - Flux V1 (V5.0 Dataset)

7/9/2025

11:00:46 AM

Mots-clés et tags associés

clothing detail,concept,detailerlora,detailifier - sd3.5/flux/sdxl/pony/(sd1.5),details,face details enhancement,flux.1 d,flux model v2,flux v1 (v5.0 dataset),fur rendering,img2img,lora,material texture,skin texture,txt2img,upscaling,zwaetschge

Kitchen interior of a 1990s Russian apartment featuring a wooden table with tea and bread, white chairs, potted flowers on the windowsill, cabinets, and a window showing nearby apartments.

Wide angle drone footage showing rows of Soviet-era apartment buildings in a sprawling city with tree-lined streets under a cloudy sky.

A wide-angle drone view of an American northeastern town featuring a rocky shore, multiple piers with docked sail boats, dense green trees, residential homes, and a visible downtown area under a clear blue sky.

A scenic swamp land with a calm river surrounded by lush green vegetation, tall trees, water lilies, and clear blue sky reflecting in the water.

Two men in business suits shaking hands in an 1980s office lobby with green marble walls, a chandelier, indoor water feature, and holding briefcases.

A stealthy female ninja in dark shinobi attire hides behind a sliding wooden door, spying on two Japanese lords seated inside a traditional manor with tatami floors and warm paper lantern lighting.

Silhouetted man at executive desk with a sleek humanoid robot assistant in a high-rise office overlooking a futuristic city with neon billboards at dusk.

Aerial wide-angle view showing a German town with medieval-style houses and a stone church tower contrasted by a large industrial complex with smokestacks releasing thick smoke at dusk.

Dieselpunk-inspired retro-futuristic sports car with headlights on, parked on a wet urban street at night, surrounded by thick yellow smog and industrial buildings.

A dieselpunk-style industrialist woman in early 20th-century attire smokes a cigarette in her warm, wood-paneled office overlooking a chaotic factory floor through a large reinforced window.

Dimly lit Soviet-era stairwell with cracked concrete steps, rusted iron railings, faded green peeling walls, potted plants on window sill, and a faded orange graffiti lambda symbol.

Prompts recommandés

<lora:detailer_flux_v1:1>

Paramètres recommandés

samplers

Euler

steps

20 - 30

cfg

1 - 2.5

resolution

896x1152

other models

detailer_flux_v1 (df0fed87b335), flux_dev (06f96f89f6), detailer_captions_flux-000005 (b374391d58b8), fluxunchainedNF4_fluxunchainedV11NF4 (aa18ab0754)

Paramètres haute résolution recommandés

upscale

1.5 - 2

steps

30 - 60

Conseils

Use the weak version for more control and stability across styles.

The strong version enhances fine details but may risk inconsistencies.

Upscaling at 1.5x to 2x with full strength and 30-60 steps improves details noticeably.

Most example images were generated via txt2img without editing, minor issues may appear.

This model is focused on improving the details in faces, skin, clothing, fur, and materials. It performs well in both txt2img and img2img/upscaling workflows.

Flux Model v2 - Weak vs. Strong:

The weak version is more flexible, allowing for a wide range of outputs.

The strong version enhances details more dramatically but has a higher risk of inconsistencies.

Both versions have their pros and cons. The strong model can really make details stand out, while the weak version offers more control and stability across different styles.

Recommended Settings:

CFG: 2.0 - 2.5

Steps: 20 - 30

SDXL/Pony:

Upscaling (1.5x - 2x) really brings out the fine details. I’ve personally had great results with full strength, but anything from 0.3 to 1 with 30 to 60 steps can make a noticeable difference.

Recommended Settings:

CFG: 2.0 - 7.0

Steps: 30 - 60

As a heads-up, most of the provided images (except for upscaled ones) were generated through txt2img without any editing, so there may be a few minor issues visible here and there.