Front view of a vintage Ford T car with headlights on, engulfed in dramatic flames, featuring deep contrasts and warm cinematic lighting.

Recommended Negative Prompts

clone, cloned, bad anatomy, wrong anatomy, mutated hands and fingers, mutation, mutated, amputation, lowres, signs, memes, labels, text, error, mutant, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, cropped, (long neck), body horror, out of frame, mutilated, tiled, frame, border, (deformed iris:1.2)

Recommended Parameters

samplers

DPM++ 2M SDE Karras

steps

28

cfg

7

Tips

Use more general words in the prompt as the model still doesn’t know a lot.

It is recommended to use weight up to 2 inclusive.

If using Automatic1111, check the “always show all networks” option in the “extra networks” settings.

Version Highlights

Minor adjustments to contrast, detail, and composition. Learning bias towards close-up portraits. For this reason, use a weight of up to 1 inclusive.

对对比度、细节和构图进行了微调。学习偏向于特写肖像。因此,使用不超过 1 的权重。

Мелкие доработки по контрасту, детализации и композиции. Уклон обучения в сторону близких портретов. По этой причине используйте силу до 1 включительно.

An ideological continuation of Advanced Enhancer. It does the same thing, but the main focus is not on detail (because SDXL is fine with that), but on composition, color, perspective, etc.

It still doesn’t know a lot, use more general words in the prompt.

It is recommended to use weight up to 2 inclusive.

Important: If you work through Automatic1111, then in the setting “extra networks” check the checkbox next to “always show all networks”.

高级增强器的理念延续。它做的事情是一样的,但主要的重点不是在细节上(因为SDXL在这方面没问题),而是在构图,颜色,透视等方面。

它还不知道很多,用更通用的词在提示中。

建议使用重量最高为 2 的。

重要提示:如果您通过 Automatic1111 工作,则在“额外网络”设置中,选中“始终显示所有网络”旁边的复选框。

Идейное продолжение Advanced Enhancer. Делает всё то же самое, но основной упор не на детализацию (потому что у SDXL с этим и так всё в порядке), а на композицию, цвет, перспективу и т.д.

Многое ещё не знает, используйте более общие слова в промпте.

Рекомендуется использовать силу до 2 включительно.

Важно: Если вы работаете через Automatic1111, то в настройке "extra networks" поставьте галочку напротив "always show all networks".

Contributor

Previous
Phandigrams - Opus I
Next
Arcane style [PonyXL] - v1.0

Model Details

Model type

LORA

Base model

SDXL 1.0

Model version

AdvancedEnhancerXLv2

Model hash

6e8beb025f

Trained words

movie still
film still
cinematic
cinematic shot
cinematic lighting
macro shot
35mm film
green eyes
blue eyes
brown eyes
close up
teeth
braces

Creator

Discussion

Please log in to leave a comment.

Images by Advanced Enhancer XL LoRA - AdvancedEnhancerXLv2

analog Images

film Images

lighting Images

portrait Images