民间浴室-Folk Bathhouse-ShowerGirls- SDXL1.0 & Flux.1D & SD1.5

SDXL1.0 버전 설명:

저가 사용하는 LoRA 가중치 계수는 0.85입니다(다른 가중치도 테스트해 보시고, 자신에게 가장 적합한 값을 찾아주세요).

핵심 트리거 단어는 아래 Flux.1D 설명을 참조하세요.

저가 사용하는 샘플러는 DPM++ 2M SDE이며, 스텝 수: 30, 프롬프트 관련 계수: 4.5입니다.

SDXL1.0 version description:

The Lora weight coefficient I use is 0.85 (you can also test other weights until you find the one that suits you)

For core trigger words, please refer to the Flux.1D description below

The sampler I use is DPM++ 2M SDE, number of steps: 30, and the prompt word correlation coefficient is 4.5

VAE 모델 기반 학습된 다운로드 링크(Download link based on VAE model training)：sdxl_vae.safetensors

---

Flux.1D 버전 설명:

LoRA 가중치 계수는 0.8-1을 참조하세요(다른 가중치도 테스트해 보시고, 자신에게 가장 적합한 값을 찾아주세요).

핵심 트리거 단어와 자연어를 결합하여 촬영 각도를 묘사할 수 있습니다.

핵심 트리거 단어: 욕실에서 벌거벗은 여성이 샤워를 하고 있으며, 몸 전체에 물 얼룩이 있고 증기로 뒤덮여 있다. 이 장면은 핀홀 카메라로 촬영되었다(이미지의 네 변에 명확한 검은 프레임이 있으며, 주변 조명은 부드럽다: 1.4). 다양한 세면용품과 병들이 주변에 놓여 있고, 세면대에는 물 얼룩이 있으며, 증기 효과는 흐릿하다. 전반적인 디테일이 풍부하고 사실적이며, 젖꼭지는 섬세하다.

저가 사용하는 샘플러는 공식 권장인 euler이며, 스케줄러: simple, 스텝 수: 30입니다.

저의 홈페이지에 게시된 Flux-Lora 이미지는 매우 안정적입니다. network_dim 및 network_alpha 설정 로직은 SD1.5 및 SDXL과 다르며, 파일이 클수록 효과가 좋은 것이 아닙니다. 팔로워분들은 안심하고 다운로드하여 사용해 주세요.

Flux.1D version description:

Lora coefficient weights can refer to 0.8-1 (you can also test other weights until you find the one that suits you)

You can describe the shooting angle through core trigger words + natural language.

Core trigger words: In the bathroom, a naked woman is taking a shower, with water stains all over her body and steam all around. This scene was shot with a pinhole camera (there are obvious black frames on the four sides of the picture, and the ambient light is soft: 1.4). Various toiletries and bottles are placed around, there are water stains on the countertop, and the steam effect is hazy. The overall details are rich and realistic, and the nipples are delicate.

The sampler I use is the officially recommended euler, scheduler: simple, number of steps: 30

The Flux-Lora image posted on my homepage is very stable. The setting logic of network_dim and network_alpha is different from SD1.5 and SDXL. It is not the case that the larger the file, the better the effect. All followers can download and use it with confidence.

중첩 트리거 단어에 사용할 수 있는 자연어 참조(See the natural language used for nested trigger words)：

：In this picture we can see a woman sitting on the floor without clothes. We can see the tap and a bucket. In the background there is a toilet seat, bucket and a wall.

：In this picture we can see a naked woman standing and holding a shower head. We can see the pipes and a wall in the background.

：In this picture I can see a woman standing naked and holding a pipe in her hand, there are buckets, there is a tap attached to the wall, and in the background there are window blinds.

:a woman taking a shower in a bathroom, holding a shower pipe in her hand. On the left side of the image there is a pipe, and in the background there are tiles on the wall.

: In this image we can see a woman standing on the floor. We can also see a wash basin with a tap, a container and a wall.

:We can see this woman standing on the floor. On the left side of the picture, there is a wash basin with a tap and a few objects placed on it. There is a mirror attached to the wall.

: In this picture I can see a naked woman standing in front and I see that, she is standing in a shower. In the background, I can also see the wall, a glass door and a white color thing. On the top right corner of this picture, I see a watermark.

:In this image we can see a person standing on the floor and holding a towel. We can also see a bathtub, a glass door and a wall.

:In this picture we can see a naked woman standing and holding a towel. Behind there is a bathtub and a curtain.

:In this image we can see a woman standing under the shower. We can also see a bathtub, a metal rod and some ceiling lights to a roof.

---

SD1.5 버전 설명:

V2 버전은 핀홀 카메라 효과와 이미지의 사실성을 크게 향상시켰습니다. 프롬프트 단어를 통해 렌즈 각도를 조정할 수 있습니다. 예: from above.

The V2 version greatly enhances the pinhole vision and the realism of the picture. You can adjust the lens angle according to the prompt words, for example: from above。

V2: LoRA 가중치 계수는 0.65를 권장합니다(개인 취향에 따라 조정 가능)

The recommended Lora weight coefficient is 0.65 (you can also adjust it according to your preference)

V1: LoRA 권장 가중치 계수: 0.65-0.8 (개인 취향에 따라 조정 가능)

핵심 트리거 단어: 이 LoRA의 핵심 개념을 더 사실적으로 재현하고자 한다면, 프롬프트 단어에 다음 가중치를 추가하세요: (이미지의 네 변에 명확한 검은 프레임이 있고, 주변 조명은 부드럽습니다: 1.4)

---

A woman is bathing in the bathroom, with water stains on her body and steam surrounding her. This scene is captured by a pinhole camera, with obvious black frames on all four sides of the picture. There is soft ambient light, a red bucket in the corner, various toiletries and bottles placed around, water stains on the countertop, and a hazy steam effect. The overall details are rich and realistic, Completely naked。

SD1.5 version description:

Lora recommended weight coefficient: 0.65-0.8 (you can adjust it according to your preference)

Core trigger words, if you want to restore the core concept of this lora more realistically, you can increase the weight of the prompt words: (with obvious black frames on all four sides of the picture. There is soft ambient light:1.4)

A woman is bathing in the bathroom, with water stains on her body and steam surrounding her. This scene is captured by a pinhole camera, with obvious black frames on all four sides of the picture. There is soft ambient light, a red bucket in the corner, various toiletries and bottles placed around, water stains on the countertop, and a hazy steam effect. The overall details are rich and realistic, Completely naked。

저의 LoRA를 좋아하신다면, “buzz”를 사용해 저를 응원하기 전에, 제 홈페이지의 작품을 자세히 살펴보세요. 즐거운 생성 되시길 바랍니다!

If you like my Lora, you can take a closer look at my homepage works before deciding whether to use "buzz" to encourage me. I wish you happy drawing!