사용 설명서를 읽으세요.

Please read the instructions.

설명서를 읽고 Google 번역을 사용하세요.

사용설명서를 읽고 Google 번역을 사용하세요.

---

가중치 Weights: 0.58-0.8, 추천 recommend: 0.65-0.7

상반신 / 카우보이 샷: 0.65-0.7

풀바디: 0.68-0.8

NSFW: 0.55-0.7

가중치가 높을수록 3D 감이 강해지고 과적합이 심해집니다. The higher the weight, the stronger the 3D feeling and the more serious the overfitting.

————————————————————————————————————————————————————————

추천 사용: 고해상도 복원 (Hires. fix), IMAGE-TO-IMAGE / controlnet-tile 확대 재생성

2.5D 스타일 체크포인트 사용을 권장합니다

【추천 체크포인트】: V2- OnlyAnime |《唯》· 화려한 애니메이션 /model/105955/onlyanime-or

【부정적-embaddings】: EasyNegative, negative_hand-neg ||| UnrealisticDream (선택 사항 optional)

【CFG】: 7

【VAE】: kl-f8-anime2

——————————————————————————————————————————————————————

【트리거 단어】

1. 긍정적 프롬프트: molianA1CB, 1girl, purple eyes, short hair, brown hair, purple_see-through_cape, white_bodysuit, see-through, see-through sleeves, gloves, boots, hairband, hair ornament,

2. 부정적 프롬프트: 3D, 3D rendering, weapon, gun, rifle, holding, holding weapon, holding gun, white background, grey_background, blue_background, reflection, gradient background, gradient, text

3. 부정적 embaddings: EasyNegative, negative_hand-neg

【참고 프리셋, 이 기반으로 수정 가능】

1. 긍정적 프롬프트: molianA1CB, 1girl, solo, standing, looking at viewer, purple eyes, short hair, brown hair, purple_see-through_cape, white_bodysuit, see-through, see-through sleeves, gloves, boots, hairband, hair ornament, medium breasts, naughty face, cowboy shot, outdoors, street, blue sky, tree, city, sunny, summer, day, sunlight, cowboy shot, masterpiece, best quality, (colorful), (finely detailed beautiful eyes and detailed face), cinematic lighting, extremely detailed CG unity 8k wallpaper, lora:molianA1CB-v1:0.65

2. 부정적 프롬프트: EasyNegative , negative_hand-neg, 3D, 3D rendering, weapon, gun, rifle, holding, holding weapon, holding gun, white background, grey_background, blue_background, reflection, gradient background, gradient, text

3. 기타 파라미터: DPM++SED Karras / DPM++2M SDE Karras / DPM++3M SDE karras, 40 steps, 512X768 / 512x1024, Hires.fix 활성화, R-ESRGAN 4X+Anime6B, 확대 배율: 1.5, 재생성 강도: 0.25-0.4

——————————————————————————————————————————————————————

【현재 문제점 및 제안】

1. (1) 원거리 샷(풀바디)을 생성할 때 캐릭터의 눈과 얼굴 정확도가 손상될 수 있으나, 아래의 【방법 3】로 수정 가능합니다.

(2) 캐릭터의 머리 장신구, 망토 및 하의 복장에 오류가 생길 수 있으나 전체적인 인상에는 영향을 주지 않습니다.

(3) sparkling eyes, naughty face, smile 등의 프롬프트를 사용해 캐릭터 인상 향상 가능합니다.

(4) 이미지 변화가 부족할 경우, 우선 euler a 샘플러로 변화가 있는 이미지를 생성한 후 【방법 3】으로 재생성하세요.

2. 원거리 또는 풀바디 장면을 생성할 때는 적절히 출력 해상도와 스텝 수를 늘리세요. 이는 캐릭터 얼굴의 정확도 향상에 도움이 됩니다.

3. 위의 【방법 2】의 효과가 만족스럽지 않다면, 생성된 이미지를 직접 IMAGE-TO-IMAGE로 전송해 확대 재생성하세요【효과 매우 좋음】. 방법은 다음과 같습니다:

(1) 원본 이미지와 동일한 프롬프트, 캐릭터 LoRA, seed를尽可能 사용하세요.

(2) controlnet--tile 활성화, 확대 스크립트 Ultimate SD upscale 활성화

(3) 참고 설정 파라미터: Steps: 30-40, 재생성 강도 0.2-0.4, 확대 알고리즘: 4x-UltraSharp 또는 R-ESGAN 4x+anime6B, 확대 배율: 1.25--

4. NSFW 콘텐츠 생성 시 실제 상황에 따라 가중치를 0.55-0.7로 낮추고, 더 많고 정확한 NSFW 관련 프롬프트를 사용하세요. NSFW 관련 프롬프트의 가중치를 높이고, purple_see-through_cape, white_bodysuit, see-through, see-through sleeves, gloves, boots, hairband, hair ornament 등의 의상 관련 트리거 단어를 부정적 프롬프트에 추가하세요.

5. 이미지의 디테일이 부족하거나 흐릿할 경우, 반복 스텝을 적절히 증가시키거나 샘플러 및 VAE를 조정하세요. 또는 【방법 3】을 이용해 I2I 재생성도 시도해보세요.

6. 부정적 embaddings로 EasyNegative 사용을 권장합니다.

7. 체크포인트의 일반성 문제 가능성

---

Weights: 0.58-0.8, recommend: 0.65-0.7, The higher the weight, the stronger the 3D feeling and the more serious the overfitting.

upper body / cowboy shot: 0.65-0.7

fullbody: 0.68-0.8

NSFW: 0.55-0.7

Recommended use: High-resolution fix (Hires. fix), IMAGE-TO-IMAGE / controlnet-tile enlargement and redrawing

It is recommended to use 2.5D style checkpoint

[Recommended Checkpoint]: V2- OnlyAnime |《only》·Colorful animation /model/105955/onlyanime-or

[Negative-embaddings]: EasyNegative, negative_hand-neg ||| UnrealisticDream (optional)

【CFG】: 7

【VAE】: kl-f8-anime2

——————————————————————————————————————————————————————————————

【Trigger Words】

1. Positive prompt words: molianA1CB, 1girl, purple eyes, short hair, brown hair, purple_see-through_cape, white_bodysuit, see-through, see-through sleeves, gloves, boots, hairband, hair ornament,

2. Negative cue words: 3D, 3D rendering, weapon, gun, rifle, holding, holding weapon, holding gun, white background, grey_background, blue_background, reflection, gradient background, gradient, text

3. Negative embaddings: EasyNegative, negative_hand-neg

[A reference preset from which you can make modifications]

1. Positive prompt words: molianA1CB, 1girl, solo, standing, looking at viewer, purple eyes, short hair, brown hair, purple_see-through_cape, white_bodysuit, see-through, see-through sleeves, gloves, boots, hairband, hair ornament, medium breasts, naughty face, cowboy shot, outdoors, street, blue sky, tree, city, sunny, summer, day, sunlight, cowboy shot, masterpiece, best quality, (colorful), (finely detailed beautiful eyes and detailed face), cinematic lighting, extremely detailed CG unity 8k wallpaper, lora:molianA1CB-v1:0.65

2. Negative prompt words: EasyNegative , negative_hand-neg, 3D, 3D rendering, weapon, gun, rifle, holding, holding weapon, holding gun, white background, grey_background, blue_background, reflection, gradient background, gradient, text

3. Other parameters: DPM++SED Karras / DPM++2M SDE Karras / DPM++3M SDE karras, 40 steps, 512X768 / 512x1024, enable Hires.fix, R-ESRGAN 4X+Anime6B, magnification: 1.5, heavy Drawing range 0.25-0.4

——————————————————————————————————————————————————————————————————————————————————————————————————————————————

[current problems and suggestions]

1. (1) When drawing a picture of a long-range shot (full body), the accuracy of the eyes and face of the character may be flawed, but it can be repaired in the following way [Method 3].

(2) There may be errors in the character's headgear, cloak and leg clothing, but it will not affect the overall character look and feel.

(3) You can try to use sparkling eyes, naughty face, smile and other prompt words to enhance the look and feel of the character.

(4) If the changes in the picture are not rich enough, you can first use the euler a sampler to get the picture with changes, and then use [Method 3] to redraw

2. When drawing distant views or full-body scenes, please increase the drawing size and number of steps appropriately, which will help improve the facial accuracy of the characters.

3. If the effect of the above method [2] is not satisfactory, please directly send the generated image to image-to-image for enlargement and redrawing [the effect is very good], the method is as follows:

(1) Try to use prompt words, characters lora, and seed that are consistent with the original picture.

(2) Enable controlnet--tile, enable the amplification script Ultimate SD upscale

(3) A reference setting parameter: Steps: 30-40, redraw amplitude 0.2-0.4, amplification algorithm: 4x-UltraSharp or R-ESGAN 4x+anime6B, amplification factor 1.25--

4. When creating NSFW content, please reduce the weight to 0.55-0.7 according to the actual situation. At the same time, use more and more precise NSFW-related prompt words, increase the weight of NSFW-related prompt words, and change purple_see-through_cape, white_bodysuit, see-through, see-through sleeves, gloves, boots, hairband, hair ornaments, and other clothing categories Trigger Words Add negative prompt words.

5. When you feel that the image details are insufficient or blurry, please increase the number of iteration steps appropriately, or adjust the sampler and VAE. You can also try the method [3] for I2I redrawing.

6. It is recommended to use negative emaddings: EasyNegative

7. Possible checkpoint versatility issues