Model
Trends
.ai
Model
Trends
.ai
die besten Open-Source-KI-Modelle
Lokal ausführen
backlighting.safetensors
144 MB · SafeTensor
Downloads werden von Civitai bereitgestellt. Manche Ersteller verlangen zuerst eine Anmeldung bei Civitai.
AD
Auf Graydient installieren / ausführen
Mitgliedschaft erforderlich
<back-light>
Über dieses Modell
STABLE DIFFUSION 1.5 · LoRAs
Erstellt von
AlchemistW
auf Civitai
晖映 Enhanced Backlighting by AlchemistW Enhanced Backlighting is a Lora used to improve the backlighting and sidelighting effects in environments with strong lighting contrast. Many models have not adjusted their noise_offset, which can result in an overly balanced overall brightness and a lack of emphasis on dark areas or photographic reflection effects, especially for 2D models. Even when the prompt for backlighting is given, the character still appears to have the same brightness as the environment, lacking a prominent highlight and making the image appear very "flat" in tone. This LORA intentionally deepens the contrast to improve this problem and has a natural light illumination improvement effect for both 2D and 3D models. Try it at a weight of 0.5 with your favorite base model as a safe starting point. Your /guidance setting may also affect the effect, so adjust both accordingly
Notizen des Erstellers
晖映 Enhanced Backlighting [Eng ver below] 607323927751959196 这是一个用来改善逆光侧光环境光照效果的Lora，许多模型的noise_offset未经调整，会导致画面整体亮度过于均衡，显得没有暗区重点或者摄影反光效果，尤其是二次元类model，即使是给出了backlighting的prompt，人物依然会显得亮度和环境保持一致，缺乏重点的突出，显得画面色调很”平“，这个LORA特意加深了对比度来改善这个问题，对于二三次元的模型都有自然光的照射改善效果。 - 特别提示：由于trigger word忘记区分，使用backlighting这个prompt会比较重的加深对比和自然光效，如果你不喜欢这种风格，可以尝试去掉它并改变Lora的权重，可以得到一些较轻的效果。计划在后续版本里通过特殊关键字解决这个问题。 - 特别提示2：这个Lora针对丁达尔现象也有不错的效果，甚至可以在二次元模型中搭配另外一个尚在完善中的Lora模拟出来。即Komorebi 日语(木漏れ日)，效果请见例图。待测试完善后会择机发布的。 - 特别提示3：Lora的权重和CFG高低都会影响阳光的效果，如果不希望画面出现太多夕阳余晖的感觉，可以通过提示词改善或者降低权重或CFG 对比效果如下： 提示词均为1girl,backlighting Enhanced Backlighting is a Lora used to improve the backlighting and sidelighting effects in environments with strong lighting contrast. Many models have not adjusted their noise_offset, which can result in an overly balanced overall brightness and a lack of emphasis on dark areas or photographic reflection effects, especially for 2D models. Even when the prompt for backlighting is given, the character still appears to have the same brightness as the environment, lacking a prominent highlight and making the image appear very "flat" in tone. This LORA intentionally deepens the contrast to improve this problem and has a natural light illumination improvement effect for both 2D and 3D models. - Special Note: Due to forgetting to distinguish the trigger word, using the prompt 'backlighting' will heavily increase the contrast and natural light effect. If you don't like this style, you can try removing it and changing the weight of Lora to achieve a lighter effect. We plan to solve this problem through special keywords in future versions. - Special Note 2: This Lora has a good effect on the Ding Darr phenomenon, and can even be combined with another Lora, which is still under development, to simulate the effect of 'Komorebi' in Japanese (sunlight filtering through trees) for 2D models. Please refer to the example image for the effect. It will be released after testing and improvement. - Special Note 3: The weight and CFG of Lora will affect the effect of sunshine. If you don't want the screen to have too much sunset glow, you can improve or lower the weight or CFG through prompt words. The comparison effect is as upside img. The prompt words are '1girl' and 'backlighting'.
Triggerwörter
backlighting
10,131 Downloads · 1,589 Likes auf Civitai
SHA256
8562C93EAE008100E8A10EE3E072D3F33C0E09EF33D534BF5BA9E20BF6764E6C
ModelTrends.ai-Modell-ID
#68
Inhalt melden
Beispielbilder
Prompts und Einstellungen, geteilt von den Leuten, die diese Bilder auf Civitai erstellt haben. Wähle ein Bild, um zu sehen, wie es entstanden ist.
Einstellungen
Sampler
DPM++ SDE Karras
Schritte
28
Guidance
7
Seed
2197519776
Größe
512x768
Prompt
13 year old girl,
Negativer Prompt
mutated hands, (poorly drawn hands:1.331),(fused fingers:1.61051), (too many fingers:1.61051), bad hands, missing fingers, extra digit,artist name,rock bad_prompt_version2 easynegative
Einstellungen
Sampler
DPM++ SDE Karras
Schritte
28
Guidance
4
Seed
1738549951
Größe
512x768
Clip skip
2
Prompt
tree,forest,
Negativer Prompt
(easynegative:1.2), (painting by bad-artist-anime:0.8), (verybadimagenegative_v1.3:1), (low quality:1.2), (worst quality:1.2),Scribbles,Low quality,Low rated,Mediocre,3D rendering,Screenshot,Software,UI,watermark,signature
Einstellungen
Sampler
DPM++ SDE Karras
Schritte
28
Guidance
7
Seed
26413083781
Größe
512x768
Prompt
13 year old girl,
Negativer Prompt
mutated hands, (poorly drawn hands:1.331),(fused fingers:1.61051), (too many fingers:1.61051), bad hands, missing fingers, extra digit,artist name,rock bad_prompt_version2 easynegative
Einstellungen
Sampler
DPM++ SDE Karras
Schritte
30
Guidance
7
Seed
2060624024
Größe
512x320
Clip skip
2
Prompt
no dyed hair, (best quality, masterpiece, perfect face, beautiful and aesthetic:1.2, colorful, dynamic angle, highest detailed face), In the gentle glow of a campfire, two girls share ghost stories, their faces lit by the flickering firelight, a sense of eerie excitement in the air. <lora:backlighting:0.5>
Negativer Prompt
FastNegativeV2,
Einstellungen
Sampler
DPM++ SDE Karras
Schritte
28
Guidance
7
Seed
2790742734
Größe
512x768
Prompt
13 year old girl,backlighting,
Negativer Prompt
mutated hands, (poorly drawn hands:1.331),(fused fingers:1.61051), (too many fingers:1.61051), bad hands, missing fingers, extra digit,artist name,rock bad_prompt_version2 easynegative
Einstellungen
Sampler
Euler a
Schritte
25
Guidance
7
Seed
3076279856
Größe
512x768
Clip skip
2
Prompt
13 year old girl,tree,outdoors,
Negativer Prompt
mutated hands, (poorly drawn hands:1.331),(fused fingers:1.61051), (too many fingers:1.61051), bad hands, missing fingers, extra digit,artist name,rock bad_prompt_version2 easynegative
Einstellungen
Sampler
Euler a
Schritte
25
Guidance
7
Seed
159594039
Größe
512x768
Clip skip
2
Prompt
13 year old girl,backlighting,
Negativer Prompt
mutated hands, (poorly drawn hands:1.331),(fused fingers:1.61051), (too many fingers:1.61051), bad hands, missing fingers, extra digit,artist name,rock bad_prompt_version2 easynegative
Einstellungen
Sampler
DPM++ SDE Karras
Schritte
30
Guidance
7
Seed
4
Größe
768x1056
Clip skip
2
Prompt
<lora:cuteGirlMix4_v10:0.4>, mix4, night, Ray Tracing, realistic, photorealistic, detailed, realistic details, depth of field, professional lighting, neon city, (1 girl:1.3), ponytail, from side, pose for camera, (realistic, photo-realistic:1.37), (8k, RAW photo, best quality, masterpiece:1.2), ultra-detailed, physically-based rendering, ultra high res, looking at viewer, full body, (black techwear jacket and red trousers with buckle and tape), soft backlighting, <lora:urbanSamuraiClothing_urbansamuraiV03:0.7>, <lora:EnhancedBacklighting_v10:0.3>
Negativer Prompt
Drawings, abstract art, cartoons, surrealist painting, conceptual drawing, graphics, (low resolution:1.3), (blurry:1.3), (worst quality:1.3), (low quality:1.3), collage, bad proportions, text, username, (watermark:1.3), letter, nude,(abs), lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, Six fingers, Low quality fingers, (EasyNegative, badhandv4, verybadimagenegative_v1.3)
KI-Modelle für Bild-, Video- und Texterzeugung
back light ist ein LoRA für die Familie Stable Diffusion 1.5 und wird auf ModelTrends.ai gelistet, einem reinen Lesekatalog für Open-Source-KI-Modelle für Bild, Video und Text. Vergleiche es mit anderen Stable Diffusion 1.5-Modellen, prüfe den Hitzewert, um den Trend zu sehen, und öffne es auf PirateDiffusion oder BitVector, um es auszuprobieren.
Nach Modellfamilie stöbern
Alle Modelle ansehen