Stable Diffusion web UI SD1.5 Recommended LoRA model
This article is quite voluminous and We would like to introduce Lora models for SD1.5 models. The models will be divided into “character-related”, “style-related”, “correction-related”, “clothing-related” and “scenery-related”.
The sample images shown for each Lora model are generated using the same prompts as much as possible using Checkpoint “Dark Sushi Mix 2.25D”. Let’s compare how well they compare to illustrations without Lora, The Lora model produces high-quality illustrations with specific prompts and beautiful details. On the other hand, illustrations without Lora provide random and unique results, but the quality is not consistent. Both are attractive, but it is important to choose the right one for your application.
Parameters
Sampling steps: 20
CFG Scale: 5
ADetailer: on
ControlNet: Model: control_v11f1e_sd15_tile
VAE: toneRangeCompressor_trcvae
Correction in Photoshop after generationPrompt
(ultra art illustrated style, masterpiece:1.3), ultra detailed, dutch angle,
1girl, beautiful face,
(medival, mage:1.1), hill_side, blue_sky,horizon,Negative Prompt
EasyNegative, (worst quality:2), (low quality:2), (normal quality:2), lowresThis setting is used as a base to generate, the additional Lora trigger words that will be introduced respectively. In post-processing, levels are adjusted to bring back the shadows and highlights compressed by VAE.
Also, please refer to this article on how to install the model.
SD1.5 Character Relationship Lora Model
If you browse Civital, you will see models of copyrighted characters. Although it is only a painting style and does not violate any rights, it is not appropriate to post sample images here. When using character models, keep them for personal enjoyment only.
SD1.5 Style Related Lora Model
Anime Lineart / Manga-like
“Anime Lineart / Manga-like” by CyberAIchemist is a popular Lora model on Cibitai. if you want to monocrome image, put monochrome in the trigger word and adjust the weight or convert it to grayscale in post.You can also adjust the thickness of the line drawing by the size of the CFG scale.
- Creator: 🔗CyberAIchemist
- Download: 18.11 MB
- Upload: 2023/3/25
- File Name: animeoutlineV4_16.safetensors
Trigger words
lineart, monochrome, Recommended settings
Model: Anything V4.5
VAE: Orangemix
Weight: 1
Sampler: DPM++ 2M Karras
Sampling steps: 20
CFG: 7
Negative embedding: EasyNegative, badhandv4
Highres fix: on墨心 MoXin
simhuang’s “墨心 MoXin” is a Lora model that becomes a Chinese ink painting style. changing the CFG Scale changes the style, so you can experiment with different styles.
- Creator: 🔗simhuang
- Download: 144.14 MB
- Upload: 2023/2/24
- File Name: MoXinV1.safetensors
Trigger words
shuimobysim,
wuchangshuo,
bonian,
zhenbanqiao,
badashanren,Recommended settings
Model: ChilloutMix, GuoFeng3.2
Waight: ~ 0.85 The style changes depending on the CFG scale.
- 1 ~ 3: Xieyi Painting: Freehand captures expression rather than detail.
- 3 ~ 7: Gongbi Painting: It is the oldest style in China with a focus on detail.
cute sketch draft style
hsj1174416191215’s “cute sketch draft style” is a Lora model that adapts rough line drawings. Although the model page does not explain anything specific, We recommend applying a lower weight.
- Creator: 🔗hsj1174416191215
- Download: 168.52 MB
- Upload: 2023/8/27
- File Name: sketch v2.safetensors
Trigger words
sketch, 3D rendering style
LONGD’s “3D rendering style” is a Lora model that can be made into a 3DCG style. If you want to make it more like an illustration, you can put 3d, realistic, etc. in the prompt.
- Creator: 🔗LONGD
- Download: 36.11 MB
- Upload: 2023/7/1
- File Name: 3DMM_V12.safetensors
Trigger words
3DMMAnime Tarot Card Art Style
Anime Tarot Card Art Style” by CyberAIchemist is a Lora model that can be made to look like tarot card illustrations. The outlines are slightly emphasized to create a Mucha-like design.
- Creator: 🔗CyberAIchemist
- Download: 36.11 MB
- Upload: 2023/3/25
- File Name: animetarotV51.safetensors
Recommended settings
checkpoint: Anything V4.5
VAE: orangeminx VAE
Negative embeddings: EasyNegative and badhandv4
waight: 0.8~1Pony Pixel Art XL & 1.5
“Pony Pixel Art XL & 1.5” by Creativehotia is a Lora model that can generate dot art styles.This model is also available for SDXL/Pony, and the model page recommends adding the quality tags score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up, for Pony. SD1.5, it does not have to be added.Hires.Fix would be difficult to use because of the crushed dots, and SD1.5 may not be suitable because it cannot generate high-resolution AI illustrations.
- Creator: 🔗Creativehotia
- Download: 36.13 MB
- Upload: 2023/7/20
- File Name: pixelartV3.safetensors
Trigger words
pixelart, pixel, Recommended settings
waight: 1Real Mechanical Parts
Real Mechanical Parts” by Reelai is a Lora model that can generate mechanical AI illustrations. This model is recommended if you want to generate AI illustrations of cyborgs.
- Creator: 🔗Reelai
- Download: 487.33 MB
- Upload: 2024/4/22
- File Name: Reel_mechanical_parts_v_1_3.safetensors
Trigger words
reelmech, mechanical parts, joints, mecha, Recommended settings
Prompt: cable, wires, reelmech and mechanical parts,
NegativePrompt: animal
Weight: 1.1~1.5KIDS ILLUSTRATION
“KIDS ILLUSTRATION” by Clumsy_Trainer is a Lora model that generates AI illustrations for children. It can generate AI illustrations with a soft touch, so they look like picture books.
- Creator: 🔗Clumsy_Trainer
- Download: 144.11 MB
- Upload: 2023/5/11
- File Name: COOLKIDS_MERGE_V2.5.safetensors
CrystallineAI-konyconi
“CrystallineAI – konyconi” by konyconi is a Lora model that adds a crystalline taste. The effect is quite strong, so if you want to add a little, adjust the weights.
- Creator: konyconi
- Download: 144.11 MB
- Upload: 2023/4/24
- File Name: CrystallineAI-000009.safetensors
Trigger words
crystallineAIPastel color
Pastel color” by nnna is a Lora model that gives a pale look to the illustration. This model is trained in 512 x 768 size using CounterfeitV30_v30.safetensors, so considering the compatibility, We think it is better to use CounterfeitV30_v30. It seems to be good at generating Showa-retro style street scenes, so it was not a good match for this medieval-style fantasy worldview.
- Creator: 🔗nnna
- Download: 36.1 MB
- Upload: 2023/8/19
- File Name: Pastel color.safetensors
Recommended settings
Model: CounterfeitV30_v30
size: 512x768
prompt: store, bicycle, no one, white background, simple background, (painting:0.5), lanterns, (watercolor:0.5), traditional media, lanterns, store, store, plants, landscapeDark Fantasy
“Dark Fantasy” by dabbinggoose is a Lora model that generates a night scene. However, since the style of the image becomes quite dark, it can be used to create a dark fantasy world, as the title suggests, rather than to make the scene darker.
- Creator: 🔗dabbinggoose
- Download: 144.11 MB
- Upload: 2023/3/28
- File Name: fantasyV1.1.safetensors
Butterfly & Flowers Multiply Style
KoRo_’s “Butterfly & Flowers Multiply Style” is a Lora model that can be used when you want to scatter butterflies. It is recommended to include delayed flowers, watercolor, in the prompt.
- Creator: 🔗KoRo_
- Download: 36.11 MB
- Upload: 2023/6/22
- File Name: butterfly.safetensors
Trigger words
butterfly style, Alpha Fantasy Touch
“Alpha Fantasy Touch” model by OrangePumpking is a Lora model that can be used when you want to add a fantasy touch. However, its use is limited because the illustrations are more realistic.
- Creator: 🔗OrangePumpking
- Download: 37.27 MB
- Upload: 2023/9/16
- File Name: BetaFantasy.safetensors
Trigger words
fantasy, art,SD1.5 Correction Related Lora Model
Detail Tweaker
“Detail Tweaker” by CyberAIchemist is a Lora model that adjusts the amount of writing in an AI illustration. Specifying a negative weight will give a flat, animated look, and specifying a positive weight will give an illustration with more detail.
- Creator: 🔗CyberAIchemist
- Download: 36.11 MB
- Upload: 2023/5/5
- File Name: add_detail.safetensors
Recommended settings
waight: -2 ~ 2Add More Details-Detail Enhancer
Lykon’s “Add More Details – Detail Enhancer” is another Lora model similar to the “Detail Tweaker” introduced earlier. It is often used when you want to write details of an illustration rather than using it for negative designations.
- Creator: 🔗Lykon
- Download: 9.11 MB
- Upload: 2023/6/2
- File Name: more_details.safetensors
Trigger words
Not Available
Recommended settings
Weight: 0.5~1Hyper detailer
JujoHotaru’s “Hyper detailer” is another Lora model that adds details just like “Add More Details” above. It does not support the negative direction, so it is basically used when you want to add details.
- Creator: 🔗JujoHotaru
- Download: 36.11 MB
- Upload: 2023/11/18
- File Name: hyperdetailer_v095.safetensors
Trigger words
Not Available
Recommended settings
Weight: 0.3~1flat2
2vXpSwA7’s “flat2” This Lora model produces a flat illustration as opposed to the three above. This model produces a flat illustration in the positive direction. It also increases the writing in the negative direction. In all of the Lora models for detail adjustment, increasing the amount of writing will result in a darker illustration, so adjust the weight and CFG scale to achieve the desired amount of writing.
- Creator: 🔗2vXpSwA7
- Download: 6.55 MB
- Upload: 2023/5/31
- File Name: flat2.safetensors
Recommended settings
Weight: -1~1Hyper refiner
This is “Hyper refiner” by JujoHotaru, the same as “Hyper detailer” introduced earlier, which is a Lora model that improves the texture of the paint and eyes in AI illustrations. It removes grainy noise and makes eyes clearer. This model is also included in “Hyper detailer”, so you can use it together when you want to fine-tune your illustrations.
- Creator: 🔗JujoHotaru
- Download: 36.11 MB
- Upload: 2023/11/14
- File Name: hyperrefiner_v090.safetensors
Contrast control
“Contrast control” by JujoHotaru is a Lora model that can adjust the contrast of generated illustrations. The sample image is slightly more strongly adjusted in the positive direction (contrast becomes darker) so that the effect can be easily seen.
- Creator: 🔗JujoHotaru
- Download: 13.02 MB
- Upload: 2023/11/25
- File Name: hotarucontrast_v100.safetensors
Recommended settings
Weight: -1.5~1.5SD1.5 Clothing Related Lora Model
concept:holographic clothing
“concept:holographic clothing” by Yueluo_ is a Lora model that generates holographic clothing. It is recommended for creating futuristic clothing. It also works well with medieval-style illustrations such as the one shown in this sample illustration.
- Creator: 🔗Yueluo_
- Download: 9.11 MB
- Upload: 2023/3/6
- File Name: holov7.safetensors
Trigger words
holographic, holographic clothing, Recommended settings
Weight: 0.7~0.9Mechanical armor
“Mechanical armor” by Kairen92 is a Lora model that generates mechanical clothing. At first glance, the model looks futuristic, but it can generate medieval fantasy outerwear. However, the background is a bit mechanical, so you can express a medieval fantasy world if you combine it with scenery Lora and other models.
- Creator: 🔗Kairen92
- Download: 108.11 MB
- Upload: 2023/8/18
- File Name: Clothing – Mechanical Armor.safetensors
Trigger words
mech4rmor, glowing, Recommended settings
Weight: 0.65 - 0.85Available prompts
mechanical knight armor,
mechanical barbarian armor,
mechanical rogue armor,
mechanical wizard robe,
mechanical coat, Dark Priest
“Dark Priest” model by Kairen92 introduced earlier is a Lora model that can generate dark priests. The sample image looks like a priest, but a dark priest can also be created if you choose.
- Creator: 🔗Kairen92
- Download: 144.11 MB
- Upload: 2023/7/31
- File Name: Clothing – Dark Priest.safetensors
Trigger words
darkpriest, robe/tabard, holding staff, Recommended settings
Weight: 0.6~0.8HX_Armour
“HX_Armour” by Hustmox is a Lora model that generates armor. There are many armors on the model page, so choose the design you like. The sample illustration is generated by mixing 47 and 57.
- Creator: 🔗Hustmox
- Download: 72.11 MB
- Upload: 2024/5/8
- File Name: HXarmour_057.safetensors
Trigger words
HXARMOURRecommended settings
Weight: 0.5~1Medieval Style Armor Suit
“Medieval Style Armor Suit” by axebro is a Lora model that produces medieval-style armor. This model does not have a wide range of designs, but it produces good-looking armor.
- Creator: 🔗axebro
- Download: 144.1 MB
- Upload: 2023/5/11
- File Name: MedievalArmor_v2.safetensors
Trigger words
medieval armor, Recommended settings
Weight: 0.8Urban Samurai
eHeAI’s “Urban Samurai” is a Lora model that generates street tech wear. This model will affect the base checkpoint model quite a bit, so adjust the weights. If you don’t want to generate a mask, you can adjust the weights by adding mask to the negative prompt.
- Creator: 🔗YeHeAI
- Download: 144.11 MB
- Upload: 2023/3/23
- File Name: urbansamurai_v0.3.safetensors
Trigger words
techwear jacket, black gloves, tactical vest, with buckle and tape, Recommended settings
Prompt: techwear jacket,
with buckle and tape,
black cap and mask,
black gloves,
tactical vest,
*holding a katana, SD1.5 Landscape Related Lora Model
As_云境 Cloud landscapes
As_7033’s “Cloud landscapes” is a Lora model that generates fantastic clouds and mountains. In the sample image of the Lora model for landscapes, the prompt dutch angle is changed to full body to depict the background.
- Creator: 🔗As_7033
- Download: 72.11 MB
- Upload: 2023/10/15
- File Name: as_cloud_04.safetensors
Trigger words
cloudRecommended settings
30+steps, eular aXSarchitectural-14Openingenvironment
“XSarchitectural-14Openingenvironment” by XSarchitectural is a Lora model that generates a dense forest. There are many trigger words, but you don’t have to use all of them, just use the words you like.
- Creator: 🔗XSarchitectural
- Download: 36.11 MB
- Upload: 2023/3/29
- File Name: xsarchitectural-14Openingenvironment.safetensors
Trigger words
very atmospheric, bright, fhd, 4k, high resolution, hyper quality, high detail, realistic, hyperrealism, ultra wide shot, canon 5d, architectural visualisation, photo realistic, beautiful lighting, CANDYLAND
“CANDYLAND” by RIXYN is a Lora model that generates a fancy world view. Since it is basically a model that generates landscapes, using it for the background of a person will even affect the person, so please adjust the weights to find the line you prefer.
- Creator: 🔗RIXYN
- Download: 144.12 MB
- Upload: 2023/6/25
- File Name: Candyland-10.safetensors
Trigger words
candyland, full background, Recommended settings
“HELPFUL PROMPTS” on the model page will enhance the world view.
Sci-fi Environments
“Sci-fi Environments” model by Ciro_Negrogni is a Lora model that generates sci-fi worldviews. Be careful not to download the SDXL model by mistake, as it generates various sci-fi worlds.
- Creator: 🔗Ciro_Negrogni
- Download: 36.11 MB
- Upload: 2023/7/9
- File Name: Sci-fi_Enviroments.safetensors
Recommended settings
Please refer to the prompts on the “example prompts” section of the model page.
PAseer's mythology wallpaper
Aseer’s “PAseer’s mythology wallpaper” is a Lora model that generates mythical landscapes and dynamic clouds.
- Creator: 🔗Aseer
- Download: 36.11 MB
- Upload: 2023/6/27
- File Name: PAseerCloudV1.safetensors
Trigger words
clouds,
Each of the following is used to produce the cloud you wish to generate.
golden clouds, Clouds of Autumn Rain, Clouds of Dreams, Clouds of aquamarine, Pink clouds, Recommended settings
Waight: 0.5~0.8
Prompt: oil painting, ink painting, pencil painting,
landscapes, outdoors, Chinese architecture, Egyptian, Norse, XSarchitectural-13Nightmagictown
“XSarchitectural-13Nightmagictown” by XSarchitectural is a Lora model that generates a medieval-style night town. You do not have to use all of the Trigger words, so use whatever words you like.
- Creator: 🔗XSarchitectural
- Download: 36.11 MB
- Upload: 2023/3/29
- File Name: xsarchitectural-13Nightmagictown.safetensors
Trigger words
a beautiful, complex and detailed medieval european village, a foreground, a rather dark sky, but bright, sacred, nestled around a mountain range where enormous amounts of golden veins are buried, landscape, indirect lighting, 4k, overall brightness, volumetric lighting, epic composition, photorealism, ultra hd, ultra detailed, Shishkin landscape
“Shishkin landscape” by Ashley_Jones_fan is a Lora model that can be used to generate a forest. The model is trained on oil painting data, so it produces trees with an oil painting look. In addition to forests, the model can also generate cabins by prompting cabin/house.
- Creator: 🔗Ashley_Jones_fan
- Download: 144.11 MB
- Upload: 2023/7/28
- File Name: SHISHKIN(LANDSCAPE)_V1.2-000001.safetensors
Submerged ruins
“Submerged ruins” by Tooru111 This Lora model generates a submerged city. It adapts well to medieval fantasy worlds, as in this case.
- Creator: 🔗Tooru111
- Download: 144.12 MB
- Upload: 2023/6/18
- File Name: Submerged ruins-v1.1-000004.safetensors
Trigger words
building, city, ruins, ocean, Edob Fairy Tale Landscape
Edob Fairy Tale Landscape” by edobgames is a Lora model that generates a fantasy worldview. This model also produces a nice worldview.
- Creator: 🔗edobgames
- Download: 144.11 MB
- Upload: 2023/10/31
- File Name: EdobFairyTaleLandscape_v1.0.safetensors
Trigger words
EdobFairyTaleLandscape, Recommended settings
They recommend downloading the “Training Data” on the model page and pasting the contents of the TXT file in the Zip file into the prompt.
DreamscapeIllustration
“DreamscapeIllustration” by kuiwaiwai is another Lora model that generates fantastic landscapes. This model also generates beautiful backgrounds.
- Creator: 🔗kuiwaiwai
- Download: 36.11 MB
- Upload: 2023/10/27
- File Name: DreamscapeIllustration.safetensors
Trigger words
dreamscape, dream, Recommended settings
mountain, water, cloudConclusion
Thank you very much for reading to the end of this article, which has become quite voluminous. Also, each model has its own license, so if you are considering commercial use, please read the license section of the model page carefully. There are many more models We would like to introduce, but We will do so at another time.
