先说结论
以下SD3指的是SD3 Medium;
- SD3对提示词更敏感（Promot Follow），提示词的冲突可以直接导致画面的崩坏；
- SD3更擅长"写字"，这得益于16 channel的Vae；
- SD3的照片更真实；
- SD3对空间位置的理解更好一些；
- SD3可适应不同风格的出图诉求；
- SD3对显存的更加友好；
- SD3认识更多的物品，比如某些品种的花；
- SD3对R级内容更加安全；
以下XL指的是SDXL-LCM;
- XL对提示词不太挑剔，画面不容易崩坏；
- XL对人体的理解更深刻，不容易出现扭曲崩坏；
- XL并不安全，它的出图范围更广；
参与对比的模型
| 模型名称 | 运行环境 |
|---|---|
| SD3 Medium | 运行地址：huggingface.co/spaces/stab... |
| SDXL-LCM | 1. 运行地址：www.shakker.ai/modelinfo/d... 2. 国内地址：www.liblib.art/modelinfo/d... |
SD3 Demo Prompt
以SD3的Demo提示词为例，对比了SD3和SDXL-LCM的出图情况。
提示词矩阵
- a female character with long, flowing hair that appears to be made of ethereal, swirling patterns resembling the Northern Lights or Aurora Borealis. The background is dominated by deep blues and purples, creating a mysterious and dramatic atmosphere. The character's face is serene, with pale skin and striking features. She wears a dark-colored outfit with subtle patterns. The overall style of the artwork is reminiscent of fantasy or supernatural genres
- Digital art, portrait of an anthropomorphic roaring Tiger warrior with full armor, close up in the middle of a battle, behind him there is a banner with the text "Open Source".
- photo of a dog and a cat both standing on a red box, with a blue ball in the middle with a parrot standing on top of the ball. The box has the text "SD3"
- selfie photo of a wizard with long beard and purple robes, he is apparently in the middle of Tokyo. Probably taken from a phone.
- A vibrant street wall covered in colorful graffiti, the centerpiece spells "SD3 MEDIUM", in a storm of colors
- photo of a young woman with long, wavy brown hair tied in a bun and glasses. She has a fair complexion and is wearing subtle makeup, emphasizing her eyes and lips. She is dressed in a black top. The background appears to be an urban setting with a building facade, and the sunlight casts a warm glow on her face.
- anime art of a steampunk inventor in their workshop, surrounded by gears, gadgets, and steam. He is holding a blue potion and a red potion, one in each hand
- photo of picturesque scene of a road surrounded by lush green trees and shrubs. The road is wide and smooth, leading into the distance. On the right side of the road, there's a blue sports car parked with the license plate spelling "SD32B". The sky above is partly cloudy, suggesting a pleasant day. The trees have a mix of green and brown foliage. There are no people visible in the image. The overall composition is balanced, with the car serving as a focal point.
- photo of young man in a black suit, white shirt, and black tie. He has a neatly styled haircut and is looking directly at the camera with a neutral expression. The background consists of a textured wall with horizontal lines. The photograph is in black and white, emphasizing contrasts and shadows. The man appears to be in his late twenties or early thirties, with fair skin and short, dark hair.
- photo of a woman on the beach, shot from above. She is facing the sea, while wearing a white dress. She has long blonde hair
出图情况
其中SD3直接用了发布时的宣传Demo图
| SD3 | SDXL-LCM |
|---|---|
标签类型的Prompt
提示词矩阵
- Photography, Fujifilm Provia 400X, heavy breathing, a beautiful girl, 24yo, skinny, pretty, closed mouth, serious, [k-pop:detailed beautiful face,detailed eyes:0.6], dark red lips, body art, Polished, Leather shorts, looking over one shoulder, full body, sunlight, wind, dark style, crop top,
- aerial Perspective, professional photography, RAW photo taken using Fujifilm Provia 400X, HD, HDR, detailed texture, Natural skin, muted colors, Fairy Tale Movie Style, the cutest cat baby in the world,wearing a small brown leather back pack,winter_clothes,hat, white background, 3d printed, hdr, 8k, myth,
- chinese ink painting,Eastern artistic conception,Celestial Bridge Leading to the Stars,
- famous artwork by (Yaacov Agam:0.998),quaint,The Forbidden Forest,picturesque,serene,Exquisite Details,Evoking a sense of nostalgia and sentimentality,Vivid Colors,mysterious
- (Chibi 3D Graphics - Kirby's Epic Yarn (game),Cute knitted toy Theme,cat,american wirehair,8k
- aerial Perspective, professional photography, RAW photo taken using Fujifilm Provia 400X, HD, HDR, detailed texture, Natural skin, muted colors, Fairy Tale Movie Style, the cutest ice Monster in the world,wearing a small brown Cyberpunk glow outfit, winter_clothes, hat, hdr, 8k,
- aerial Perspective, professional photography, RAW photo taken using Fujifilm Provia 400X, HD, HDR, detailed texture, Natural skin, muted colors, Fairy Tale Movie Style, the cutest cactus baby in the world,wearing a small brown glow Cyberpunk armor, winter_clothes, hat, white background, 3d printed, hdr, 8k, myth,
- Excellent character design work, Hyperrealistic art cinematic film still photography in the style of detailed hyperrealism photoshoot, fantasy, hyper detailed, highly detailed, high budget, bokeh, cinemascope, moody, epic, gorgeous, film grain, grainy, Extremely high-resolution details, photographic, realism pushed to extreme, fine texture, incredibly lifelike, An excellent character design work, the duck queen evolved from a duck, (named_ZhangBingNv:0.1), with white skin covered with small moss, bright glowing large eyes, green algae hair, dark green lipstick, wearing a red hoodie made of crystal scales, exquisite mechanical leather pants, mysterious jungle background,
- (sfw:1.32),(Prototype Illustration:1.098),(Icon Design:1.26),2D Icon,white background,button theme,Pink,Olive,Peach,Glowing,Stop Sign,Bright Colors
- famous artwork by (Steve McCurry:1.003),picturesque,0Bg1},quaint,serene,intricate details,emitting a sense of destruction,high definition,mysterious
出图情况
| SD3 | SDXL-LCM |
|---|---|
自然语言类型的Prompt
提示词矩阵
- A young woman with short dark hair and bangs is the central figure in the image. She is wearing a white dress with a floral pattern and a white blouse. The woman is standing in front of a window, which is adorned with white curtains. The window is positioned on the left side of the image, and the woman is looking directly at the camera. The background is a simple white wall, which serves to highlight the woman and the window.
- A young woman stands in a dimly lit room, her gaze directed off to the side. She wears a black leather vest, a black tank top, and black shorts, with a tattoo visible on her left arm. The room is bathed in a soft, blue light, and a window on the right side allows a glimpse of the outside world.
- A young woman stands in a dimly lit room, her gaze directed off to the side. Her hair is styled in a braid, and she wears a black dress with a plaid pattern. The room is bathed in a soft, warm light, with a window on the left side and a door on the right, suggesting an indoor setting.
- A person is standing in a lush forest, wearing a white lion costume with a large head, sharp teeth, and a fierce expression. The costume includes a white top and shorts, and the person's arms are outstretched, as if ready to run. The forest around them is dense with trees and foliage, creating a serene and tranquil atmosphere.
- A young woman with dark hair is captured in a moment of quiet contemplation, her gaze directed off to the side. She is dressed in a black halter top, which contrasts with the blurred background of a waterfall. The waterfall, with its cascading water, serves as a serene backdrop to the woman's serious expression.
- A young woman stands in a dimly lit alleyway, her gaze directed off to the side. She is dressed in a black kimono, its long sleeves and collar adding an air of mystery to her appearance. The alleyway is lined with wooden houses, their windows glowing warmly against the encroaching darkness. The sky above is a canvas of orange and pink hues, suggesting either sunrise or sunset. The woman's position in the center of the alleyway draws the viewer's attention, and her expression is serious, adding an element of intrigue to the scene.
- A woman in a white dress stands in a foggy forest, her back to the viewer. The forest is dense with trees, their branches heavy with leaves, creating a canopy of green. The ground is carpeted with fallen leaves, adding a touch of autumnal charm to the scene. The woman's long, dark hair cascades down her back, and her white dress contrasts with the surrounding greenery. The fog obscures the background, adding an air of mystery to the scene.
- A person in a black robe stands in front of a large, ornate eye sculpture, holding a book and appearing to be reading. The eye sculpture is a large, intricate piece of art with a gold frame and a blue eye, which is the central focus of the image. The person is positioned in front of the eye sculpture, creating a sense of depth and perspective. The background is filled with bookshelves, suggesting a library or study setting.
- The image presents a floating island with a house on its surface, surrounded by a vibrant rainbow arching over the island. The island is nestled in a body of water, with a pink and blue wave surrounding it. The sky above is a clear blue, dotted with fluffy white clouds. The image is a digital illustration, with the artist's signature "LIBLIB" visible in the bottom right corner.
- A male figure stands in a dark forest, wielding a sword and surrounded by a burning fire. The figure is adorned in a black cloak and horns, with a red eye and a menacing expression. The figure's gaze is directed towards the viewer, creating an atmosphere of intense confrontation.
出图情况
| SD3 | SDXL-LCM |
|---|---|
SD3高级配置
SD3的配置如下
SDXL-LCM高级配置
因算力紧张，所以尺寸和步数缩减了一些