top of page
20230506 Lora-model
20250225 LAZYSUN_FLUX1.0_橙色蓝色霓虹灯条未来城市
LAZYSUN_FLUX1.0_Orange and Blue Neon Strip Future City
Based on the sci-fi style of FLUX.1, this project explores a future city theme, with elements of machinery, gaming CGI, and cyberpunk-inspired neon aesthetics.
The dataset provided primarily consists of illustrations depicting futuristic cities. Therefore, pairing it with photographic-style CKPTs yields exceptional results.
When attempting to generate robots, the outcome has also proven to be effective.
The dataset quality is high, which results in a similarly strong visual output.
Recommended steps: 30-35
Recommended weight: 0.5-0.9
Suggested keywords: Future city, Future science and technology, Very beautiful art style, Cyberpunk, Avatar style
20241107 LAZYSUN_FLUX1.0_欧美人物韩式厚涂风格原画LAZYSUN_FLUX1.0_Western Characters in Korean Thick Painting Style
Based on the FLUX.1 Korean thick-paint illustration model.
The main focus is on Western characters, handsome men, and beautiful women, all rendered in a Korean-style thick painting aesthetic.
The dataset primarily consists of thick-painted Korean illustrations, with some Western-style thick painting elements. Combining this with photographic CKPTs achieves excellent results.
When attempting to generate scenes, the output has also been promising.
Recommended steps: 20-30
Recommended weight: 0.7-0.9
Suggested keywords: Impasto, Fantasy Art, Scene content
20241102 LAZYSUN_FLUX1_洒金国风插画_青金色淡彩中式美学LAZYSUN_FLUX1_ Golden Ink Chinese-style Illustration_ Qingjing Aesthetic
Based on the FLUX.1 Eastern aesthetic illustration model.
The primary focus is on Qingjing hues, golden ink, traditional Chinese landscapes, architecture, and ink wash elements.
The dataset mostly features scenes, skies, and flying birds, making outdoor scenes particularly effective.
However, attempts to generate human figures have also shown positive results.
Recommended steps: 20-30
Recommended weight: 0.7-0.9
Suggested keywords: Shajin, Golden, Scene content
Large model suggestion: Generally, the models are similar in performance.
Resolution suggestion: 1024x1024
20240709 LAZYSUN_SDXL_煎蛋模型_friedegg
LAZYSUN_Fried Egg Model_Friedegg
Based on the SDXL physical object model.
The main focus is on fried egg elements.
The dataset contains a variety of fried egg illustrations, but many images also feature fries, hamburgers, and vegetables. I assumed these were all food-related, so I didn’t process them too much. I even fried multiple oddly-shaped eggs in my kitchen for the purpose of this project.
Some re-drawing efforts were particularly interesting.
Recommended steps: 20-30
Recommended weight: 0.7-0.9
Suggested keywords: Friedegg, Scene content
Large model suggestion: Generally, the SDXL models work well.
Resolution suggestion: 1024x1024
bottom of page