In 2023, the first year of AI painting.
First, let me share my thoughts on AI painting. As a former digital art worker with an art background, I personally don't oppose AI painting, because as a tool, it's really useful, time-saving, and cost-effective. It saves me time on production and eliminates the need to hire others.
Unfortunately, for most art practitioners, we are indeed in a time of change. AI painting will eliminate most people who love art but currently lack sufficient skills to make a living from it. It will then return art to the realm of the top talents. This is good for art, but not so good for most practitioners. Just like today, people learn AI painting because, on a commercial level, it has already replaced traditional painting. This is seen from an observer's perspective. China has long been the world's largest producer of digital art outsourcing, providing inexpensive painting services to game and animation companies worldwide. To a large extent, AI drew from the blood of Chinese painting workers during its initial training. Now it's replacing their work. After AI emerged, many outsourcing companies began to lay off employees and replaced them with AI. Combined with the current employment environment, this is why many domestic practitioners hate AI so much.
However, every technological leap has always eliminated the previous generation of workers. Just like the invention of combines eliminated professional wheat harvesters, the sand of the times, when it falls on an individual, becomes a mountain. The wheels of time crush all those who cannot escape. People can only go with the flow. So, young friends, when the combine harvester is already in front of you, don't complain. Put down your sickle and become one of the first combine harvester drivers!
===========================================================
170 selected portrait sketches from joint exams have been refined and compiled.
The base model uses ChilloutMix. I have tested it and it works well on most excellent realistic models. Anime and 2.5D models can also be used, but it goes against the original intention of the title "Joint Exam Sketch Portrait". Please use cautiously;
The V1.0 material set includes sketches of various angles, genders, and age groups, with no half-length portraits or still life objects. Therefore, it may not produce good results when exported as a half-length portrait or still life;
It works best when used with tools like ControlNet to control actions;
Trigger words: portrait sketch, adding prompt words like graphite (medium), sketch can intensify the effect;
Iteration steps (Steps) are recommended to be 20~40, or 40 or more depending on the situation;
Sampling methods (Sampler) are recommended to be DPM series or Euler, Euler a, restart;
Sizes are recommended to be between 512~1024;
Prompt word guidance coefficient (CFG Scale) is recommended to be 7~12;
The LoRA weight can vary depending on the situation, generally around 1, and different models may have different degrees of fit;
For high-resolution repair (Hires. fix), the re-drawing amplitude can be adjusted between 0.3~0.5 for maximum structure preservation or 0.5~0.7 for more freedom; this is up to you;
If needed, you can turn on After Details.
===========================================================
Works in this series:
"The Tears of Artists: Joint Exam Sketch Portrait | Style LoRA"
https://civitai.com/models/267797/the-tears-of-artists-joint-exam-sketch-portrait-or-style-lora
"The Tears of Artists: Joint Exam Color Still Life | Style LoRA"
"The Tears of Artists: Joint Exam Figure Sketch | Style LoRA"
https://civitai.com/models/267852/art-students-tears-joint-exam-sketch-or-style-lora