ControlNet uses
ControlNet is a highly modular, flexible open source framework for robot control that supports a variety of sensors, actuators, and communication protocols. ControlNet can be used in a variety of applications, including but not limited to:
Industrial automation: ControlNet can be used in industrial automation systems, such as assembly lines, packaging and handling tasks. It can help achieve efficient, precise and reliable robot control, thereby improving production efficiency and product quality.
Unmanned driving: ControlNet can be used in the control system of unmanned vehicles, supporting data collection and processing of various sensors (such as radar, lidar and cameras), as well as actuators (such as motors and brake) control. It can realize functions such as autonomous driving, path planning and obstacle detection.
Home automation: ControlNet can be used in home automation systems to control and manage smart home devices. It can communicate with various smart home devices (such as light bulbs, curtains, air conditioners and security systems) to achieve functions such as remote control, automation and energy saving.
Healthcare: ControlNet can be used in healthcare areas such as surgical robots and rehabilitation equipment. It enables precise control of surgical operations, patient positioning and rehabilitation training.
Logistics and warehousing: ControlNet can be used in the logistics and warehousing fields to control tasks such as the handling, storage and sorting of goods. It can improve logistics efficiency and accuracy, reduce costs and risks.
ControlNet provides a highly modular framework that supports the integration of various sensors, actuators, and communication protocols, allowing developers to quickly build and expand robot control systems. It also provides a wealth of libraries and tools to help developers quickly implement various control algorithms and applications.
The use of ControlNet in Stable Diffusion
The use of ControlNet in Stable Diffusion is mainly to control and adjust the output of the model, and to optimize the performance of the model in deep learning. ControlNet can be understood as a controller that is used to set, operate and limit states on the data flow and post-processing models, and control the movement direction, movement trajectory, shape and size of the diffusion process. It can finely control the process of global flow pattern formation and the detailed expression of diffusion elements, improving visual expression through interpretability and diverse rendering effects.
ControlNet has strong flexibility and can adapt to different needs. For example, it can control the shape, size, color, transparency, etc. of diffusion elements, and can also dynamically change according to specific conditions. In addition, ControlNet is efficient and enables real-time rendering without sacrificing accuracy.
In the specific scenario of Stable Diffusion, ControlNet can help generate more natural, vivid, and rich images, improve the quality of the generated images, and meet the different needs of users. At the same time, ControlNet can also help optimize the performance of the model and improve the speed and efficiency of model generation.
Positive prompt words:
Star face, long black hair, beauty, wearing a white shirt, upper body frontal photo, ultra-clear, cute, lolita, natural black pupils, bright eyes, Chinese style, well-proportioned, regular facial features, no stretching, first love, light blue Color background, tie, campus, desks and chairs, school uniform, long hair to waist, smile, dimples
Reverse prompt words:
(semi-realistic, cgi, 3d, render, sketch, cartoon, drawing, anime:1.4), text, close up, cropped, out of frame, worst quality, low quality, jpeg artifacts, pgly, duplicate, morbid, mutilated, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, blurry, dehydrated, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, fused fingers, too many fingers, long neck
Question picture
You can see that there is a problem with the character's fingers in the picture below, so we need to use ControlNet to fix it.
ControlNet repair operations
Select the image to generate the image, which is img2img.
1. Open ADetailer
The ADtailer here needs to check [Enable ADetailer] to enable high-definition repair, and the corresponding ADetailer model selects [mediapipe_face_full].
2. Turn on ControlNet
Don’t skimp on resources here, choose the highest quality.
3. Start repairing
Click [Generate] to start.
Contrast effect:
Original picture:
Repair picture:
The effect is obvious.
Summarize
The repair operation is a very important content. Many people generate pictures that are crooked. So through this operation, it is practical to directly use the satisfactory pictures we have generated before for repair.