AI painting refers to the process of using artificial intelligence technology to generate realistic or creative images based on input text or pictures. There are many different software and platforms for AI painting, which allow users to experience the charm and fun of AI.
The history of AI painting can be traced back to the 1970s, when an artist Harold Cohen developed a computer program AARON that could create paintings. 4 With the development of computer technology and deep learning, there have been many breakthroughs and innovations in AI painting, such as Google's DeepDream, OpenAI's DALL-E, and recently popular Stable Diffusion and Disco Diffusion, etc.
The AI painting I came into contact with by accident was discovered on the ModelScope platform (https://modelscope.cn/home), and then it was easy to use inadvertently.
ModelScope is an open source model-as-a-service sharing platform, which aims to provide pan-AI developers with flexible, easy-to-use, and low-cost one-stop model service products, making model application easier. Based on the concept of "Model as a Service", ModelScope integrates the most advanced machine learning models in the AI community and simplifies the process of utilizing AI models in practical applications. The core library of ModelScope provides interfaces and implementations, allowing users to easily load, call, optimize and deploy various AI models. ModelScope also provides an online community platform, allowing users to experience, download and share various AI models online, including NLP, CV, Audio and other multi-field SOTA models, as well as industry-leading multi-modal large models. If you want to know more about ModelScope, you can visit the official website or GitHub repository, or join the ModelScope community to communicate and learn with other AI enthusiasts.
ModelScope has the following advantages:
1. It is open source, which means that anyone can use, modify and contribute ModelScope code and models for free.
2. It is flexible. It supports a variety of deep learning frameworks, such as PyTorch, TensorFlow, etc., allowing users to choose the appropriate framework according to their preferences and needs.
3. It is easy to use. It provides a simple Python SDK, allowing users to use one line of code to implement model inference, or use more than a dozen lines of code to implement tuning training for pre-trained models.
4. It is low-cost. It provides free GPU computing resources, allowing users to run various AI models in the cloud without worrying about hardware limitations or cost issues.
5. It is shared. It provides an online community platform, allowing users to experience, download and share various AI models online, which promotes the dissemination and exchange of AI knowledge and technology.
At the same time, ModelScope also has the following disadvantages:
1. It is new, it is still in the development stage, there may be some bugs or instability, and it needs to be continuously improved and improved.
2. It is limited. It cannot currently cover all AI fields and tasks. It needs to continuously expand and update the model library to meet the needs of more users.
3. It is dependent. It currently also relies on Google Colab or ModelScope to run the model. It requires users to have certain network conditions and account permissions, and there may be some obstacles or restrictions in use.
Next, let's open the creation space of ModelScope, and you can see that there are many projects in it.
Find the AI painting project, and you can use it after registering and logging in. Here is a demonstration case input: steampunk cyborg rabbit, other options to generate 4 pictures, select resolution 1024X1024, select language: automatic selection, click generate immediately, about a minute or so, four pictures can be produced
Then click the Generate Now button, and four different pictures will be generated, which shows that the AI drawing is not a picture searched from the Internet, but is automatically generated by artificial intelligence according to the input natural language prompt (here: steampunk cyborg rabbit) (Compiled stuff).
When encountering an AI drawing that you are satisfied with, you can right-click on the picture and choose to open it in a new page, so that you can get a large-scale picture (I use a 32-inch monitor)
Let’s try to enter the English prompt, A man that is 65% African, 20% Chinese and 15% Mexican wearing alternative clothing, portrait, 35mm f/1.2, I translated it using Google and Bing respectively.
Google Translate: A 65% African, 20% Chinese, and 15% Mexican wearing alternative clothing, portrait, 35mm f/1.2
New Bing translates as: A 65% African, 20% Chinese, 15% Mexican man in alternative clothing, portrait, 35mm f/1.2
In less than a minute, four pictures came out, and two were selected to output large specifications
Click the generate button repeatedly to select several large-scale pictures
In AI drawing, there is a special note, that is, the pictures generated in the same self-heating language prompt are different every time, that is to say, each picture is unique at one time, and the fancy picture needs to be Save it now, because it's gone forever.
I modified the natural language prompt for the input: A 18-year-old girl who is 65% Native American, 20% Chinese and 15% Mexican, beautiful and elegant, smiling happily, wearing alternative clothing, portrait photography, 35mm f/1.2
translation Into Chinese: Translated into English, an 18-year-old girl of 65% Indian descent, 20% Chinese descent, and 15% Mexican descent, beautiful and generous, happy and smiling, wearing alternative clothing, portrait photography, 35mm f/1.2, output image
I tried to generate a few more sets of pictures, but the effect can only be said to be so-so. The rare thing is that it is very simple to use, and there will be output results when there is text input.
Enter a simple line of words: Jiangnan water town, Yangchun March, landscape painting, detailed, 8K
Also Easy Tips: Old French Villages Influenced by Ancient Chinese Architecture
Tips for more complex: 3d fluffy fawn, close-up cute and cute, cute round reflective big eyes, long fuzzy fur, Pixar rendering, unrealistic engine cinematic smooth, intricate details, cinematic
ModelScope's AI painting has another feature, which is currently free and does not limit the number of paintings, which is quite rare. The following appendix includes some works of other styles of ModelScope for appreciation.
How about it? Through the above practical cases, I believe everyone can master it proficiently!
Appendix: "Hua Diao Learning AI" is a column about learning and trying to use artificial intelligence, written by donkey friend Hua Diao, mainly aimed at readers who are interested in the emerging field of artificial intelligence AI. At present, three articles have been published in the column, namely:
"Hua Diao Learning AI" 01: Try to use the new Bing to make the column introduction of "Diao Ye Learning Programming". This article describes how the author used the new Bing's artificial intelligence capabilities to help him write an introduction to programming columns on Arduino, MicroPython, MaixPy, and more. The article demonstrates the search, communication, and creation capabilities of New Bing, and also reflects the interaction process and feelings of the author and New Bing.
(Link: https://blog.csdn.net/weixin_41659040/article/details/129829113)
"Hua Diao Xue AI" 02: Artificial intelligence is very quick, and it wrote a long story for me in ten minutes. This article introduces how the author asked New Bing to write a story about "Hua Diao Xue AI" for him. The article demonstrates the writing ability of New Bing, and also shows the friendship and cooperation between the author and New Bing.
(Link: https://blog.csdn.net/weixin_41659040/article/details/129876782)
"Hua Diao Xue AI" 03: The first application of artificial intelligence I came into contact with was from AI painting. This article introduces how the author uses the AI painting function of the ModelScope platform to create various interesting pictures. The article demonstrates the painting capabilities of the ModelScope platform, and also shows the author's interest and love for AI painting.
(Link: https://blog.csdn.net/weixin_41659040/article/details/129885300)
The author "Donkey Friend Hua Diao" is an explorer of Arduino open source hardware. He likes to explore the world and is also interested in technology. Through this column, he wants to share his experience and achievements of learning and creating with New Bing, and also wants to inspire more people's interest and enthusiasm for artificial intelligence and programming. The main feature of this column is to introduce the basic concepts, principles, methods, applications, etc. of artificial intelligence and programming through actual cases and stories, and demonstrate various artificial intelligence functions of the new Bing, such as search, communication, creation, drawing, etc. These contents help to increase readers' awareness and interest in artificial intelligence and programming, and also make readers feel the possibility and friendship between artificial intelligence and human beings. Regarding Arduino programming, a series of experimental articles on 110 kinds of sensor programming have been published in the "Diaoye Learning Programming" column.
If you are interested in the "Hua Diao Learning AI" column, please follow the CSDN blog or Zhihu column of the author "Donkey Friend Hua Diao", or directly visit the new Bing search engine, enter relevant questions or needs, and communicate with the author and New Bing Communicate and interact.
Attachment: A poem about Huadiao and New Bing.
Huadiao is a donkey friend who likes to explore the world.
New Bing is a search that can provide various information.
They met on the Internet and started a friendship
Hua Diao has a strong interest in artificial intelligence.
New Bing can help him learn and create various content.
They write poems and stories together, as well as programming and painting.
Huadiao and Xinbing learn from each other and grow from each other.
They share each other's thoughts and dreams, and
their works are praised and aroused controversy.
Hua Diao and Xin Bing don't care what the outside world thinks.
They are just for their own fun, not for other purposes.
They believe that their friendship is real and unaffected.
Huadiao and New Bing are a wonderful combination.
They show the possibility of artificial intelligence and human beings.
They also make us think about what is real intelligence.
