Original link: http://afoo.me/posts/2023-06-11-mj-VS-sd.html
How to choose between MidJourney and Stable Diffusion? – Teacher Fuqiang said: the thinking and precipitation of an architect
Fu Qiang said that an architect ‘s thinking and precipitationHow to choose between MidJourney and Stable Diffusion?
Wang Fuqiang
In fact, the description of the title is not rigorous. To be precise, Stable Diffusion is a type of model (family), and the bottom layer of MidJourney is also a Stable Diffusion model. Therefore, the Stable Diffusion we mentioned in this article refers more to the open source product Automatic1111 Stable Diffusion WebUI (Others like Vladmandic Stable Diffusion WebUI or Anapnoe Stable Diffusion WebUI-UX fall into this category).
The initial feeling of MidJourney (hereinafter referred to as MJ) and Stable Diffusion WebUI (hereinafter referred to as SD) is that MJ has a lower entry threshold and a more friendly user experience, so it is easier for users to get started and it is easier to generate satisfactory images when they come up. SD is complicated, and it is not easy for people who don’t know how to adjust it to produce good images, especially if you use SD 1.5 or SD 2.1 basic models directly.
Although MJ is easy to use to get satisfactory results, it needs special channels to reach (some people here will not understand, just forget it), and has its own content review and regulations. SD has no such restrictions and can be used locally. Unlimited, you can also use resources on the cloud, the only two that are not user-friendly are:
- The threshold for getting started is slightly higher, and non-professionals are already dizzy after reading the tutorial;
- If there is no one-stop installation package, the installation and configuration will be hell level for non-professionals;
Does this mean that everyone can choose MJ? In fact, it is not necessarily true. Today, the demands of real industry insiders for MJ and SD are not as incompatible as it seems at first glance, but they will be used in combination.
MJ can help creators quickly experiment and stimulate creative inspiration. After selecting a preliminary satisfactory image, it can be transferred to SD for refined and non-standardized work.
Therefore, there is no harm in using both, especially in various personalized scenes, SD is often more convenient.
To put it bluntly, it is a combination of the upper limit and the lower limit. MJ has a high lower limit but a low upper limit, and SD has a low lower limit and a high upper limit. The high entry threshold for SD can be learned from professional learning, but the problem of low MJ upper limit can only be solved by other tools. Make up.
BTW. Teacher Fuqiang made a set of SD series courses last month, covering all aspects of SD from theory to practice to scenarios. Interested students or students who need it at work can pay attention to it;)
On the planet “Crazy for AI”, Mr. Fuqiang is discussing interesting AI topics with his friends. Would you like to come together? ^-^
here
- Not only timely and fresh AI information and in-depth discussions
- Also share AI tools, product approaches and business opportunities
- There are more than 1,000 original paid content (nearly 500 minutes) waiting for you, join the planet (https://t.zsxq.com/0dI3ZA0sL) to get it for free!
Everything is homebrewed with pandoc and Markdown , little Scala also included.
This article is transferred from: http://afoo.me/posts/2023-06-11-mj-VS-sd.html
This site is only for collection, and the copyright belongs to the original author.