Machine Power Report
Editor: Sia
Big model replaces the editor assistant, and is more reliable than Carrot Run as a driver. Much more.
It is said that self-media uses photos of Trump’s assassination as illustrations and charges 2,100 yuan per picture!
The days of just taking a picture to illustrate an article are gone forever.
However, the creators’ troubles did not end there.
After writing a public account article, the next ten minutes are the most head-scratching:
Title shop, prime location, every inch of land is valuable. In addition to highlighting the highlights of the content, you also need to ensure that the title is attractive enough to click and read. This is much more difficult than writing an article!
What? The aroma of wine is not afraid of the deep alley?
In the Red Sea public account circle, this doesn’t work.
Don’t be discouraged either! After running in for a while, we found that the large model is quite effective in solving these two daily challenges.
-1-
Create a qualified title in the shortest time
For the convenience of chatting, let’s take this reporton the new robot basic model company Skild AI as an example.
If you know the highlights that need to be highlighted in the article, but you just don’t know how to create an attractive permutation and combination, you might as well tell the big model:
1. This is an article for a WeChat official account and needs an attractive title. ;
2. The title should highlight the data advantages, financing amount and robot GPT
This is Claude 3.5 sonnet’s idea, not bad! 
Can’t even figure out what the highlights are?
Then let the big model help you find it first, and then let him or her focus the found highlights on the title.
Claude 3.5 sonnet’s ability to summarize is very good.
Don’t even bother to think about the highlights? Then just "follow the gourd and draw the scoop"!
Some people say they like the title of "This Site" style. Wait, what does that smell like? We haven’t really counted it yet.
Let the big model analyze it: technology and big IP should be prominent and eye-catching...
Hmm, it seems like this.
gave several similar titles to the new reports. It really has an internal flavor, and the first three can be used almost directly, except that they are a bit long. 
What if you prefer Weibo hot search style titles?
We take the hot searches at a certain time on July 15 as input and let the large model analyze the characteristics of these hot search titles.
Next, give this article a hot search style title. Is it okay? At least, we have a basic plan for further improvement (everything is difficult at the beginning): 
You can also ask for more alternatives: 
Some people like the professional and serious style of financial media headlines, please arrange it.
We collected the headlines of Caixin Weekly cover reports from three years and let Claude 3.5 sonnet analyze them.
Short, pun, major topic, problem-oriented..., several key points have been captured: 
These are several titles created by the big model for our report.
What’s in it? Are the styles of the first three titles indeed somewhat similar? 
-2-
Imitating pictures, AI focuses on one province
In order to avoid copyright disputes, more and more people are starting to use AI to generate the cover of articles.
In many cases, rather than letting the AI be clueless and imaginative, we prefer that it directly imitates some ready-made pictures.
Generally, we will first let the large model look at pictures and speak, give prompt words, and then throw these prompt words to the picture generator.
This is the real experience of a colleague:
If you find that the prompt words are not easy to use, you can ask the large model (Claude 3.5 sonnet) to "carefully describe the details and content of the picture":
Then, use these Description thrown to the image generator:
This image shows a unique and contrasting scene:
Background: The background of the image is a gray concrete wall that looks rough, mottled and textured.
Seat: There is a simple black bench or platform in the center of the picture, which also looks to be made of concrete or some kind of hard material.
Character: A young man sits on the right side of the bench. He was wearing a dark suit, turtleneck sweater and black-rimmed glasses. His posture is relaxed, with his hands on his legs.
Robot: On the left side of the bench, sits a small white robot. The robot has a round head with red eye-like lights. There are some colorful decorations on its body that look like a control panel.
Light: The light in the picture is very distinctive. It comes from the right side, casting shadows of people and robots on the wall and ground. This light and shadow effect adds drama and depth to the image.
Ambience: The whole scene gives a futuristic and technological atmosphere. Humans and robots sit side by side, seeming to hint at a future where humans and machines coexist.
Composition: The composition of the picture is simple and powerful, symmetrical from left to right, with a blank space in the middle, which enhances the contrast and connection between humans and robots.
This picture cleverly displays the theme of the coexistence of humans and artificial intelligence through a concise and meaningful scene setting, triggering thinking about the future development of science and technology.
Of course, you can also use Kuaishou The large model of the self-developed Wensheng diagram is available, and it is now available for free for a limited time. The reputation is also good.
However, the large graphable model is not like DALL-E-3 and cannot digest the large amount of description content generated by Claude 3.5 sonnet.
Therefore, when entering the "Creative Overview", it must be shortened and modified:
A simple black bench, with a young man sitting on the right side of the bench. He was wearing a dark suit and turtleneck sweater, and black-rimmed glasses, with his hands on his legs. To the left of the bench, sat a small white robot with a round head with red eye-like lights. . There are some colorful decorations on its body that look like a control panel. The light in the picture is very distinctive, coming from the right, casting shadows of people and robots on the wall and ground.
Then "pad the image", adjust the degree of similarity, and generate the result: 
However, for this picture, the effect is very poor.
In addition to the sudden increase in people and machines, if you look closely at one of the characters, you can see that the character has a Lego-like face. 
Recently, technology media The verge reported on the first generative biographical film, the protagonist is Brian Eno, the music producer of David Bowie, U2, Coldplay and others.
Every time this movie is played, the movie materials will be recombined according to a special algorithm, so what you see is different every time.
This collage-style cover design caters well to the theme of the movie and also shows the versatility of a biographical figure.
We first let Claude 3.5 sonnet describe the image carefully, and then threw its description to DALL-E-3 to generate the image.
How to say? It feels neat and a bit dull. 
We have simplified the description of Claude 3.5 sonnet into one sentence, thrown it to the large model, and put the picture on it.
This time, the effect of Ketu is obviously better. 
We also tried to imitate the creative and unique cover of The Economist.
This time, the visual effect of Tutu (lower right) is better than that of DALL-E-3 (upper right). However, poor spelling ruined the result.
We used DALL-E-3 to generate a similar Styled pictures that express LLM pose a threat to personal privacy and data.
-3-
Text and fingers
are still the weakness of the picture generator
The previous cases have told us:
If the picture must contain words and text, it will be difficult for TA to do it!
You see, Ketu still misspells OpenAI, and DALL-E-3 often does the same.
It is very difficult to correctly display "Siemens" in the picture of DALL-E-3, whether in English or Chinese.
Generate a picture about Siemens Industrial Copilot, the two key words are also wrong. 
Although I knew earlier that it involves finger details, the image generator is still prone to problems.
But I never expected that even a picture like a flower picking its nose would be difficult to succeed. 
This is the work of DALL-E-3, it is really "amazing".
Throwing it to Ketu, TA was in a difficult position, and he was too embarrassed to put his fingers in front of his nostrils: 
As for the gun in his hand, well, it only had six fingers: 
This is not surprising.
Image generators usually use diffusion models to reconstruct images from noise and learn patterns that cover more pixels. Naturally, they perform poorly in generating details such as text and fingers.
Of course, this does not mean that text generators must be spelling masters. Although the underlying technologies behind image and text generators are different, they have similar difficulties in spelling and other details.
After all, we still lack basic common sense of the physical world and the language world.
The above is the detailed content of Trump's assassination photo costs 2,100 yuan? ! Article titles, accompanying pictures, AI will give you a dozen for free. For more information, please follow other related articles on the PHP Chinese website!
![[Web front-end] Node.js quick start](https://img.php.cn/upload/course/000/000/067/662b5d34ba7c0227.png)
