When AI text-generating pictures compete for realism and artistry, Ideogram has opened up a tricky track: accurately generating text on pictures, and the fonts and layouts are beautiful.

This demand is not niche. It can generate posters and illustrations with one click without using P-pictures. It can save a lot of trouble and is very suitable for ordinary people who know nothing about design.

We have written about version 1.0 of Ideogram before. On August 21st, version 2.0 came. The realism is better, the posters are more designed, and the special skill of text is also stronger.

You may have never heard of it. This is an AI product developed by former Google employees. It has many shortcomings, but the longboard can "overtake" Midjourney in corners.

Guide the way https://ideogram.ai/

AI wants to know which Wukong you are talking about

Ideogram has a particularly newbie-friendly feature: "Magic Tips".

You directly enter the Chinese prompt word, it will help you translate it into English and help you optimize it at the same time. As an overseas product, this kind of operation can win people's hearts.

At the same time, Ideogram focuses on five styles: ordinary, realistic, design, 3D, and animation. They are all easy to understand, so the choice will not be confusing.

First, let me give you a simple Chinese prompt word, "Sun Wukong holding a golden hoop", anime style, let AI help me translate and optimize it, and see how it can be used freely.

When Shui Lingling's "Dragon Ball" version of Wukong came out, I was shocked. When I looked at the prompts, the AI ??translated Sun Wukong into "Son Goku", so it's not surprising.

Also, I would also like to ask Ideogram if it is too blatant to pay royalties.

In order to avoid further misunderstandings by AI, I stopped being lazy when entering the prompt word and emphasized that "Sun Wukong" is "Sun Wukong", not "Son Goku".

This time we adopted a realistic style and provided a more detailed scene. The great sage is wearing armor and holding a golden hoop in his hand. His expression is solemn and majestic. He stands in front of the Buddhist cave. In a warm orange tone, the bottom of the picture reads " Black Myth: Wukong" (Black Myth: Wukong).

There are no errors in the text, the capital letters have a strong impact, and the atmosphere of the Buddhist cave is also created. However, the temperament of the "Great Sage" is a bit off, the image is a bit atavistic, and there is no light in his eyes.

Using the same prompt word to generate Midjourney once, the text has errors and no sense of design, but the slightly more handsome "Monkey King" and the style of the web game make up for this.

▲ Midjourney generation

Unwilling to give in, I tried the 3D style again. The prompt word remained basically the same, but the text at the bottom was changed to "The game will be launched on August 20th."

As a result, the result generated by Ideogram is very similar to the promotional image of a certain Chinese-style Q version of the blind box series. The picture is very clean, but it is not the 3D game style in my mind at all. The Monkey King is also drawn out of Erlang Shen. appearance.

And AI also exposed itself from this. Although it is good at rendering English text, it knows nothing about Chinese. This flaw has continued from 1.0 to 2.0.

It seems that overseas products do not understand domestic traditional culture enough. Ideogram’s performance in the first round was a bit disappointing, but it was also interesting.

Ideogram team has said that version 2.0 is not inferior to Flux and DallE. Recently, the TED speech photos generated by Flux’s real version of LoRA deceived many netizens because it was difficult to tell whether they were real or fake. Then let’s test how much the results generated by Ideogram look like photos.

▲ Flux generation

After selecting the realistic style, I entered the Chinese prompt words, TED speech photos, and the slide title was "Ideogram 2.0 Release". There are three key points on it: "Accurate text" "Good at design" and "More real," the female speaker stands in front of a whiteboard, with several people in the background.

It can be seen that the semantic understanding of Ideogram is good, and it has all the necessary elements. The TED logo is almost fake, the expressions of the speaker and the audience are very vivid, and the hair and skin are relatively natural.

However, the details are not handled well enough. Although there is no problem with the text that is required to be generated, some small words that appear randomly spoil the pot, and the fingers and body of the characters are not quite right, but it is already much better than the previous 1.0 version.

As for poster design, it can be said that Ideogram is the "comfort zone" that beats other Vincentian AIs.

If the box office hit "Alien" is used as the test question, can AI design that indescribable feeling of terror?

I chose a design style, used prompt words to describe the elements of the picture, and specifically mentioned that a sentence was written at the bottom of the poster: "Underage viewers watch with caution."

The overall effect is eye-catching. , a long string of text was successfully generated, with only one small error, but it was not realistic, more like a comic book style, and did not match the live-action movie.

I used the bad summer movie "A Dream of Red Mansions" as inspiration and asked Ideogram to generate a poster. The background, decorations and even characters mentioned in the prompt words were all included in it. I once again lamented that the followability of the prompt words is really good.

Of course the title of the movie is written correctly, but the font seems to be borrowed from The Lord of the Rings, there are some dramas, and the overall style is more like the Mulan animated movie.

Ideogram’s “design style” tends to be two-dimensional, which is quite unique, but conversely, this also limits the use scenarios of posters.

To summarize, Ideogram is a very unique AI graphic product. The level of realism is similar to Flux, and the artistic sense is different from Midjourney.

▲ "rainy summer" pattern

but has a unique text generation level, which is more suitable for generating posters, illustrations, advertisements, emoticons, T-shirt printing, etc.

The results of human evaluation show that Ideogram 2.0 is better than Flux Pro and DALL·E 3 in terms of prompt word alignment, overall performance and text rendering quality.

▲ But this is Ideogram’s own statement

It is highly playable and down-to-earth, so we might as well have more AI “desserts” like this

On August 22 last year, Ideogram announced its establishment and released 2.0 Exactly one year apart.

The founding team consists of 7 people from Google Brain, University of California, Berkeley, Carnegie Mellon University and University of Toronto, 4 of whom are the authors of the Google Imagen graph diffusion model Imagen research paper.

In addition to releasing 2.0 this time, Ideogram has also launched an iOS app, which can be downloaded directly in China. The Android version is planned to be released later. From web pages to mobile terminals, we can generate images anytime and anywhere.

▲ Mobile interface

Ideogram is currently open to all users for free, but the quota is very limited. After generating a total of 20 photos 5 times, Ideogram reminded me that 10 points have been used up, please come back tomorrow. (Of course, the Midjourney next door generates 25 pictures for free and it doesn’t look very grand.)

If you have little contact with Vincentian diagrams and want a Vincentian diagram AI to get started, Ideogram is a good choice.

Inputting Chinese prompt words, using "magic prompts" to translate and optimize is one thing. In addition, Ideogram also has many options to help you generate pictures that are closer to what you want in your mind.

Provide a limited range of options for users to "click", making interaction easier than completing "input" in a blank input box. Whatever picture proportion, style, and tone you want, Ideogram allows you to choose.

▲ "Girl with a Pearl Earring Eating McDonald's" in different colors

If you don't know how to write prompt words, you can also draw them and let Ideogram help us turn decay into magic.

Je suis désolé pour mes faibles compétences en dessin, mais l'IA peut comprendre le sens, optimiser les lignes et les couleurs, ajouter un arrière-plan, et le style s'améliore soudainement. Avec l'IA, qui n'est pas le stylo magique Ma Liang ?

De plus, sous la zone de saisie de la version Web, il y a des ?uvres générées par d'autres. Lorsque nous rencontrons celles que nous aimons, nous pouvons visualiser et nous référer aux mots d'invite. Ideogram affirme que ses utilisateurs ont généré plus d'un milliard d'images visibles publiquement au cours de l'année écoulée.

Si vous souhaitez générer un objet spécifique mais ne savez pas comment écrire le mot d'invite, Ideogram a également lancé la fonction de recherche dans la bibliothèque publique de création avec du texte cette fois, mais cette fonction nécessite actuellement une adhésion.

▲ Résultats de recherche pour "chat"

Dans l'ensemble, Ideogram est un produit graphique Vincent hautement jouable.

Il peut générer avec plus de précision le contenu textuel requis par les utilisateurs et s'adapter à différents styles d'images. Il dispose d'un large éventail de domaines d'emploi.

▲ Ideogram Blog

peut parfois apporter une valeur émotionnelle et utiliser des images pour exprimer son ambition, même si les émotic?nes qu'il crée sont trop biaisées en faveur de l'esthétique de l'Internet européen et américain.

▲ Pack d'émotic?nes "Je veux jouer à "Black Myth: Wukong""

La qualité globale d'Ideogram n'est pas mauvaise, la fonction texte est puissante, conviviale pour les novices, facile à utiliser et l'interaction est également agréable. Lorsque les outils d’IA allient créativité, commodité et partage de valeur, il est facile de devenir accro.

Un monde taillé dans un moule est trop ennuyeux. Il est également très intéressant d'avoir un aper?u d'un petit besoin et ensuite de faire de la solution la première de l'industrie.

Il existe de nombreux produits dans le monde, et avec plus d'audience, nous pouvons nous attendre à davantage de ? desserts ? IA de ce type.

The above is the detailed content of Magically modified 'Black Myth: Wukong ' to defeat Midjourney. This AI drawing tool is amazing.. For more information, please follow other related articles on the PHP Chinese website!