Shulou(Shulou.com)11/24 Report--

[new Zhiyuan Guide] Adobe's AIGC graphic platform Firefly has recently been upgraded to Firefly 2, which improves image quality, introduces vector graph generation, and adds a number of new features, greatly improving the user experience and providing better tools for creative and design work.

Firefly, Adobe's AIGC biograph platform, has recently undergone a major update. Adobe directly named it Firefly 2.

First of all, the image model quality of Firefly image 2 has made a great leap forward.

The wrinkles of marshmallows can even be seen in the picture of the new model on the right.

The details of bird feathers and depth of field in the new model on the right are much better than those of the previous generation model. The Firefly Image 2 model produces significantly higher quality images, especially when rendering realistic human bodies, involving high-frequency details such as leaves, skin textures, hair, hand and facial features. Images generated using the Firefly Image 2 model have higher resolution and brighter colors and color contrast.

Moreover, the function of generating vector graphics is added, which can generate vector graphics directly through prompt.

Adobe also adds several new features that can significantly improve the user experience:

Generate matches: images with similar styles can be generated based on the sample images uploaded by the user.

Image settings: you can also set details for the generated pictures, including depth of field control, motion blur, field of view adjustment and generation.

Prompt optimization tips: for users' Prompt, the system can automatically generate prompts and suggestions to help users better express their needs and get a more ideal image output.

Adobe claims that Firefly 2 has greatly improved the quality of image output, especially in terms of leaves, skin textures and facial features, with more realistic and fine details than previous generation models.

Especially for portrait pictures, the rendering quality of the human body is enhanced by improving the structure of skin, hair, eyes, hands and body, providing better color and more accurate dynamic effects.

The skin of the characters generated by the new model on the right is more real and natural, and the depth of field effect of the background is more obvious.

In the picture on the left, the light and shadow effect of the new model is more realistic and consistent, and the background details are more natural. The details of the clouds in the picture are more natural and richer, and the light and shadow are softer. The details of the characters in the picture on the right are more realistic, and the hand description is more natural. The effect of marshmallows is more real and more detailed.

From the point of view of the picture content, it is clear that the new model is more accurate in understanding and depicting what a CRT display is and what a TV is. The previous generation engine basically depicts only one display screen, while the new generation engine is more reasonable in drawing the details of the display.

From the perspective of picture style, the light and shadow generated by the new model is more natural, more realistic, and the tone is more consistent.

Like several other bioglypes AI, Firefly is also free to choose the model version of the biograph. Although version 2 is theoretically better than 1, there is no problem if users want to use version 1 to create a graph.

The actual measurement of three new functions gives users the ability to finely control the picture. The change of the image model is a big upgrade of Firefly for the "overall quality" of the model output.

But the most important thing for users of bioglyph AI products is the ability to control the image generation process. Adobe's decades of accumulation in these areas is obviously not comparable to Midjourney,DALL-E 3, a novice in image products.

For this update of the generation of matching, image settings, prompt prompt suggested three functions, we also carried out a wave of measurement.

After using it, the biggest feeling is that the combination of PS+AI pictures is really cool!

First of all, you can use all these features by registering through Firefly's official website.

Entrance address: https://firefly.adobe.com/ inspire / images

Generate matching function:

Similar to the "style converter" function of other AI, the generation matching function can generate pictures with high similarity in style and content according to a picture given by the user, which can be understood as the "image to image" function of the Adobe version.

For example, when we want an eye image based on the style of a sketch pencil, we just need to provide a sketch of the same style, and then enter a prompt, such as "one's eyes".

Firefly 2 will then create four sketch pencil-style images that match the style we expect.

Use the picture of "neon theme" with the dog prompt to generate the "neon puppy".

Use the original picture of the paper-cut style, together with the prompts of the cat to generate the paper-cut-style cat.

In addition, you can also adjust the style of the image according to your personal preference and transform it into a vivid three-dimensional style, giving the image more vivid and attractive characteristics.

This feature allows you to easily customize and create unique images according to the requirements of different projects or the diversity of personal ideas.

Image settin

In addition, Firefly also provides the function of secondary editing of the generated images, which is equivalent to the ability of PS+AI pictures, and provides a very convenient workflow for users who require high picture details.

For example, you can use the image of the paper-cut kitten just generated by the system to edit the material at any time, including adding text, graphics, or other elements.

This flexibility allows you to further personalize and refine the generated content as needed, creating images that match your unique ideas to meet your specific needs.

This function provides users with more choice and control in terms of creativity and customization.

In addition, the visual intensity, color hue, lighting, synthesis can be further modified. For example, on the basis of the generated cat image, the lower right corner can change the hue, lighting and other effects.

If you want to generate a golden kitten, you can choose from the color and tone options of Firefly 2.

You can also synthesize more elements in your creation, such as scenes shot through windows.

Prompt word suggestion

As the biggest bottleneck on the human side of AI, Adobe also gives its own solution to this update.

If the user's input prompt is too short or not enough to describe the user's needs clearly, the system provides the function of prompt word suggestion.

For example, when we type a short prompt like "a woman is on top of the mountain during sunset", the system will automatically generate five different statements that will present a variety of ways to express similar situations for us to choose and further refine our creative needs.

This function allows us to customize and expand our creative ideas more flexibly, increasing the possibility and diversity of creativity.

Together, these aspects demonstrate the more powerful performance of Firefly 2 released by Adobe. It not only improves the image quality, introduces the vector graph generation function, but also adds a number of functions, greatly improves the user experience, and provides more excellent tools for creators and designers.

Reference:

Https://news.adobe.com/news/news-details/2023/Adobe-Releases-Next-Generation-of-Firefly-Models/default.aspx

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.