First, let's start with a website: https://rightbrain.art/imageToImage#
Advantages:#
-
First of all, it's free, no need to spend money to buy an account.
-
No need for magic internet access, those who understand, understand.
-
Low threshold, use Chinese directly, no need to translate into English.
-
There is a large library of models, no need to be troubled by prompts.
-
Can be modified and generated locally, this is MJ's weakness.
-
Can generate poses, pose whatever you want.
Text-to-image#
We are all familiar with text-to-image. In Vega AI, when creating a text-to-image, you first select a model, then add a text description, and you can quickly generate an image with high consistency. In this respect, it is much better than Midjourney. Take a look at the two sets of images below, do you feel familiar? Some people say that these images are vulgar and unbearable, and I also think so. But in fact, some people on Xiaohongshu quickly gained over 100,000 followers using these images. Some public accounts don't even have substantial content, they just post these types of pictures every day. The speed at which they gain followers is beyond my reach as someone who works hard every day to write valuable content. The images are definitely not generated using Midjourney or Stable Diffusion, as they are too slow. They are all generated using Vega AI. The effect is good and there is plenty of quantity. It only took me 4 minutes to generate the two sets of images below...
|
001.jpg |
002.jpg |
|
001.jpg |
002.jpg |
Image-to-image#
Image-to-image, also known as placeholder image, in Vega AI, you can directly send the images generated by text-to-image to image-to-image. Then you can choose a model you like and generate the style you want with just one click. The following image is an example where I sent the previously generated images to image-to-image and directly chose a blind box model to generate. It's super convenient, you just need to click once.
|
001.jpg |
002.jpg |
Image-to-image supports local modifications#
What's even more amazing is that image-to-image in Vega AI supports local modifications. You just need to use a brush to mark the area you want to modify, and then modify the new description. This allows you to easily change the appearance with just one click. The most important thing is that the parts outside the marked area will not be affected at all, maintaining the complete consistency of the main image. This is much more precise than Midjourney, where any changes, big or small, will affect the overall image.
Mark where you want to make changes
|
001.jpg |
002.jpg |
Conditional image generation#
Conditional image generation allows you to directly colorize line drawings, or even provide only an image feature or depth map, and it will generate an image that matches the conditions you provide. It's like a miracle. Let's try a male image this time (Note: The line drawings in the following examples are from the internet. If there is any infringement, please let me know and I will delete them!)
|
001.jpg |
002.jpg |
Pose generation#
Character model
Pose well and add a description to generate an image
Control the pose, the character's art style is up to you
Style customization#
The most powerful feature, in my opinion, is here. You can train the model with a large number of images of the same type. The more images you feed it, the more accurate the style model will be. This model training, which does not require programming skills, is quite user-friendly and extremely powerful. Once you train a style model that you like, you can generate as many images with the same style as you want. Now you know how those batch-produced beauty images are created.
Training a style model is like having a batch printer
Vega AI's Style Plaza provides a massive collection of style models that can be used directly.
Finally, this article only scratches the surface and serves as an introduction. From the perspective of overall image quality and art, Vega may still have some gaps compared to AI painting techniques like Midjourney and Stable Diffusion. However, it is more convenient and has a lower threshold. Due to space limitations, there are many other features of Vega AI that I haven't introduced. Feel free to explore and exchange ideas with each other.