5. How to generate images via Xipu AI Gallery-Text to Image
An example of Gallery in XIPU AI platform
Overview:
This guide will show you how to generate images using the Gallery feature in XIPU AI.
What to know:
The Gallery feature allows users to generate images by inputting text descriptions and blending the images.
Table of Contents
Step 1: Open Gallery
Visit https://xipuai.xjtlu.edu.cn and click the Gallery icon on the top.
Step 2: Click 'Start Creating' on the Gallery page
Step 3: Input description and adjust parameters to generate images
- Input description in the Image Description area
Suggestion: Use the Assistant feature to help you optimize your input description. - Adjust parameters
- Click Relax/Fast button
Parameter details:
- Doubao, Midjourney, and DALL-E 3 are all AI image generation tools, but they each have different strengths. Doubao offers a user-friendly experience and strong capabilities in understanding prompts, making it convenient for quickly creating diverse visual content. Midjourney focuses on artistic and stylistic interpretations, making it especially suitable for visually striking and creative images. DALL-E 3 emphasizes generating images from detailed textual descriptions with a strong focus on realism and accuracy.
Midjourney parameters:
- Upload lmage: Upload reference images
- Image Weight: Choose image similarity
- Aspect Ratios: Width-to-height proportions
- Version: Choose different versions of Midjourney (Differences between the versions)
- Stylize: Controls the level of artistic style
- Chaos: Adjusts the randomness and variability
- Seed: Determines the starting point for the image generation process, influencing the randomness and consistency of the results
- No: Specify certain features or subjects that should not appear
- Tile: The ability to repeat a texture pattern seamlessly across the generated image
- Public Drawing: The results will be displayed in the public gallery by enabling it
- Fast: Image generation time is about 1 minute, 50 points.
- Relax: Image generation time is about 5 minutes, 20 points.
[Optional - Midjourney] Split the image or continue to adjust the image
Split the image:
Click the U1\U2\U3\U4 button to split the image and obtain a single image.
Regenerate
Adjust the image:
Click the V1\V2\V3\V4 button to unify the four images into the style of the specified image.
Further adjustments to the parameters can be made to achieve a satisfactory effect.
Parameter details：
- Upscale (Subtle): Enhances the image with subtle improvements in detail.
- Upscale (Creative): Enhances the image with more creative adjustments and artistic flair.
- Vary (Subtle): Generates a new image with slight variations from the original.
- Vary (Strong): Generates a new image with more significant changes from the original.
- Vary (Region): Allows you to create variations focused on a specific area of the image.
- Zoom Out 2x: Zooms out the image to show a wider view, doubling the original size.
- Zoom Out 1.5x: Zooms out the image to show a wider view, increasing the size by 1.5 times.
- Custom Zoom: Allows you to specify a custom zoom level for the image.
-
Expand the image up, down, left, and right.
[Optional] Image blend
Add images to blend
- Open the 'Blend' tab
- Add images
- Click the Relax/Fast button to blend images
Tips:
- Prompt is best used in English; using Chinese may result in the inability to generate derivative images.
- Derivative Image Limitations in Midjourney: Only images within the last 10 hours can have the button to generate related derivative images. Recommendation: If you want to generate derivative images, please do so as soon as possible.
- Task Limit: No more than five tasks at a time. If there are five ongoing tasks, you cannot continue generating.
Next steps:
Now that you have a general idea on how to create an image via the Gallery feature, you may be interested in the following related articles: