Mann-E Dreams is the latest SDXL-based model launched by Mann-E Platform, an Iranian generative AI startup. This model utilizes thousands of mid-process images to achieve high-quality image generation. The development team, led by Mann-E founder and CEO Muhammadreza Haghiri, spent months collecting, labeling, and training the model with virtually no censorship and was tested with Automatic1111.
Demand group:
"The target audience is designers, artists and creative professionals who need to produce high-quality images. The model's fast generation capabilities and high image quality make it ideal for creative work."
Example of usage scenario:
Designers use Mann-E Dreams to create unique advertising images.
Artists use this model to generate artwork with a personal style.
Creative professionals use this model to quickly create graphic content for social media.
Product features:
CLIP Skip setting, either 1 or 2, 1 provides better results.
Steps are set to 6-10, usually 8 steps works best.
CFG Scale is set in the range of 2-4.
Supports 768x768 and 832x832 image sizes. For 16:9 ratio, try 1080x608.
The sampler supports DPM++ SDE Karras.
Compatible with SDXL 1.0 LoRas, ControlNet, IPAdapter, InstantID and other technologies.
Usage tutorial:
1. Visit the Mann-E Dreams model page.
2. Adjust CLIP Skip, Steps, CFG Scale and other parameters as needed.
3. Select an image size and sampler.
4. Enter a text prompt that describes the image.
5. Click Generate and wait for the model to output the image.
6. Adjust according to the generated results until you are satisfied.