How to generate

Shedevrum uses a neural network to help you generate an image or apply filtrums to your photo.

To create an image using the neural network:

1. Tap at the bottom and select Image.

2. Open the list of models and choose one:

   • v.2.7 creates sharper and more detailed images compared to previous versions.

   • v.2.5 generates images with text in Latin script and automatically selects the optimal style for your prompt.

   • v.2.0 creates text in Latin script for images and can mix different styles in a single image.

3. Describe the image you want to get:

   • Describe the image in your own words. The more detailed your description is, the more relevant the generated image will be. For example, instead of "a lamb in the sky", enter "a fluffy lamb cloud flying in the sky, cute, gentle, colorful starry sky, 4k quality". You can describe the image in any language.

   • To generate a description, tap .

4. Tap Create. The neural network will start generating the image. It usually takes under a minute.

   Note

   Images are generated in low quality, but the quality and sharpness are improved when you publish the image.

5. Change the generated image:

   • To generate another image from the same description, tap at the bottom.

   • To change the description, tap your prompt and edit it. Tap Create.

   • To delete the image, tap in the menu below.

6. Prepare your image for posting:

   • Add hashtags — tap in the menu under the image.

   • Control who can see your post and prompt — tap in the menu under the image.

7. Select one of the generated images and tap Publish at the bottom of the screen. The image will appear in your profile and in other users' feeds.

   After the image is published, you can share or download it.