FAQ — Image · Video · Enhancement · Editor
Summary: A collection of frequently asked questions and answers about image generation, video generation, enhancement tools (upscale, inpaint, outpaint, background removal), and editor/export.
3. Image Generation
What is the difference between templates and styles (LoRA)?
A template is an image published by another user — you copy the same settings to quickly generate similar images. A style (LoRA) is an AI model trained on a specific art style or brand, useful for repeatedly generating images with a consistent tone. For details, see the Style Generation Guide.
How do I write a good prompt (length / keywords / negative prompts)?
The more you describe the subject, style, lighting, colors, and mood together, the closer the result will be to what you want. The "Prompt Enhance" feature automatically expands a simple idea into a detailed prompt. If you want to exclude certain elements, explicitly state what to exclude in the idea field.
What is the recommended resolution for reference images and sketches?
A resolution of 512×512px or higher is recommended for reference images; JPG and PNG formats are supported. The clearer the image, the more accurately the AI reflects composition, colors, and style. You can add images by dragging and dropping, selecting via the file browser, or pasting a screenshot.
What is the "Imagine More" option?
"Imagine More" is an option where the AI automatically adds detailed descriptions based on your prompt to generate richer images. Enabling this option produces more varied and creative results from the same prompt. It is off by default during inpaint operations to preserve the context of the original image.
4. Video Generation
Why is an image required for video generation?
XBrush video generation works by adding natural motion to an image. The starting image becomes the first frame of the video, and the AI converts it into a 5- or 10-second clip. Higher quality source images produce more natural video results.
How does the choice of engine affect the result?
Each engine has different video generation styles, motion quality, and supported lengths. The default engine, Kling, provides stable motion and balanced quality. The list of available engines and their characteristics can be found in the engine selection menu on the video generation screen.
Why are videos limited to 5 or 10 seconds, and what are the tips?
Currently, XBrush supports 5-second and 10-second video generation due to AI model characteristics. The supported length may be fixed depending on the engine selected. If you need a longer video, try combining multiple clips in the editor.
What is prompt relevance and how do I adjust it?
Prompt relevance is a value between 0 and 1 that determines how closely the video reflects the text description. Values closer to 0 prioritize natural movement from the image; values closer to 1 follow the prompt description more closely. Start with a value between 0.5 and 0.7 and adjust from there.
What should I do if the downloaded video won't play?
Downloaded video files are in MP4 format. If playback fails, try installing a modern media player such as VLC Media Player or Windows Media Player. If inline playback is not supported in some browsers, download the file and check it with a separate player.
5. Enhancement Tools (Upscale / Inpaint / Outpaint / Background Removal)
What are the effects and limits of upscaling?
Upscaling uses AI to increase image resolution while filling in pixel details. Unlike simple enlargement, AI fills in pixels to produce a sharper high-resolution image than the original. However, if the original image quality is very low or severely blurry, there are limits to perfect restoration.
Tips for creating a good inpaint mask?
Drawing the masking area slightly larger than the part you want to replace helps produce a more natural result. The more specifically you describe what to replace in the idea field, the more accurate the result. Setting the inpaint strength to 0.8–0.9 helps it blend naturally with the original image.
What is the difference between portrait and product mode in background removal?
XBrush background removal uses AI to automatically detect the subject and remove the background. The AI detects the main subject without a separate portrait/product mode distinction. The clearer the contrast between subject and background, the more accurate the result. Use the inpaint tool for fine adjustments.
Tips for natural-looking outpainting?
Specifying the style and color palette of the original image in the idea field during outpainting leads to more natural expansion. It is better to set the expansion direction step by step rather than all at once. The sharper and higher resolution the original image, the higher the quality of the expanded result.
6. Editor · Export
Subtitle/font usage guide (including Korean fonts)?
From the text menu in the editor, you can select from a variety of fonts including Korean fonts, and freely adjust size, color, alignment, and line spacing. Text styles and effects such as bold, italic, and underline are also supported. You can use the AI writing feature to automatically rewrite text or adjust tone.
What are the recommended resolution/frame rate/codec/bitrate?
The editor supports PDF, SVG, PNG, and JSON formats for download. PNG is recommended for image output and saves at the original resolution. Choose SVG for vector format or versatile sizing, or PDF when combining multiple pages into one file.
Tips for optimizing large file exports?
Complex layers and high-resolution images can increase export time. PNG is high quality but may result in large files; SVG is a vector format and remains sharp regardless of size. For multiple pages exported at once, PDF format is the most convenient option.
More FAQs