Workspace
Summary: The XBRUSH Workspace is the main creation screen where you generate images, videos, and audio, organize results in folders, and manage your assets. Access it by clicking Workspace in the top navigation bar.
ℹ️ About engine names in this guide. XBRUSH adds and retires generation engines frequently. Any engine name below is the default at the time of writing — the engine selector on screen is authoritative, and the home page lists the models currently offered.
Workspace Layout
A session in the workspace (xbrush.ai/workspace). The tab bar shows Home and the open session tab; your request, the agent's reply (with its plan and estimated cost), and the result card (Generate #4) flow like a conversation, with the input box — Chat · Direct · Attach, Send · New line · @ Pick character — pinned to the bottom.
The workspace consists of three areas.
| Area | Location | Role |
|---|---|---|
| Tab Bar | Below the header | Folder tree icon · Home tab · session tabs (close with ×) · + new session · Search |
| Task Feed | Center | Shows your requests, the agent's replies, and generated result cards in time order |
| Input Box | Bottom | Enter prompts, switch between Chat / Direct, Attach files, and send |
- In Chat mode, describe what you want and the agent picks the engine and options for you.
- Turn on Direct mode (Ctrl+/) to open a settings panel inside the input box, where you choose the model and options yourself. Generating from Direct mode starts a new session — the panel says so ("Generating starts a new session").
In Direct mode, the bottom bar of the input box shows:
| Item | What it does |
|---|---|
| Agent (robot icon) button | Hands the request to the AI agent instead of running the panel settings yourself |
| ↵ Run · ⇧↵ New line | Press Enter to run, Shift+Enter for a new line |
| Generate ◆ 5.2 | The run button, with the estimated credit cost of the current settings (the number changes with engine, count, resolution, and so on) |
In Workspace, a "task" and a "session" are the same thing — one conversation and everything it produced. See Create and Edit in Plain Language.
Content Type Tabs
Select Image / Video / Audio at the top of the Direct panel.
- The sub-tabs to the right and the settings below change based on the selected type.
- The panel remembers the sub-tab you opened last. Close it with the × at the far right; the ⓘ icon next to it shows help for the current tab.
- Results of every type appear as result cards in the task feed.
Image Tab — Generate
Click Direct in the input box and choose Image › Generate. The panel is a compact strip inside the input box: the category and sub-tab row on top, the prompt line ("Describe what you want to create."), the option chips, and a bottom row with Chat | Direct, "Generating starts a new session", the agent (robot) button, Run · New line, and Generate with its credit cost.
The Generate option row reads, from left to right:
+ Edit from image · GPT Image 2.5 Flare · 3:4 · Count 1 2 3 4 · 1K · Quality Medium · Settings
Each item is described below.
AI Engine
- Select the AI engine to use. Default: GPT Image 2.5 Flare
- Clicking the engine button opens the engine list, which shows the credit cost of each engine.
Size
- Select the aspect ratio for the generated image (default 3:4).
Number of Images
- Choose how many images to generate at once with Count: 1–4 (default 1).
- Higher numbers consume more credits.
+ Edit from image (optional)
- Click + Edit from image to add a reference image, and the AI will use it as a visual reference during generation. (Older screens labeled this "Reference image".)
Resolution, Quality, and Transparent Background
Resolution (1K) and Quality sit on the option row. Click Settings to open the popover with Transparent background — "Outputs an RGBA PNG with no background".
| Option | Description |
|---|---|
| Resolution | Choose 1K / 2K / 4K. Default 1K. |
| Quality | Choose Low / Medium / High / Extra high / Max. Default Medium. |
| Transparent background | Under Settings. Outputs an RGBA PNG with no background (no extra charge). |
Image Tab — Edit
The Edit tab lets you modify existing images using AI. Options, left to right: Source Image · required · Reference sketch · GPT Image 2.5 Sunburst · Count · 1:1 · 1K · Quality High · Settings; the action button is Apply.
How to Use
- Click + Source Image · required at the start of the option row, or drag an image into the panel.
- Enter what you want to change in the prompt input.
- Click the Apply button.
Key Options
| Option | Description |
|---|---|
| Reference sketch | Upload a sketch image or draw directly with the built-in drawing tool to guide the edit |
| AI Engine | Default GPT Image 2.5 Sunburst |
| Count | Choose 1–4 (default 1) |
| Aspect ratio | Shown as 1:1 by default; choose another ratio if needed |
| Resolution | Choose 1K, 2K, or 4K |
| Quality | Choose Low / Medium / High / Extra high / Max (default High) |
| Settings | Extra options for the edit |
Image Tab — Outpaint
The Outpaint tab lets you expand the area outside an image using AI. Options: Source Image · required · FLUX.2 Pro · Count; the prompt line reads "Describe what you want to add." and the action button is Generate.
How to Use
- Upload an image with + Source Image · required; the expansion editor opens automatically.
- Choose a ratio preset and 1K/2K, adjust the expansion frame, then click ✓ Confirm. To adjust it again, click Edit in the panel.
- Describe what should appear in the expanded area using a prompt, then click Generate.
Key Options
| Option | Description |
|---|---|
| Edit | Reopen the expansion editor to adjust the area |
| Reset Outpaint | Reset the expansion area |
| Clear All | Remove the uploaded image and expansion settings |
| AI Engine | Default FLUX.2 Pro |
| Count | 1–4 (default 1) |
Image Tab — Sub-tabs
The Image type now has six sub-tabs: Generate · Edit · Outpaint · Remove BG · Upscale · Layer Split. Layer Split is shown directly in the tab row — there is no More menu.
| Tab | What it does | Documented in |
|---|---|---|
| Remove BG | Removes the background automatically | Below on this page |
| Upscale | Raises the resolution of an image by a chosen factor | Below on this page |
| Layer Split | Splits an image into separate editable layers | Below on this page |
ℹ️ The HD Upscale and Object Removal tabs are no longer in the Direct panel. To raise resolution, use Upscale. To erase something from an image, ask the agent in plain language (e.g. "remove the person in the background") or use the Edit tab.
Image Tab — Remove BG
The Remove BG tab has no prompt line and a single option, Source Image · required; the action button is Proceed.
How to Use
- Upload an image with + Source Image · required.
- Click Proceed to remove the background automatically.
- The result appears in the task feed for download or further use.
Image Tab — Upscale
The Upscale tab raises the resolution of an uploaded image. Options: Source Image · required · RealESRGAN_x4plus (upscaler) · 2× (scale); the action button is Proceed.
How to Use
- Upload an image with + Source Image · required.
- Check the upscaler (default RealESRGAN_x4plus) and the scale (default 2×); click either chip to choose another upscaler or scale.
- Click Proceed to generate a higher-resolution version.
Edit in Canvas
Opens the selected image in the layer editor, where you can add text, shapes, and effects and arrange them as layers by hand.
How to Use
- In the task feed, open the ⋮ menu on the result card of the image you want to edit.
- Choose Edit in Canvas.
- The layer editor opens with the image placed as the base layer.
For the full set of editor tools, see the Editor guide.
Image Tab — Layer Split
Splits an image into separate layers — subject, background, and other distinguishable elements — so each can be moved or replaced independently. Options: an optional prompt ("(optional) what to split — e.g. person, logo · empty = automatic"), Source Image · required, Seedream 5.0 Pro (with a −10% badge), and 1K; the action button is Split layers.
How to Use
- Choose the Layer Split sub-tab.
- Upload an image with + Source Image · required.
- Optionally type what to split out (e.g. person, logo). Leave it empty to split the whole image automatically.
- Click Split layers. The separated layers appear in the task feed, ready to take into the layer editor.
💡 Layer Split is the step before compositing. Split first, then use Edit in Canvas to rearrange or replace what you pulled apart.
Video Tab
Video › Generate creates videos from a prompt plus optional reference images, videos, and audio. Sub-tabs: Generate · LARP · AI Edit · Upscale. Options: Seedance 2.5 · Reference images 0/30 · Reference videos 0/10 · Reference audio 0/10 · 16:9 · 720p · 5s · Draft 480p (off) · Audio (on) · Settings.
The Video type has four sub-tabs: Generate · LARP · AI Edit · Upscale.
With the default settings, Generate is set to Seedance 2.5 · 16:9 · 720p · 5 seconds · Draft 480p off · Audio on, with reference image 0/30, reference video 0/10, and reference audio 0/10 — about 454C per run at these defaults.
Key Options (Generate)
| Option | Description |
|---|---|
| AI Engine | The engine preselected in the panel is the current default. Engine names and versions change often, so read the selector rather than this table — the home page lists the models currently offered. |
| Aspect Ratio | Auto / 21:9 / 16:9 / 4:3 / 1:1 / 3:4 / 9:16 |
| Video Length | Slider in seconds, default 5 seconds |
| Draft mode (480p) | Turn on to generate a quick 480p draft (off by default) |
| Audio generation | Choose whether to generate audio along with the video (on by default) |
| Reference images | Up to 30 (optional). The model can generate from the prompt alone, without an image. |
| Reference videos / Reference audio | Up to 10 each (optional) |
| Settings | Extra video options |
| Resolution | Choose 480p / 720p / 1080p |
- Describe the scene and the motion you want in the prompt input.
- Point to references with
@. Attached references get names such as @Image 1 and @Image 2. Use those names in the prompt to say which reference plays which role — e.g. "@Image 1 and @Image 2 are taking a selfie together". - For more details, see the Video Generation page.
Video Tab — LARP
Same video, your character as the lead. Video › LARP: Mode (Swap the lead | Restyle) · Source video 0/1 · Character photos 0/3 · Quality Best · Matches source · 720p · Keep original audio · + sheet 45 · if needed, with the notice "Use only videos and photos of yourself or people who agreed." and the Make video button.
LARP (formerly Recast) takes a source video and remakes it with your character in the lead role, keeping the original shots and motion.
- In the Direct panel, choose Video › LARP.
- Choose the Mode: Swap the lead replaces the main person with your character; Restyle redraws the video in a different style.
- Add the video to follow with + Source video (1 video).
- Add up to 3 photos of your character (or the person you want in the lead) with + Character photos.
- Optionally describe changes in the prompt line ("Describe changes (optional)").
- Check the remaining chips — Quality Best, Matches source · 720p (the output follows the source video), Keep original audio, and + sheet 45 · if needed (a character sheet is added only when needed) — then click Make video.
⚠️ Use only videos and photos of yourself or of people who have agreed to it.
You can also start LARP from the LARP row on the Workspace home or open it directly at https://xbrush.ai/workspace?open=video-recast — see LARP.
Video Tab — AI Edit
Video › AI Edit changes an existing video from a written instruction. Options: Source Video · required · Gemini Omni 1.1 Flash · Keep original (keeps the original audio); the action button is AI Edit.
- Choose Video › AI Edit and upload a video with + Source Video · required.
- Describe what to change in the prompt line ("Describe what to change.").
- Click AI Edit. The edited video appears in the task feed.
To raise a video's resolution instead, use Video › Upscale: Source Video · required · RealESRGAN · 2×, then Proceed.
Audio Tab
The Audio tab generates voices, music, sound effects, and lip sync. Sub-tabs: Voice · Music · SFX · Lip Sync. Shown here is Voice with the Gemini engine and the default voice 광고 보이스오버 1; the prompt line reads "Enter text to convert to speech."
The Audio tab has four sub-tabs.
| Sub-tab | Description |
|---|---|
| Voice | Convert text to speech (TTS) — Gemini · voice 광고 보이스오버 1 |
| Music | AI music generation — Reference Image · Lyria 3 · Settings |
| SFX | AI sound effect generation — From Video | From Text · Source Video · required · Pixverse Sound Effects |
| Lip Sync | Apply voice lip sync to an image — Source Image · required · Voice audio · required · InfiniteTalk · 480p |
Voice Tab — Voice Selection
- The default AI engine is Gemini, and the default voice is 광고 보이스오버 1 (a Korean-named voice, "ad voiceover 1"; the name stays in Korean even in the English UI).
- The voice list and the options you can adjust depend on the selected engine. When you switch engines, the list and options change with it — check the panel for what is available.
- For more details, see the Audio Generation page.
Folder Structure
Click the folder tree icon at the left of the tab bar to open the folder popover.
- The top row shows the current folder (e.g., Home · Tasks 10 · Subfolders 2) and a + button.
- The tree below shows the number of tasks in each folder; click ▸ to expand subfolders.
- You can also create a folder with + New folder next to the breadcrumb on the Home tab, or with the New folder card in the Folders section.
Creating a Folder
- Click the + button at the top of the popover to display a folder name field inside the tree.
- Enter a name and press Enter to create the folder.
Creating a Subfolder
- Hover over a folder row to reveal the ⋮ button on the right.
- Choose New subfolder from the ⋮ menu to add a new folder inside that folder.
Renaming a Folder
- Choose Rename from the folder row's ⋮ menu.
- The row turns into a text field — enter the new name.
Moving a Folder
- Choose Move… from the folder row's ⋮ menu to open the Move 'folder name' dialog.
- Pick the destination in the tree. If needed, use + New folder to create a new folder there.
- Click Move here.
Deleting a Folder
- Choose Delete from the folder row's ⋮ menu.
- In the Delete Folder dialog, type the folder name to confirm and click Confirm.
- Deleting a folder moves its subfolders and their tasks to Trash together. The folder's library assets move to the team home library.
Asset Management
Content generated at once appears in the task feed as a result card.
- The bottom row of each card shows the label (Generate/Edit) · number (e.g., #4) · runner avatar · ▾ · 👍 · 👎 · 💬 comments · time · ⋮ (card menu).
- Expand ▾ to see the AI engine, prompt, and reference image used.
- Hovering over an image shows Copy, Download, and ★ (Star) icons at the top right, ↳ (Add as reference) at the bottom left, and ⋮ (image menu) at the bottom right.
Expanding ▾ shows the settings used for the generation.
A result card has two menus. Actions on a single image live in the ⋮ on the image; actions on the whole card live in the ⋮ in the card's bottom row. See Result Card ⋮ Menu below.
Task Comments
Members can leave comments on individual generation results within the team workspace. This is useful for providing feedback or notes on specific outputs without leaving the platform.
Leaving a comment on a result card
- Click the 💬 comment icon in the bottom row of a result card to open the comment popover.
- Type in the "Write a comment…" field and click the send button to post.
- All team members can view and reply to comments on shared tasks.
Sharing Settings
- Sharing is done at the team level in Team Settings.
- See the Team Management page for details on inviting members and managing roles.
Result Card ⋮ Menu
The image menu opens when you hover over an image and click ⋮ at its bottom right. Hover over Use to open the follow-up actions.
| Image menu | Description |
|---|---|
| Use › | Upscale · Remove BG · Animate · Edit in Canvas (opens the editor) · Filter |
| Add as reference | Puts this image into the input box as a reference (same as the ↳ button at the image's bottom left) |
| Download | Saves the image file |
| Share | Creates a public link for the result. Only available when the team has turned on Allow public result sharing (see Team) |
| Star | Marks the result with a star |
| Save to Library | Saves it to the library |
| Delete | Deletes this image |
The card menu opens from the ⋮ at the right end of the card's bottom row.
| Card menu | Description |
|---|---|
| Copy Settings | Copies the settings used for this generation |
| API Code | Shows an example request (curl) that reproduces this result through the Public API. Replace YOUR_API_KEY with your team API key; Copy and View API docs are available. See API |
| Move To · Duplicate | Moves the result to another folder or duplicates it |
| Compress | Compresses the card's results |
| Select multiple | Selects several results to act on at once |
| Delete Contents | Deletes the card's results |
Image detail view
Click a result image to open the detail view. The image fills the center (arrows to move between images, zoom controls at the bottom), and the right-hand panel holds the details and actions.
- Details — label (Generate/Edit), runner, Model · Resolution · Created At, Prompt (Copy Prompt, Show more)
- Use — Upscale · Remove BG · Animate · Edit in Canvas · Filter · Copy Settings
- Send — Add as reference · Download · Copy Image · Share · Publish to SNS (see Social Media)
- Organize — Star · Save to Library · Delete
- Shortcuts — arrows move · S star · D download · R reference · Esc close
Next Steps
- Image Generation — Detailed image generation guide
- Video Generation — Detailed video generation guide
- Audio Generation — Detailed audio generation guide