Project Workspace

The project detail page is the main production surface. It combines preview, prompt controls, asset selection, generation history, and project-level actions.

Prompt help: For surface-specific prompt examples for image, vector, video, voice, music, InvokeAI, smart resize, and composites, read the Prompt Writing Guide.

Workspace Layout

AreaPurpose
Center previewShows the selected or primary image, vector, video, audio, or output result. Video opens in a stable modal/player when previewed.
Prompt dockEnter the user prompt, select output type, choose format, attach assets, choose voice/music/video options, and generate.
Right inspectorShows project properties, brand kit, typography/palette/aspect context, quality gate context, and export/settings actions.
Bottom railTabs for Assets, Timeline, Variants, History, and Uploads/Camera.

Prompt Dock Output Types

The prompt dock has five primary output types:

OutputUse it forImportant constraints
ImageRaster images, social posts, posters, hero graphics, composites with logos/photos.Accepts image and vector/SVG assets. Does not accept audio or video assets in image mode.
VectorLogo marks, simple icons, and badges.Prompt-only in the current build. Remove attached assets before generating vector output.
MusicStandalone instrumental background music using Audiocraft.Creates reusable project audio. Use it before a video if you want a music bed.
VoicePiper voiceover from script text.Script-only in the current build. Remove attachments before generating voice.
VideoSlideshow videos, title-card videos, and composed video variations.Video rules depend on selected video mode. See Video, Voice & Music.

Image Generation

Use image mode for static raster output and composites. Available modes include freeform, social, poster, and hero-style images.

  1. Select Image.
  2. Choose image mode.
  3. If using social mode, choose a social format or let the prompt resolve one.
  4. Attach compatible image or vector assets.
  5. Write a simple prompt such as Overlay the logo on the beach photo and create an Instagram launch post.
  6. Generate and preview the result.

Image Social Formats

Image mode supports platform-aware social dimensions:

Instagram Square 1080x1080 Instagram Portrait 1080x1350 Instagram Story 1080x1920 Instagram Reel Cover 1080x1920 Facebook Feed 1200x630 Facebook Cover 1640x924 LinkedIn Feed 1200x627 LinkedIn Banner 1584x396 X Landscape 1600x900 YouTube Thumbnail 1280x720 YouTube Channel Art 2560x1440 TikTok Cover 1080x1920 Pinterest Pin 1000x1500

Vector Generation

Use vector mode for simple logo marks, icons, and badges. Current vector mode is prompt-only. Do not attach images, videos, or audio when generating vectors.

Brand Kit Selection

The project may inherit a brand kit from its campaign. The prompt dock can also use a generation-level override. Use overrides carefully.

Asset Rail Tabs

TabUse it for
AssetsAll project assets. Preview, download, delete, use, or refine eligible assets.
TimelineVideo composition inputs and timeline-oriented work.
VariantsGenerated variants, including InvokeAI refinement outputs where available.
HistoryPast jobs and generation state.
Uploads/CameraProject uploads and campaign/project-associated phone camera assets.

Export And Download Actions

The project workspace exposes real export actions. These actions download files through authenticated Studio routes, so stay signed in while using them.

ActionWhere it appearsWhat it does
Download assetsProject header and Export settings modalDownloads a ZIP containing the current project assets.
Export project CSVProject header and Export settings modalDownloads a CSV manifest of project assets and job metadata.
Download selected assetPreview area and Export settings modal when an asset is selectedDownloads the selected asset file.
Export settingsInspector panelOpens a modal with supported export/download actions. It does not show unsupported fake toggles.

Resize Workflow

Resize opens a target picker. It does not simply scroll to the prompt box. Choose one or more target sizes, then generate resized assets through the Studio job system.

The deployed 2026-06-29 build keeps the preview action bar above passive video previews, so Resize, Preview, and export actions remain clickable even when a video is selected in the central workspace.

Timeline Workflow

The Timeline tab is a functional MLT-backed video editor for project assets. It can assemble multiple visual, logo, text, caption, music, and voice assets into a rendered MP4, and it exposes split, duplicate, trim, transition, transform, crop, opacity, keyframe, audio mix, caption, and export profile controls.

  1. Open the Assets tab.
  2. Click Timeline on each asset you want to use. For a complete video proof, add a visual clip or image, a logo asset, music, and voiceover.
  3. Open the Timeline tab.
  4. Add text and captions where needed.
  5. Check render readiness. Required features such as visual scene, logo overlay, text/CTA, captions, music bed, and voiceover must show OK.
  6. Click Render timeline.
  7. Review the generated final video asset in the Assets tab.
Verified production behavior: The Itskatzu TikTok/Reel project has been tested on the deployed app with video, logo overlay, text/CTA, captions, music, and voiceover. The rendered output completed as a 1080 x 1920 MP4 with an audio stream and an Immich-backed final video asset.

InvokeAI Image Editing

Eligible image and vector assets can show a native InvokeAI editor action. Use it for image polish, variants, painted-mask inpaint, or source-protected outpaint when the runtime capabilities are enabled.

  1. Select an image or vector asset.
  2. Choose Refine with InvokeAI or the visible InvokeAI editor action where available.
  3. Use the canvas controls: source, mask, generated, and accepted layers.
  4. For inpaint, paint the area to edit with the brush or erase tools, then run inpaint.
  5. For outpaint, expand the canvas with the outpaint handles, then run outpaint.
  6. Compare before/after results and accept the variant you want to reuse in the project or campaign.
What InvokeAI edits: Secant chooses the source asset, mask/canvas, prompt packet, negative prompt, and acceptance rules. InvokeAI performs the generative image edit. Video editing stays in the MLT timeline editor.