Aocus User Guide

A prompt on its own gives you text-to-image. Add a reference image and it becomes image-to-image: the model draws on the subject, composition, or style of your references.

1. 1Describe the image in the prompt box — the more specific you are about subject, setting, composition, lighting, and style, the better.
2. 2(Optional) Add reference images: click the Image button to the left of the prompt, or drag and drop or paste images straight into the box. PNG, JPEG, and WebP are supported.
3. 3(Optional) Type @ in the prompt and pick a reference, or click a reference thumbnail to insert it as @Image1.
4. 4At the bottom of the box, choose the Model, Resolution, and Ratio.
5. 5Click the generate button in the bottom-right corner (it shows the credit cost) and confirm.

New images appear in Latest results below the prompt box and are saved to the history panel on the left (on a phone, tap History next to the title). Each history card lets you:

• Click the image to open the preview: zoom in and out, view the original, and download it. Press Esc to close.
• Click the prompt to put it back in the prompt box, or use Copy prompt and expand it to read it in full.
• Use reference image: put this generated image back in the prompt box as a reference and keep iterating. Clicking one of the card's original reference thumbnails reuses that reference instead.
• Generate video: take this image to Video Studio as source material.
• Favorite, Download, Copy details, and Delete image. If a generation failed, click Retry.

Re-edit

Brings the prompt, references, and settings of this generation back into the prompt box so you can tweak and generate again.

Reuse prompt

Brings back only the prompt and leaves everything else as it is.

Text to video

Just a prompt, describing what's on screen, the action, the camera movement, and the pacing.

Image to video

Upload an image as the Start Frame, and optionally another as the End Frame; the video moves from one to the other.

Reference to video

Add several images, videos, and audio clips as references and use @ in the prompt to say what each one is for — e.g. “@Image1 mimics the moves in @Video1, with the voice from @Audio1”.

Models support different kinds of input, and the prompt box adapts to the model you pick: some show start/end frame slots, others show image / video / audio reference buttons, and some let you switch Mode between Frames and Reference.

1. 1Pick a Model at the bottom of the prompt box first — the box switches to the inputs that model supports.
2. 2Write your prompt and add start/end frames or references as needed. Click a reference thumbnail to insert its @ mention into the prompt.
3. 3Set the Size (landscape, portrait, and other ratios), Quality, and Duration.
4. 4Click the generate button and confirm. Videos usually take a few minutes; the history panel shows progress and the estimated wait.

• Click the cover to play a preview, watch it fullscreen, or copy the video ID.
• Remix: bring this video back into the prompt box as a reference video, edit the prompt, and generate again — useful for local edits, extending a shot, or swapping elements.
• Click the prompt to put it back in the prompt box; in the preview you can also use Re-edit or Reuse prompt.
• Download, Favorite, Copy details, and Delete. Failed jobs show the reason; click Retry to try again.

• Start with the subject and setting, then the action, then the camera language (shot size, movement, pacing).
• When referencing several files, spell out each one's role: who provides the character's look, what provides the motion, what provides the setting or music.
• For continuous shots, use Remix to extend from the previous clip.

Generations in Video and Image Studio are filed by workspace. Click the workspace name at the top of the history panel to:

• search for and switch to another workspace;
• create a New Workspace for a new project;
• click the ··· next to a workspace to Pin, Rename, or Delete it.

• Switch between Detail list and Grid view.
• Scroll down to load older items automatically.
• On a phone, tap History next to the title. A green dot on the button means something is still generating.

1. 1Click Canvas in the sidebar to see your canvas projects.
2. 2Click New project and give it a name, or open an existing one. Each project is its own canvas and opens in a new tab.
3. 3From the project list you can Pin, Rename, or Delete a project.

Each piece of content on the canvas is a node — either an image node or a video node.

1. 1Click Add image node or Add video node in the left toolbar, or right-click an empty spot to add one.
2. 2Write a prompt in the box under the node, pick a model and ratio (and a duration for video), then generate. Click Expand for a larger editing area.
3. 3When it's done, the result appears right in the node.

You can also Upload local images or videos from the toolbar (videos up to 15 seconds), or open the History panel and drag past generations onto the canvas.

A connection means “reference”: the upstream node's content is used as reference material when the downstream node generates.

• Drag the + on a node's edge onto another node to connect it as a reference.
• Click the +, or drag it onto empty space, to open the Reference from this node menu: choose Image or Video to create a new node that references this one.
• Video nodes also offer Extract first frame, Extract last frame, and Extract current frame, turning a frame into a new image node.
• Type @ in a prompt and pick @Image1, @Video1, and so on to say how each reference should be used.
• Hover over a connection to reveal the scissors, then click to cut it.

Select a node that has content and a toolbar appears above it:

Multi-angle

(Image) Reshoot the same frame from a new camera position. Pick a preset (eye-level medium, 3/4 oblique, hero low, over-the-shoulder, high overlook, trailing back), or use the sphere and sliders to set Shot scale, Orbit, and Height, plus Dutch tilt under Advanced.

Lighting

(Image) Relight the frame. Adjust global Brightness and Color, drag the Key light direction, turn on Rim light, or apply presets such as Rembrandt, Butterfly, Golden Hour, and Cyberpunk Neon. In Smart mode, describe the light or mood in a sentence and optionally upload a lighting Reference image.

Annotate

(Image) Mark up the image with Pen, Rectangle, and Text. Saving creates a new annotated image node you can use as a reference to show the model what to change.

Extract frames

(Video) Capture the first, last, or currently playing frame as an image node — handy for continuing into the next shot.

Download / Fullscreen

Download the original file or view it fullscreen.

The director stage is a 3D blocking tool: position characters and cameras before you generate, and capture storyboard frames with exactly the composition you want.

1. 1Click Add director node in the left toolbar, then click Open director stage on the node (or double-click it).
2. 2Use Add to place a Character, Props & set pieces (tables, chairs, doors, windows, trees, cars, and more), and a Camera.
3. 3Select an object and Move, Rotate, or Scale it. Characters can take pose presets such as stand, walk, run, or sit under Pose, with fine control over head, torso, arms, and legs in Pose adjust.
4. 4Select a camera to set its lens, heading, pitch, and roll — the bottom-right preview updates live. Turn on Thirds and Safe overlays to help framing, and switch between 3D and Top views.
5. 5Click Capture shot to save what the camera sees, then Add to canvas or Download it from the shot list.

• Left-drag on empty space to box-select several nodes; drag a node to move it.
• Pan with a two-finger swipe on a trackpad or by dragging with the middle mouse button. Zoom with a pinch, or hold Ctrl while scrolling.
• The bottom-left controls include Pan canvas (left-drag pans while it's on), Auto-arrange (then choose Keep or Revert), Snap to grid, Minimap, Reset zoom, Zoom in, and Zoom out.

Drama Agent requires an active subscription. The flow is: new drama → Script Outline → Asset Library → Episode Videos. Use the step bar at the top to move between steps you've already reached.

1. 1Click Drama in the sidebar, then New Drama under My Projects.
2. 2Choose Upload script (txt / docx, up to 100,000 characters, drag and drop supported) or Paste text (50–100,000 characters).
3. 3Set the Aspect Ratio (9:16 portrait or 16:9 landscape) and a Style Preset: Western, Chinese, Chinese Period, Japan/Korea, or Southeast Asia live-action cinematic, or Cyberpunk, Japanese Anime, or Chinese Donghua.
4. 4Click Confirm & Analyze. The analysis cost is shown next to the button.

The agent reads the whole script and extracts characters, scenes, and episode structure. Short scripts usually take about a minute; long ones can take 10–15 minutes. You can leave the page and come back — the outline appears when it's ready. If analysis fails, click Retry analysis; one retry usually does it.

The outline shows the Logline, Full outline, Characters, Scenes, and Episodes, along with the aspect ratio, resolution, and episode strategy (Explicit episodes from script or AI auto-segmentation). When it looks right, click Next: Asset Library. To start over with a different script, click Re-upload.

The Asset Library lists every character and scene in the series. Their generated images anchor the look of every later shot, keeping people and places consistent.

• Generate All creates every character and scene image in one go; use Regenerate on a single card to redo one.
• Edit a character: change the Name, Character description, and Voice, and optionally upload a reference image (jpg / png / webp / gif). Click Save & regenerate to update the look.
• Add look: a different appearance for the same character in other parts of the story. Costume keeps the face; Phase allows the face, hair, and eyes to change. You can pick a Reference look and fill in Applies to episodes (e.g. 1, 2, 5-7).
• Edit a scene: change its Name, Description, and Atmosphere.
• Add a scene state: another version of the same place — intact → ruined, day → rainy night, empty room → wedding décor. The new state uses the main scene image as a reference so it's recognizably the same location.
• Versions: every regeneration is kept. Browse them in the version strip and click Adopt this version to make one current.

When you're happy with the assets, click Next: Storyboard to move on to the episode videos.

The episode list shows how many segments each episode has and how far along they are.

1. 1Click Generate Storyboard on an episode. The agent splits it into segments and shots (the credit cost is shown on the page).
2. 2When the storyboard is ready, click Open Editor to open the episode editor — or click Generate N to render all of the episode's pending segments at once.

• The left column shows the characters and scenes in this episode; click one to edit it. Click + to add a character or scene from the Asset Library so you can reference it in the script.
• The middle column holds the segments and their scripts. Click Edit Script to change a segment: type @ to reference characters and scenes and to add a duration to each shot, and pick the previous segment's last frame as a reference for smoother transitions.
• Each segment must total 4–15 seconds before you can Save Script.
• Click Generate Video to render the current segment (billed per second of video; the rate is shown), or Generate Videos to render every segment at once. Not happy? Click Regenerate.
• In the preview on the right, switch between versions, click Use this version to choose the one that goes into the final cut, or Download this version.

Compose All

Available once every segment in an episode is done. Joins the adopted version of each segment in order and downloads a single MP4. If segment formats don't match, you get a ZIP of the segments instead.

Export Subtitles

Exports the episode's dialogue and narration as a subtitle file (.srt).

• Tabs across the top: Image, Video, Canvas, and Drama, each with a count.
• Images and videos are grouped by date and load as you scroll. Click any item to preview it.
• Hover over a card to Add to favorites or Remove favorite.
• Click Select to pick several items, then Favorite or Delete them together.
• The Canvas and Drama tabs let you open, rename, pin, or delete those projects.

Plans differ in monthly credits, how many jobs can run at once, how long files are kept, and whether output is watermarked. The pricing page has the details for each plan.

• Subscriptions don't renew automatically. They end when the term is up, with no further charges.
• Subscription credits are valid for 30 days.
• You can upgrade at any time; to downgrade, wait until your current plan ends.

• Top up with a credit pack from the Buy credits tab of the pricing dialog.
• Buying credits requires an active subscription. Purchased credits are valid for 365 days.
• After paying you'll land on a confirmation page. If it says the payment is still syncing, wait a moment and click Retry check.

1. 1Click your avatar to open the user menu and choose Redeem.
2. 2Enter the code and click Confirm redeem. A code can grant a subscription, a credit pack, or credits.

• Click the credit number at the bottom of the sidebar (or Credits in the user menu) to see your credits overview, when each batch of credits expires, and All, Spend, and Gain records.
• Starting a generation charges the amount shown on the button. If it fails, the credits are refunded automatically and a matching refund record appears.
• Purchased and redeemed credits need an active subscription. When your subscription ends they show as “Locked”, and they unlock again when you resubscribe (as long as they haven't expired).

Upgrade / Current plan

See your current plan and open the pricing page.

Credits

Your balance and credit history.

Redeem

Enter a redeem code.

Contact us

Email addresses for support and partnerships.

Language

Switch between 中文 and English. When signed out, you can also switch at the bottom of the home page.

Generation confirmation

When on, a confirmation dialog appears before every generation. It's on by default; you can also switch it off from the dialog itself.

Theme

System, Light, or Dark.

Sign out

Sign out of your account.

• When you see “A new version of Aocus is available”, click Refresh to get the latest version.
• If an image or video fails to load, click Tap to retry to load it again.

No. When a generation fails, its credits are refunded automatically — you'll see the refund in your credit history.

Usually the prompt is empty, or the selected model doesn't support that combination of settings. With start/end frames, you need at least a start frame, and you can't mix frame inputs with general reference media.

The prompt or reference media may contain sensitive or copyrighted content, or photos of real people. Adjust the wording or swap the media and try again.

Usually a few minutes; each card shows an estimated wait. If you submit more jobs than your plan runs at once, the extra ones queue.

How long files are kept depends on your plan, and download links expire. Download anything important to your own device.

Yes. Image and video generation and the Asset Library work in mobile browsers. Canvas and the drama editor involve a lot of dragging and multi-panel work, so we recommend a computer for those.

Support (account, billing, bug reports): support@aocus.ai. Partnerships: contact@aocus.ai. We usually reply within 1–2 business days.