Skip to content
Studio · ~5 minutes

Small things that save time and money

Not big features but Studio's small habits: say a prompt by voice, see the exact price before you click, get a frame description for free, understand how long to wait and how many generations you can run at once. This page collects what Studio has now and is confirmed by the code.

Updated:

In short
  • A microphone in the video and image composers: “Dictate the prompt” — free, your browser does the recognition; in browsers without recognition there is no button.
  • Paid actions (4K, “Frame analysis”, “Take style”, “Remove background”, editing a long clip) first open a window with the price, your balance and the action button; until the price arrives the button is disabled.
  • “Get a prompt from the frame” is free; the limit is 30 times per hour and 120 per day.
  • The “Generate” button shows the price and the time (“~3 min”); Ctrl+Enter starts a generation.
  • You can keep at most 3 unfinished generations at a time.

1Prompt dictation

In the video composer (and the image composer), in the row of icons, there is a microphone. The tooltip on hover is “Dictate the prompt”. Click, speak, and the recognised text is appended to the field after a space, to what is already written. The same button stops it, now with the tooltip “Stop dictation”.

  • Free. This is the browser's own speech recognition: no tokens are spent and no sound is sent to Gibrid servers.
  • The language follows the interface language: Russian, English, Spanish, Indonesian or Arabic.
  • Where there is no recognition (for example Firefox and some Safari versions) the button is not drawn at all — Studio does not promise what will not work.
  • The microphone: the browser asks once for permission; if you refused, you get “No microphone access — allow it in your browser settings”.
  • If the speech was not recognised you get “Couldn’t recognise the speech — please try again”.
TipDictate sentence by sentence and re-read the field: recognition sometimes confuses names and titles, and fixing the text by hand is faster than regenerating a clip.

2A window with the price instead of “paid operation”

Before, the browser asked just “This is a paid operation — continue?” before paid actions. Now Studio shows a window with the sum that the server calculates with the same functions as the charge. In the window: a Price row, for clips a Clip length row, a Balance row and the buttons Cancel and the action (for example Upscale).

The window opens for these actions:

ActionWindow title and buttonHow the price is calculated
Image to 4K (menu “Upscale to 4K (paid)”)“Upscale to 4K” → the “Upscale” buttonBy the result size: up to 2048 pixels on the long side — 3 tokens, up to 4096 — 6, more — 12.
Clip to 4K (menu “Enhance to 4K (paid)”)“Upscale the clip to 4K” → the “Upscale” buttonBy clip length: about 3.6 tokens per second (a 10 s clip — about 36 tokens, 30 s — about 109).
“Frame analysis”“Frame analysis” → the “Review” buttonAn upper estimate: “less than 1 token” or “up to N”; charged by actual use, usually less.
“Take the style of this frame”“Style of this frame” → the “Take style” buttonThe same: an upper estimate by the model's use.
Editing a clip longer than 15 s“Editing a long clip” → the “Continue to edit” buttonThe sum for the whole clip, noticeably more than an ordinary edit — more on the clip editing page.

Until the server returns the price the action button is disabled: Studio no longer allows “paid, but we won't say how much”. If the price could not be obtained the window says “Couldn’t get the price — check your connection and try again”. If the balance is not enough you see “Insufficient balance” and the button is disabled.

Remove background in the image edit mode works the same way: a “Remove background” window with the note “We’ll cut out the subject and make the background transparent”, the price (“less than 1 token” — a flat $0.02) and your balance. More on the quick edits page.

⚡The sums are by the server's calculation today and may change; the right figure is the one written in the window. 1 token ≈ $0.025 and your balance is in the Studio header. More about tokens and top-ups on the Billing and plans page.

3“Prompt from frame” — free

For any finished image or clip you can get a frame description: a neural network looks at it and writes a prompt that will reproduce a similar frame. In the work's menu this is the item Get a prompt from the frame (marked “free”), in the viewer panel the button Get prompt. While it reads you see “Reading the frame…”, then the prompt goes into the composer field and the message “Prompt added to the prompt bar” arrives.

  • No charge and no price window.
  • The limit: 30 times per hour and 120 times per day per person — protection from abuse.
  • Only this kind of analysis is free. “Frame analysis” (through a director's eyes) and “Take style” stay paid but very cheap — the window shows their price.

4Price and time on the “Generate” button

The video launch button shows the price in tokens, calculated by the server for your current settings (model, duration, quality, frame, sound), and next to it the time, for example “~3 min”. The button's tooltip: “Usually ready in ~3 min · Ctrl+Enter”. The Ctrl+Enter shortcut (Cmd+Enter on a Mac) starts a generation without reaching for the mouse.

The time is the median over our clips in 30 days; on the card of a running generation it turns into “~1 min left” and “Almost done”. Details are on the page about the “Done” signal. Choosing a suitable model by price and time is helped by picking by goal.

5No more than 3 generations at once

A person can have at most three unfinished generations — waiting or running (videos and images are counted together). A fourth is not accepted: the server answers “Currently 3 of your generations are running — that is the limit. Wait for any of them and you can start the next one”. As soon as any finishes you can start more. (The text of this message comes from the server, in Russian.)

  • A “reel from photos” counts as the whole limit while it is being assembled: its frames go out three at a time and it takes all three places.
  • Series — “Compare”, “Animate selected” and chains — send clips one by one and run into the same limit; whatever did not get out is not paid for.
  • A clip that is already running is counted once (not twice): before, a second send in a row could be refused because of this; now it is fixed.

?FAQ

Why don't I see the microphone in the composer?

The dictation button appears only in browsers with built-in speech recognition (for example Chrome, Edge, newer Safari). Firefox does not have it, and Studio does not draw a button that will not work.

Is dictation paid?

No. Your browser recognises the speech, no tokens are spent and no sound is sent to Gibrid servers.

How much does upscaling to 4K cost?

The price is shown in the window before you run it. For images it depends on the result size (3, 6 or 12 tokens), for clips on the length (about 3.6 tokens per second).

What does “Charged by actual use, usually less than shown” mean?

For actions where the use is not known in advance (“Frame analysis”, “Take style”) the window shows an upper estimate. The real charge is calculated from actual use and is usually less.

Why can't I start a fourth generation?

The limit is 3 unfinished generations at a time. Wait for any to finish and start the next.

Is getting a prompt from a frame paid?

No, it is free: the limit is 30 times per hour and 120 per day. “Frame analysis” and “Take style” stay paid; the window shows their price.

Related

Open Studio →