AI Studio makes images, video, talking clips, speech, and music for you — right here on this computer, with no uploads and no per-image fees. You write what you want in plain words, press a button, and the result appears in your library a little later. Some photo jobs are sometimes handed to a second computer to finish faster — there is nothing for you to do, and the results, prices, and library are exactly the same either way.

1. Open the AI Studio address in your browser (your operator gives it to you). You will see the sign-in screen.
2. Enter your email and password, then press Sign in. New here? Press Create account, choose a password, and click the confirmation link we email you — the account activates the moment you click it.
3. After signing in you land on the Dashboard: your recent jobs, recent creations, and shortcuts. The menu on the left is where everything happens.

The left menu lists the studios. Each one works the same way: pick a tab, fill in a description (and sometimes a photo or audio clip), press the big button, and watch the progress card on the page or the Jobs page. Every finished creation also appears in that studio's Library tab.

Two tabs: Photo creates a new image from a text description; Variation starts from a photo you upload and re-imagines it.

1. In the left menu press Visual Design Studio — it opens on the Image tab.
2. In the prompt box, describe the picture: subject first, then scene, then mood. Example: “a red fox sitting on a mossy rock at dusk, soft rim light, photorealistic”.
3. Optional: pick the shape (landscape / vertical / square), and fill the negative box with what you don’t want (e.g. “blurry, text”).
4. Press Generate. A progress card appears; when it finishes, the image shows below with a Download button.
5. Not happy? Press Generate again — the same words give a different take each time. Keep the best one; everything lands in the Image Library tab automatically.

The Avatar tab of Visual Design Studio is the character-savvy image tool. Three tabs: Photo from words (describe a person or character from scratch), From a reference (build a new picture from photos you upload — faces, outfits, styles), and Edit (change an existing picture).

1. Press Visual Design Studio in the left menu, then the Avatar tab.
2. From a reference: press the add slots to upload your photos (a face, a jacket, a background… — up to 16). Write what to make from them — e.g. “the woman from the reference wearing the red jacket, standing in a rainy street” — and press Generate.
3. Edit: upload the picture you want to change as the canvas, add extra reference photos if needed, and describe only the change — “change the jacket to black leather, keep the face and pose”.
4. To keep a face recognisable, use a clear head-and-shoulders photo as a reference and say “keep the face and identity” in the prompt.

Video AI makes your clips; the AI Utilities menu holds every other moving-image tool. Turn a photo into a moving clip (Animate a photo) or create a clip from words alone (Video from words). Clips are a few seconds long and take a few minutes to make.

1. Press Video AI.
2. Animate a photo: press Choose image and pick the first frame — a clear photo works best. Then describe the motion: “the fox pads forward and looks up as fireflies drift”.
3. Video from words: skip the image and describe the whole shot.
4. Pick the shape, the length (the slider snaps to the lengths that render cleanly), and the frame rate. Press Generate clip.
5. Watch the progress card. When it finishes, a video player appears with a Download button; the clip is also in the Video Library tab.

Give the Lip Sync tab of AI Utilities a portrait photo and a voice clip of someone speaking, and it returns a video of the person in the photo saying those words — mouth moving in sync, with your original audio kept exactly as you recorded it.

1. Press AI Utilities in the left menu, then the Lip Sync tab.
2. Press Choose portrait and pick a clear, front-facing photo — the face should fill a good part of the frame, eyes visible, not turned away.
3. Press Choose voice clip and pick the recording (mp3, wav, m4a, ogg…). Keep it under 15 seconds — the video follows the clip’s length automatically.
4. In the scene prompt, describe how the person looks and speaks: “a woman in a sunlit kitchen speaks calmly to the camera, natural hand gestures”.
5. Pick the shape (landscape / vertical / square) and press Make it talk.
6. A progress card follows the render. When it finishes you get a video player — the person now speaks your clip. Download it, or find it later in the Video Library tab.

Tips: record the voice in a quiet room (the sync follows the recording as-is), and keep the scene prompt consistent with the photo — a photo of a chef in a kitchen works best with kitchen words.

The character tabs in AI Utilities make a character speak and act: Face Video animates a portrait from a short description, Character Swap replaces a person in an existing clip with a new character, and Continuous stitches longer scenes together.

1. Press AI Utilities in the left menu.
2. Choose the tab (Face Video / Character Swap / Continuous).
3. Add the photo — and the source clip for Swap — then describe who the character is and what they say or do.
4. For Swap, you can preview first: Check mask runs SAM tracking only (~1 minute per 3 s, ~2 credits per second instead of 20) and shows exactly which area the swap regenerates — cyan is regenerated, orange is the tracked person. Adjust what SAM tracks and the edit margin until the mask fits, then swap.
5. Swap also offers Part to use (swap only a section of a longer clip — proportionally faster) and Fast mode (DMAD 6-step sampling, roughly 3× faster with slightly softer detail).
6. Press generate and wait — these are the longest jobs; watch the Jobs page for progress. Finished clips appear below and in the Video Library.

Four tools: Audio to Text turns a recording into text, Text to Speech reads your words aloud in a chosen voice, Voices manages the voice list, and Music composes an instrumental track.

1. Press Audio Studio in the left menu.
2. Audio to Text: upload an audio or video file — you get a timestamped transcript.
3. Text to Speech: paste your script, pick a voice and speed, press generate. A player appears with the narration; download it or use it in Movie Production House.
4. Music: describe the mood (“calm piano over soft strings”), pick a length, and a track is composed for you.
5. Everything lands in the Audio Library tab.

Make any image larger and crisper. Fast upscale sharpens what is already there; Generative detail re-imagines fine detail while enlarging (slower, richer).

1. Press Visual Design Studio in the left menu, then the Upscale tab.
2. Upload the image (or pick one from your library).
3. Choose Fast upscale or Generative detail and the enlargement factor.
4. Press generate — the enlarged image appears below and joins the library.

Movie Production House runs a whole story end to end: paste or write a script, split it into scenes, assign characters, then generate the images, video, and narration for every scene — with the same characters looking the same across scenes.

1. Press Movie Production House in the left menu, then New project and give it a name.
2. Paste your script and press Break into scenes — the script is split into numbered scenes you can edit.
3. In Cast, add your characters (upload a face photo or generate a character sheet) so every scene keeps them consistent.
4. Per scene, press Generate image (and video / narration if wanted). Jobs queue automatically and fill in as they finish.
5. Use the Timeline to review, re-order, and export the finished piece.

The library holds two curated collections: Image Prompts and Video Prompts — each with its own keyword cloud filter. Any description you like can be saved to your own list, then reused later: search it, or press Use in Image / Video / Clip to load it straight into a studio (video collection prompts load into Video AI).

1. In any studio, press Save prompt next to the prompt box (add tags like portrait or product).
2. Open Script Studio in the left menu to search, edit, or reuse them.

Every studio has a Library tab with everything it made for you (newest first, 100 per page). Cards show a preview, a download button, and Archive.

• Archive hides an item. It is reversible and nothing is deleted.
• The Archived view (a chip above the grid) lists hidden items with Restore and Delete forever.
• Delete forever asks for a second click and is final — there is no trash bin below the Archived view.
• Archived items are kept for a while, then a nightly sweep clears them out. Restore before that window to keep them.
• Re-generating from an existing item creates a new version; the library keeps the active one and the history stays traceable.

Every creation is a job. The Jobs page shows what is queued, running, and finished, with progress and a cancel button. Big jobs (video, lip sync) run one at a time on purpose; small jobs (audio, transcribing) can overlap. A job waiting in the queue is normal — press Cancel only if you changed your mind.

Each creation costs credits, shown as an estimate before you press generate — you always confirm the price first. The price follows the options you selected: pick a different model, workflow or engine and the upfront estimate changes with it. For example on the Image tab a plain SDXL render costs 10 Cr, Krea-2 Turbo (bigger model, more steps, higher quality) costs 15 Cr, and the Qwen Image 2.1 workflow costs 40 Cr. Video-family jobs (video, lip sync, character clips) are priced per second of output, so the estimate scales with the duration you chose. The Pricing page lists every rate, and your remaining balance is in the left menu. New accounts start with a welcome balance.

What happens to the credits once a job ends — the shown estimate is what you pay in each of these cases:

Refunds need nothing from you: they are issued automatically in the same step that marks the job failed or cancelled, and appear under Credits → Transactions. The price quoted before you press generate is the price reserved — retries of a failed job bill as a new job only after a new estimate is shown and confirmed.

Everything you can do in this web app is also available as an HTTP API, so a script or another app can generate scripts, images, videos, and narration for you. You drive it with an API key: open Settings (the Account tab), find the API keys panel, press Create key, and copy it immediately — the full key is shown only once. Keys start with aist_, can carry an expiry and a daily/total request limit, and you can revoke one at any time without affecting your login.

Every call goes to the same address you are using right now — https://aistudio.bishnubhatta.stream — so a script on any computer can reach AIStudio. Put your key in the Authorization header of each request. The flow is the same as clicking in the app: submit a job, then check on it until it finishes. For example, to generate a script for one of your projects:

curl -X POST "https://aistudio.bishnubhatta.stream/jobs/submit?job_type=script-generation&priority=INTERACTIVE" \
  -H "Authorization: Bearer aist_YOUR_KEY" -H "Content-Type: application/json" \
  -d '{"project_id": 3, "brief": "a lighthouse keeper finds a message in a bottle"}'
# -> {"job_id": "..."}

curl -H "Authorization: Bearer aist_YOUR_KEY" \
  https://aistudio.bishnubhatta.stream/jobs/THE_JOB_ID
# -> status QUEUED/RUNNING, then COMPLETED with the result in "output"

The same key works for everything. The calls behind the most common tasks:

Jobs cost credits exactly as in the app — the estimate call shows the price before you run anything, and a key that hits its daily limit gets a clear “quota” error rather than a surprise. Project, scene, and job numbers are the same ones shown in the web app.