Manufaktura

Paste a link and it says what it found before fetching anything — then pulls the largest stream Instagram holds, or an iPhone-friendly one.

A photo and a prompt become a clip. Three engines — Kling, Omni Flash, Seedance — a length in seconds, sound on or off, and the price shown before you press anything.

A still photo and a soundtrack become a talking avatar through HeyGen. Every field their interface fills in for you is filled in here and shown before a penny is spent — including the audio length, which is the only way to catch the wrong track.

Transcribes the speech, cuts it into lines and burns them into the frame. Style, font and position are all on the preview before you render.

Upload a wide video or paste a YouTube link, and it finds the strongest stretch by what is said — then stands it up as 9:16, whole frame or centre crop.

A talking head comes in flat and comes out cut: four scenes, wide and close in turn, the close-ups pushed in on the face. Same length, same soundtrack, untouched.

A shelf of ready effects — a click, a shutter, a bell, a cash register. Put each one anywhere on the timeline, set how loud it is, hear it against the video before you render. They mix with the clip’s own sound rather than replacing it.

Animated “subscribe”, “save” and “share” graphics laid over the video: pick as many as you like, drag each into place, set its size and whether it shows at the start, at the end or all the way. The preview shows exactly what the file will hold.

Two tracks — video and sound. Drop the files in, drag the clips about, trim them by the edges, split with the blade, set a volume for each, watch it play, then render one file. Free — ffmpeg does the assembly.

The speech is transcribed and every sentence end is offered as a cut on the timeline — keep the ones you want, drop the rest. At each cut the picture dips to black, comes straight back and jumps a touch, the way a real cut does.

A video and a track in, the video with the track laid under its own sound out. The music volume is tried on a live preview. ffmpeg only — free.

The camera drifts a few centimetres: what is near moves more than what is far, and the flat clip gains depth. A depth model on our own server — free.

Any number of videos in, one file out — in the order of the list, with no gaps and the sound in sync. Different sizes and frame rates are fitted to the first clip. ffmpeg only — free.

Finds every cut by how much the picture changes, and gives each scene its times and a sharp frame from its very beginning. ffmpeg only — free.

Step frame by frame with the arrows, grab stills at the clip's own resolution, download them one by one or all at once.

Offers a few frames worth using — sharp, and with a face in them where there is one — and puts your title on the one you pick. Ten looks, and you place the title yourself.

Write what should be in the picture, attach reference photos if you have them, pick the format and quality. Nano Banana 2, with the price shown before you press anything.

The photo is fitted into the frame by dragging and zooming; text blocks come in any number, with presets and your own fonts.

Three voices — Leo, Victor ru and Victor En — with an emotion to pick. Each can be heard before you spend a generation on it.

A waveform you can zoom to a single frame, cuts, deletions, tempo and silence removal. Any piece downloads on its own.

Not a translation: the meaning, the order and the length stay, while the idioms, the numbers and the turns of phrase become the ones that language actually uses. The result is a plain text field — edit it as you like.

A file or an Instagram link goes in, the spoken script comes out — ready to edit and copy.

The schedule and autoposting live next door, in the content plan. This opens it — the accounts, the slots and the queue are all there.