Prompts for every level

You say what should change, and everything else in your video stays. That is the whole idea of Optic, and it is why a first prompt can be one plain sentence. A professional prompt is a few sentences more, each doing one job — not a different language, just the same sentence with more said. If it is in your video, you can revise it: a person, a coat, a lamp, the words on a sign, the light, a whole cast. See What you can revise.

This page shows the same wishes written at three levels, so you can find your own level, see exactly what the next one adds, and know when the simple one is already enough.

#Find your level

Whatever the level, the price is the same: it depends only on the seconds you chose and the quality, never on how much you write. It shows on the Revise button before you press, it is held rather than spent, and every credit comes back if no film is delivered.

#When one sentence is enough

The simple prompt works when all four are true:

• there is only one subject your words could mean — the dog, the sign, the man;
• the change holds for the whole stretch;
• what it becomes is clear from a few words — a colour, a plain object, a common look;
• nothing right next to it is at risk of drifting with it.

When one of them is not true, you are at the next level — and the next level is only the missing part said out loud.

#A person

• A first try: "Replace the man with the person in @Image 1."
• Getting good: "Replace the man in the blue shirt with the person in @Image 1, exactly as the picture shows him. The woman beside him stays as she is."
• Professional: "Replace #Focus 1 with the person in @Image 1 and @Image 2, exactly as the pictures show him, for the whole stretch — @Image 1 is his face, @Image 2 his coat from behind. The woman beside him, the table and everything on it stay as they are."

What each level adds: the first is enough when there is one man. The second says which man and names the one thing beside him that could drift. The third marks him so there is no doubt at all, gives a picture for the front and one for the back, and says the change holds for the whole stretch. At every level the performance is your video's — his eyes, lips and expressions follow the original, and he moves as the old man moved — so none of the three describes how he acts.

Putting a real person in? The Content Policy is plain: not a real person's face, body or voice without their consent.

#Clothes and fabric

• A first try: "Make her coat black."
• Getting good: "Change her coat into a long black wool coat with a wide collar. Her scarf and her bag stay as they are."
• Professional: "Change the coat of #Focus 2 into heavy black wool with a visible weave, matte, with no shine. The left side of the jacket under it, which shows from 0:06 to 0:11, is dark green velvet. Her scarf, her bag and her boots stay as they are."

What each level adds: colour alone is a first try. The second adds the material and the cut, and holds the scarf and the bag in place. The third names the texture the way a costume department would, reaches one side of one garment, and says when that side is seen, in the clip's own clock. Fabric, weave, shine and the state of a garment — wet, muddy, worn — are all yours to name.

#The place, or one thing in it

• A first try: "Make the lamp on the table brass."
• Getting good: "Change the lamp on the table into a tall brass lamp with a green glass shade, lit, in the same place. The table and everything else on it stay as they are."
• Professional: "From 0:00 to 0:09 the room is a sunlit café: wooden tables, a long counter at the back, plants in the window. The two people, the camera and the light on their faces stay as they are. The picture frame on the wall at 0:21, #Focus 3, holds a black-and-white photograph of a harbour instead of the flowers. From 0:30 to the end, the inside of the car is cream leather with a wooden dashboard."

What each level adds: one object, named, is a first try. The second says what the new object is made of and where it sits, and holds the rest of the table. The third is a whole scene's worth: the room in one stretch, one frame on one wall at one moment, the inside of a car later — each its own sentence, each with its own seconds. The people, the camera and the cuts stay exactly where your video has them.

#The look

• A first try: "Make it look like evening."
• Getting good: "Make it look like early evening: warm low sunlight from the left, long soft shadows, a slightly orange sky. The people and the camera stay as they are."
• Professional: "For the whole stretch the look is a cold overcast morning: flat grey light, muted colours, no hard shadows. From 0:18 to the end, the neon sign over the door, #Focus 1, is lit in pink and throws its colour onto the wet pavement under it. Skin tones stay natural; the words on the sign stay as they are."

What each level adds: a time of day is a first try. The second describes the light as you would see it — direction, warmth, shadows, sky. The third sets one look for the stretch, adds a second light that comes on at a given moment, and names what the new light must not touch.

#Text on screen

• A first try: "Remove the subtitles."
• Getting good: "Change the words on the white sign above the door from OPEN to CLOSED, in the same lettering, colour and size."
• Professional: "Replace the logo on the van door, #Focus 2, with the logo in @Image 1, at the same size and angle, for the whole stretch. From 0:05 to 0:12 the caption at the bottom of the frame reads NEW IN STORE in the same white lettering; before 0:05 and after 0:12 it is removed. The number plate stays as it is."

What each level adds: removing is a first try. The second gives the exact new words and holds the lettering. The third brings your own artwork for a logo, times a caption to the seconds it should be seen, and says what happens to it outside those seconds.

#A crowd or a cast

• A first try: "Everyone at the table wears black."
• Getting good: "Everyone at the table wears plain black clothes. The man at the head of the table keeps his glasses. The room stays as it is."
• Professional: "#Focus 1, the fox that talks from 0:03 to 0:09, becomes a grey tabby cat of the same size in the same red scarf. #Focus 2, the rabbit beside it, becomes the owl in @Image 1. The bear and the kitchen stay as they are."

What each level adds: one change for the whole crowd is a first try. The second carves out one exception and holds the room. The third changes several characters of an animated cast in one prompt, one sentence each, each marked with Focus so no description is needed twice. You do not describe how the cat speaks: its mouth and eyes follow the fox's, at the same moments, and the voice is your clip's own.

#When you write for a series

The same prompt and the same pictures, video after video, is a professional habit rather than a feature: a picture is the whole new look, so giving the same pictures and the same words is the surest way to keep that look from one video to the next. Keep one copy of your prompt and your pictures. For the next video, change only the times and the Focus numbers; leave every word that worked.

#If you come from other video models

Optic reads tags, paragraphs and shot lists; for the surest film, turn them into plain sentences that name only what changes. Nothing here needs a special syntax beyond @ for a picture and # for a focus.

Some habits from models that start from nothing simply have no work to do here, because Optic already keeps everything you do not name:

• Describing the whole scene. Optic sees your clip. Describe only what changes; every extra sentence is one more thing that might change.
• Camera directions. The camera stays as your video has it — every move, stop and framing. A request about the camera alone is not a revision.
• Quality tags — "ultra-detailed, masterpiece, sharp focus". The quality is a chip you choose, 720p or 1080p; words do not change it.
• Long lists of what to avoid. Optic already keeps everything you do not name, so the list has nothing to hold back. Say what you want to see instead.
• Brackets and weights. Plain words carry the same meaning. Put the thing that matters most in its own sentence.

The one real difference: you are revising, not starting over. The story stays. To change what happens in the shot, use the chat.

#A template for each level

A first try:

Make [what] [how].

Getting good:

Change [what changes] into [what it becomes — colour, material, shape, style].
It is [which one — "the woman on the left", or #Focus 1].
[Only if something next to it might drift: keep [what] as it is.]

Professional:

[What changes] becomes [what — material, texture, colour, cut, state], [for the whole stretch | from 0:12 to 0:20 | from 0:30 to the end].
It is #Focus N[, the subject in @Image N, exactly as the picture shows it — and anything in the picture you do not want seen].
[Next change, its own sentence, its own seconds, its own focus or picture.]
[What is next to a change and must stay: … stay as they are.]

#Your next step

Open Studio, press Revise a clip, bring your clip and confirm the stretch. Paste the template for your level into the words box and fill it in. Check the price on Revise, read your prompt once from the top, and press.

For the five parts in detail, read Writing Optic prompts. For more pairs of a weak prompt and a strong one, Prompt examples. For what your pictures do, Your pictures. For what you can make with Optic, What you can use Optic for.

#Frequently asked questions

#Do I have to work my way up through the levels?

No. The first two levels are enough for most changes. The professional level is there for the day a job needs it — several subjects, a series, exact seconds.

#Does a longer prompt cost more?

No. The price depends only on the seconds you chose and the quality. It is on the Revise button before you press, it is held rather than spent, and it comes back in full if no film is delivered.

#Can I mix levels in one prompt?

Yes. One sentence can be a first try — "Remove the subtitles." — and the next a professional one with a focus and seconds. Each sentence stands on its own.

#Can I write in my own language?

Yes. Write in any language; the structure is the same.