MiniMax H3 Product Ad Prompts That Sell

A product ad is usually four shots, not one: a reveal, an in hand shot, a demonstration and a lifestyle frame. These are MiniMax H3 prompts for each, written to be pasted and adapted. The most important rule appears in all of them: attach your product photo rather than describing it.

Why attach a photo instead of describing the product?

Because a description gets you something that resembles your product, and resemblance is not good enough when a customer is about to compare the ad to the product page.

Attaching a single image switches H3 into image to video, where your still becomes the literal first frame and the clip animates forward from it. Your label stays your label, your colour stays your colour, your bottle shape stays your bottle shape. Nothing is invented.

Two practical notes. The output aspect ratio follows your image, so crop to your placement before you upload. And do not re describe what is already in frame; describe what happens next.

Our general guide to turning product photos into video ads covers the technique across models.

The reveal shot

Opens an ad or lands after a hook. No people, all attention on the object.

With a photo attached:

The camera pushes in slowly from a low angle as soft key light rises from the left, catching the edge of the product. Shallow depth of field, rich blacks, clean highlights. Quiet room tone, one soft click, no music. 

Text to video, if you have no photo yet:

A matte black wireless charger sits on a dark wooden desk beside a closed notebook. The camera pushes in slowly from a low angle as soft key light rises from the left, catching the edge of the product. No people in frame. Shallow depth of field, rich blacks, clean highlights. Quiet room tone with a single soft click as a phone is set down, no music. Horizontal framing.

Why it works: one light source, one camera move, nothing else competing. Reveal shots fail when people stack a rotation, a push and a light change into five seconds.

The in hand shot

This is the one that sells scale. A product photographed alone could be any size. A product in a hand is instantly understood.

A hand holds a small amber glass serum bottle up against a plain warm wall, turning it slowly so the light moves across the label. Extreme close up on an 85mm lens, static camera, soft diffused daylight from the left, shallow depth of field. Quiet room tone, no music. Vertical framing.

Keep the motion to one slow turn. Fiddly hand work, unscrewing a cap or pressing a small button, is still unreliable across every model in this class and will cost you renders.

The demonstration shot

Shows the thing the product actually does. This is the beat most ecommerce ads skip and then wonder why conversion is soft.

A woman in her thirties stands at a bathroom counter and presses a pump on a white bottle, releasing cream into her palm, then rubs her hands together. Medium close up, static camera at eye level, soft even daylight from a window on the left, clean natural colour. Bathroom room tone, the quiet click of the pump, no music. Vertical framing.

A man in his thirties sets a slim power bank on a desk and connects a phone to it with a short cable. A small indicator light turns on. Close up on a 50mm lens, static camera, warm desk lamp light from the right, shallow depth of field. Quiet room tone, a soft connector click, no music. Horizontal framing.

Note the pattern: one action, one camera position, and the sound of the action written explicitly. Those small sounds do a lot of work in making a product feel real.

The texture close up

For anything where material quality is the argument. Fabric, food, skincare, leather, ceramics.

Extreme close up of a thick cream being spread slowly across a matte ceramic surface with a small spatula. No people visible beyond a hand. Static camera on an 85mm lens, soft directional light from the left raking across the surface to show texture, shallow depth of field, clean natural colour. The quiet sound of the spatula against ceramic, no music. Vertical framing.

The key phrase is "raking across the surface". Soft light from the side reveals texture. Flat frontal light hides it, which is why so many product close ups look like plastic.

The lifestyle frame

Shows who the product is for. Sells the identity rather than the object.

A man in his late twenties sits on a rock at the edge of a pine forest at golden hour, taking a metal water bottle from a backpack and drinking. No dialogue. Wide shot settling to medium, slow pull back, warm backlight through the trees, lens flare, teal and orange grade. Wind in the trees, distant birds, no music. Horizontal framing.

A woman in her early thirties pours coffee into a ceramic mug at a kitchen counter in early morning light, then wraps both hands around it. No dialogue. Medium shot, slow push in, warm low angled sunlight through a window behind her, soft haze in the air. The sound of pouring liquid and quiet kitchen room tone, no music. Vertical framing.

Lifestyle frames are the one place where a heavier grade earns its keep, because you are selling a feeling rather than pretending to be unproduced footage.

Should product beats have dialogue?

Usually not.

The spoken line belongs in the hook, at the top of the ad, where it has to earn attention. Once someone is watching, the product beats work better with ambience and an on screen caption than with a voice explaining what they can already see.

The exception is a demonstration where the point is not visually obvious. There, one short line helps.

How do you assemble these into an ad?

A structure that works for most ecommerce offers:

1. Hook. A person, a specific line, handheld. From the UGC prompt templates.
2. Product beat. Reveal or in hand, five seconds.
3. Demonstration. The thing it does.
4. Lifestyle or end card with your offer.

Generate each beat separately at five seconds, then assemble in the editor. Generating separately is not a limitation to work around, it is the right way to do it, because you can reshoot one beat without regenerating the whole ad.

Add captions before you ship. Most social video is watched muted at first contact, which our guide to adding captions covers.

What should you avoid asking for?

On screen text, prices and logos, because lettering is unreliable across this whole class of models. Add them in the editor. Also avoid stacking camera moves, crowded scenes with several people handling the product, and any fine motor action like threading, clasping or unscrewing.

Frequently Asked Questions

Should I describe my product or attach a photo?

Attach a photo whenever you have one. Description gets you something that resembles your product. A photo attached as the first frame gets you your actual product.

What product shots does an ecommerce ad actually need?

Usually four: a reveal, an in hand shot, a demonstration of the thing it does, and a lifestyle frame that shows who it is for.

How do I stop the model inventing my packaging?

Stop describing it and attach it. Image to video uses your still as the literal first frame, so the label and colours stay correct.

Should product shots have dialogue?

Usually not. Product beats work better with ambience and a caption. Save the spoken line for the hook shot that opens the ad.

What lighting reads best for product video?

Soft directional light with one clear source. Hard even light flattens the object and removes the edge definition that makes it look real.

How long should a product beat be?

Five seconds is usually plenty. The product beat supports the hook, it does not carry the ad on its own.

Build one full ad

Take your best selling product, generate the four beats above, and assemble them. It is one sitting and you end with a complete ad rather than a clip. The walkthrough for the image driven beats is in how to use image to video. Shopify's blog has useful background on what product content is meant to do, and Meta's advertising guidance publishes current placement specs worth checking before you render.