Make Objects Talk

Write a script in the first person as an object. A trainer, a coffee cup, a bottle of hot sauce. The studio builds it out scene by scene, with that object as the character in every shot. It is a short-form format that people watch to the end because they want to know where it is going.

Start creating

How to make a talking object video

Write the script in the object’s voice
This is the whole trick, and it happens before you touch any setting. Write in the first person as the thing itself: "Nobody has washed me in four months." "I was bought for a marathon. I have been to the shops." The comedy and the sales pitch both come from the object having a point of view, and no model can supply that for you.
Set the length and language
Length governs how many scenes the script is broken into, so a longer video means more shots rather than slower ones. Language is chosen here too, which makes the same joke portable across markets.
Choose how the scenes look
The media type decides whether each beat is generated imagery or drawn from elsewhere, and you can have the scenes animated rather than left as stills. Stills cut fast and cost less to iterate on; animated scenes give you movement in the shot, which suits a slower, more cinematic read.
Pin your object as the character
The character step is optional but it is what makes the format work commercially. Upload a photo of your product, pick one from your library, or use one of the premade characters, then switch on consistent character so the same object appears in every scene rather than a new interpretation of it each time.
Add music and generate
Background music sets the register. A wistful piano and a bouncy loop turn the same script into two different videos. The finished piece lands in your library, ready to trim, caption, and schedule out to YouTube, TikTok or Instagram.

What people make with it

A product that argues its own case

The pitch nobody skips, because it does not sound like a pitch. Your product narrates its own day, complains about the competition, or explains why it exists. The viewer takes in the features without being sold at.

Neglected-object comedy

The gym membership card, the exercise bike, the herbs at the back of the cupboard. Self-aware, faintly sad monologues from household objects are a durable short-form format, and they are entirely writing. That means they are cheap to try.

Explaining a service without a presenter

When the thing you sell is dull to look at, giving it a personality is the easiest way to hold attention. An insurance policy that speaks about itself is more watchable than an insurance policy in a slide.

Restaurant and food posts

A dish talking about how it is made is a better menu than a menu. The format suits food particularly well because the object is already appealing and the script only has to be warm.

Brand mascots that do not need drawing

Pin the same object across a run of videos and you have a recurring character. Consistency across posts is what turns a one-off joke into a format people recognise in the feed.

Series that keep the same prop

Because the character can be held steady across scenes, an episode is a script change rather than a rebuild. That is what makes a weekly slot realistic instead of aspirational.

What is a talking object video?

A talking object video is a short film in which an inanimate thing is the narrator. The object addresses the camera in the first person. Describing itself, complaining, remembering, selling. The humour or the persuasion comes from the gap between an ordinary thing and an unexpectedly specific personality.

The format is old. Anthropomorphised objects have carried advertising for a century, and children’s animation longer than that. What has changed is the cost: giving a bottle a personality used to mean an animator, a rig and a schedule. Now it means a script and a picture of the bottle.

Mechanically, this studio works from your script outward. The script is split into scenes, each scene becomes a shot built around your object, and those shots are assembled into a video with a music bed under it. The object you supply is the constant. That is the point of the consistent character switch, which the interface describes plainly as using the same character across scenes.

One thing to know before you plan around it: this studio has no voice selection. The options panel says so directly. Voice is not used here. What you are assembling is the visual performance and the music. If you want a spoken performance from a voice you picked, that is what the talking avatar studio is for.

Talking objects vs a talking avatar

These two look adjacent and are not interchangeable, and picking the wrong one is the usual reason a video underperforms. A talking avatar puts a person on screen delivering your argument. Object Talk puts the subject itself on screen delivering it. The question is whether your content is an opinion or a thing.

If the message is "here is what I think about this", a person should say it. Opinions land better attached to a face, and a face carries the parasocial weight that makes people follow. If the message is about a specific object, giving the object the voice is stronger. This product, this dish, this prop. The viewer then spends the whole video looking at exactly the thing you want them to remember. An avatar describing your product asks the viewer to imagine it; an object describing itself does not.

There is also a practical asymmetry in the audio. The avatar studio picks a voice and a language and delivers your words spoken. This one does not. Voice is absent by design, and the audio you control is the music. That makes Object Talk a visual and written format: the script drives what you see and the pacing, and the register is set by the track underneath. Plan for that rather than discovering it afterwards, because it changes how you write. Some lines depend on vocal delivery. Sarcasm carried by tone, a pause for timing. Those are riskier here than lines that are funny on the page.

The third comparison is against simply filming your product, which is the option people skip past too quickly. If the product genuinely looks good and the message is "look at this", film it. Generated video earns its place when the shot you want does not exist and would be tedious to stage. The trainer in six different rooms across ten years, the cup on a windowsill in weather you do not have. Object Talk is at its strongest when the script asks for a sequence nobody is going to set up in real life.

A reasonable pattern once you are running all three: the avatar handles opinion and recurring commentary, Object Talk handles anything where the product is the subject, and real footage handles the moments that need to be real. They are not competing for the same slot, and the fastest way to waste a week is to make one of them do another’s job.

Getting better results

Give the object one grievance
The scripts that work are specific and slightly aggrieved: a thing that wants something, resents something, or remembers something. "I am a great mug" is not a video. "I am the mug you only use when the good ones are dirty" is.
Write in the first person, always
The moment the script slips into third person, the format collapses into an ordinary product video with an odd camera angle. One "this bottle contains" does it. Every sentence should be something the object could say about itself.
Photograph the object properly before you pin it
The character image is inherited into every scene, so its faults are inherited too. Plain background, even light, the whole object in frame, sharp. A cluttered snap taken at an angle gives the model less to hold on to, and the object drifts between shots.
Turn on consistent character for anything commercial
For a joke, drift does not matter much. For a product, it is fatal: a viewer who sees three different versions of your packaging has learnt nothing about your packaging. The switch exists for exactly this, and there is no reason to leave it off when you are selling.
Let each line be its own shot
The script is divided into scenes, so it pays to write in beats rather than paragraphs. One idea per line gives the studio a clean place to cut, and short beats cut together better than long ones regardless of what generated them.
Choose music before you judge the edit
With no voice track, the music is doing the emotional work in this format. More than it would anywhere else. The same sequence reads as funny under one loop and melancholy under another, so try the opposite of your instinct once before settling.
Land the ending
The strongest versions of this format turn at the end: the object reveals why it has been talking. That final line is where a viewer decides between scrolling on and sending it to someone, and it is worth more rewriting than the opening.
Keep the frame upright
The studio defaults to a 9:16 frame because this format lives on TikTok, Reels and Shorts. Only widen it if the video is destined for a site or an email, where an upright clip looks stranded.

Models available

Veo 3.1 — Google's photoreal generation, with sound. Up to 12 seconds.
Nano Banana 2 — Google's best stills. Generation and editing, to 4K. The scenes your object appears in.
Kling 3 — Cinematic motion that holds a character across a cut.
Seedance 2.0 — ByteDance, up to 24 seconds. The long shots.
Grok Video — xAI. Fast image-to-video, and text-to-video.

Questions

Can I use a photo of my own product?: Yes. Upload it as the character, or choose one you have already saved to your library, and switch on consistent character so the same object carries through every scene instead of being reinvented each time.
Is there a voiceover?: Not in this studio. It has no voice picker. The options panel states that voice is not used, so what you are choosing here is the visuals and the background music. If you want a script delivered by a voice you selected, the talking avatar studio is built for that.
Do I have to supply an object at all?: No, the character step is optional. Leaving it out gives the model freedom to invent the subject, which is fine for a joke and wrong for a product. If the specific thing matters, pin it.
Can the scenes move, or are they stills?: Either. The studio can animate the generated scenes rather than leaving them as static frames. Stills are quicker to iterate on when you are still testing the script; animation is worth it once the writing is settled.
What length should I aim for?: Short. The format depends on a viewer wanting to know where the monologue is going, and that curiosity has a half-life of about twenty seconds. Longer versions work when the script genuinely turns partway through.
Does it work for things that are not products?: Yes. The object can be anything with a personality you can write. Household objects, tools, vehicles, food, a building. The constraint is the writing, not the subject.
Can I post it straight to social?: The video lands in your library, where you can trim it, add captions and schedule it out to YouTube, TikTok and Instagram.
Can I use the videos commercially?: What you make is yours. Check the terms for the specifics before you build a campaign on it.