Model, Creator or AI: What Your 2026 Campaign Actually Needs
In short a 2026 campaign budget buys one of three things: a booked model, a creator who shoots and posts from a personal account, or generated imagery. Each option prices differently on speed, usage rights and recognisability. Generated imagery is strongest on volume product work without a face. A booked model is strongest where a recognisable person carries the message.
Search for the cost of hiring a model and roughly a third of the first page is published by companies that sell image generation. Photta, Blendnow and Koozee all rank for it, and all three answer the question the same way: a live shoot costs thousands, generation costs very little, so why book anyone.
The arithmetic is not wrong. What it leaves out is that the three options are not substitutes for each other, and a brand that treats them as interchangeable usually discovers the difference after the campaign is live.
At Ambressa we book models and we run creator campaigns, so we have no particular reason to defend one against the other. What follows is the comparison we give clients who arrive with one budget and three ways to spend it.
What are you actually buying in each case?
Three different products that happen to produce images.
A booked model
You are buying attendance, direction and a licence. The model arrives at a call time, works to your art direction, and grants you scoped use of their likeness. You control the frame, the styling and the retouch. You also inherit the paperwork: a day rate, a usage fee, and sometimes an exclusivity fee, which is how usage rights are priced across the region.
A creator
You are buying an audience and a voice, and the images are a by-product. The creator shoots on their own phone, in their own flat, in their own register. Control drops sharply. In exchange you get distribution you cannot buy any other way, plus the comment section, which is the part brands consistently underrate.
Generated imagery
You are buying pixels and speed. No call sheet, no travel, no licence conversation with a person, because no person was there. Revisions cost minutes. Nothing is scheduled and nothing can be late.
How do the three compare on the axes that matter?
| Booked model | Creator-model | Generated | |
|---|---|---|---|
| Lead time | Days to weeks, availability driven | Days, subject to the creator's own calendar | Same day |
| Cost driver | Day rate plus usage | Per deliverable | Per seat or per credit |
| Revisions | Reshoot, expensive | Limited by agreement | Effectively free |
| Usage rights | Negotiated by media, territory, term | Same four axes, often skipped | Clean unless derived from a real person |
| Best at | Hero, OOH, packaging, anything with a face | Paid social, TikTok Shop, review formats | Volume e-commerce, flat-lay, colourways |
| Weakest at | Volume SKU work | Tight art direction | Anything where a person must be believed |
Where does generated imagery genuinely win?
More places than agencies like to admit, and pretending otherwise costs credibility with anyone who has actually tried it.
Flat-lay product, where nothing human is in frame. Background extension, when the crop is wrong for a placement that came late. Colourways of a garment already shot on a real body, where the fit and drape are photographed and only the fabric changes. And high-volume e-commerce, where a catalogue of several hundred SKUs needs consistent images and nobody is examining a face.
In all four the saving is real, no likeness is involved, and there is no usage conversation to have. A brand that books a model for those jobs is spending money on a problem that does not need a person in it.
Where does it stop working?
The moment the picture asks the viewer to believe a person.
A generated face has no reputation. It has never been seen anywhere else, it cannot be looked up, and it has no comment section. For a hero image on a billboard that may not matter. For a skincare claim, a testimonial, or anything a local audience will read closely, it matters a great deal.
Southeast Asia reads casting closely. Audiences in Kuala Lumpur, Jakarta and Manila can place a face by market, sometimes by community, and they notice when a campaign gets it wrong. This is the same reason casting for the region is a market question rather than a lookbook question. A generated composite that belongs to no market in particular tends to read as belonging to none of them.
One trap worth naming
If you generate imagery from photographs of a real model, that output is still built on that person's likeness. The licence you signed for the original shoot almost certainly said nothing about training or synthesis, because most booking templates in the region predate the question. Generating from it does not remove the usage conversation. It moves the conversation to ground where your paperwork is weaker.
Put the permission in the booking if you intend to do it. Asking afterwards is a negotiation you will lose.
What is the creator-model hybrid?
A creator who is booked like a model but shoots like a creator, and it is the fastest-growing of the three.
Soona prices this format per deliverable rather than per day and flags it as new for 2026. You commission a set number of assets, the creator produces them in their own environment, and you receive usage for paid media. No studio, no call sheet, and output that already looks like the platform it will run on.
The trade is control. You are briefing rather than art directing, and the images will look like the creator's feed because that is the point. For paid social this is usually an advantage. For a campaign key visual it usually is not.
This is also where the two halves of our own work meet, and where clients most often arrive with the wrong assumption: that a creator campaign and a model booking are alternatives. On most briefs they are sequential. The model shoot produces the assets that hold up on a billboard, and the creator campaign produces the volume that keeps the account alive between campaigns.
Does the answer change from market to market?
It changes more than most regional plans allow for, and the variable is rarely cost.
Malaysia and Singapore are the two markets where a brief most often lands on a booked model, because campaigns there tend to run across paid media and retail at the same time, and that combination is a usage question before it is a casting question. Both markets also carry a genuine expat and Chinese-speaking audience alongside the local one, so a single campaign frequently needs two or three casting variants rather than one hero face.
Indonesia, the Philippines and Vietnam sit further towards creators. Platform-native video carries more of the buying journey, the formats that convert look like the feed rather than like a campaign, and a polished studio asset can read as an advertisement in the wrong way. Thailand sits between the two and moves by category, with beauty and food behaving differently from retail.
Generated imagery is market-neutral in the only sense that matters here: it fails for the same reason everywhere. It has no local reading. In a region where audiences place a face by market and community, that absence is more visible than it would be in London or Berlin, and it is why generation belongs on product work rather than on the people in a campaign.
A practical consequence for anyone planning six markets from one budget. The shoot that produces hero assets does not need to happen six times, but the creator layer usually does need to be local, because the credibility being purchased is local by definition.
How should a brand choose?
Start from the deliverable, not from the budget. Three questions settle most briefs.
Is a face carrying the message? If yes, a real person, and the choice is between model and creator. If no, generation is on the table immediately.
Does the audience need to recognise the person, or only believe them? Recognition points to a creator with an existing following in that market. Belief points to a well-cast model who reads as local.
Where will the images run, and for how long? Paid media across several markets for two years is a usage question, and usage questions favour a booked model with a properly scoped licence. A six-week TikTok push does not.
Most campaigns end up mixing at least two. That is normal, and it is cheaper than it sounds when the shoot is planned as one production rather than three separate procurement exercises.
Frequently asked questions
Is AI-generated imagery cheaper than booking a model? For product work without a face, substantially, and the comparison tables published by generation tools are broadly fair on that point. For anything where a person carries the message the comparison breaks down, because the two options are not producing the same asset. Compare against the job, not against the day rate.
Do we need usage rights for generated images? Not for likeness, provided the output is not derived from photographs of an identifiable person. If it is, the original licence governs, and most booking paperwork written before 2025 does not mention synthesis at all. Check before you generate, not after.
Can a creator replace a model for an e-commerce catalogue? Rarely. Catalogue work needs consistent lighting, consistent framing and a neutral background across hundreds of images, which is studio production rather than creator production. Creators are strong where the informality is the message.
Will audiences in Southeast Asia notice generated faces? Increasingly, and faster than in most Western markets, because the audience is used to reading casting for market and community. A face that fits no market in particular is the tell people report first.
Can Ambressa quote all three? We book models and run creator campaigns, and we say plainly when a brief does not need a person in it. Ambressa does not sell image generation, so that recommendation costs us the booking, which is roughly the point of asking us.
About the author
Mikle Iva is the founder of Ambressa, an influencer and brand ambassador agency working across Malaysia, Singapore, Thailand, Indonesia, the Philippines and Vietnam. He handles the agency's casting briefs and creator selection.
About Ambressa
Ambressa is an influencer and brand ambassador agency working across six Southeast Asian markets: Malaysia, Singapore, Thailand, Indonesia, the Philippines and Vietnam. Ambressa holds a deep roster of commercial models and creators, local, Chinese-speaking and expat, and matches them to the audience a brand is trying to reach. One partner, one brief, work handled end to end.
Tell us what you are shooting and where it will run: ambressa.com/model-agency-malaysia