All rates below are third-party 2026 figures describing other operators. Where we stand appears in the closing section, and it is a zero.
100 dollars a day is about 3,000 dollars a month. The two phrasings point at an identical bank balance and at completely different working lives.
A monthly figure can arrive in a single lump on the 28th. A daily figure cannot pretend like that. It asks the harder question, which is what happens on an unremarkable Tuesday in February, with nothing launched and no campaign running, that finishes with money attached to it.
That is the subject. Not size - 3,000 a month is small beside what this genre normally advertises - but rhythm. Repeatability and scale are separate properties, and most writing in this category swaps one for the other without saying so.
Here is every route, divided out in public, plus one where the division stopped returning an answer.
Route one: virtual staging, the shortest sum in the survey
You take a photograph of an empty room in a property listing and furnish it in software.
Published figures for this year: the tools cost an operator between 1 and 15 dollars an image. What the client pays lands anywhere from 16 dollars at the cheap end to 75 at the top, so a package of three to eight images bills at roughly 60 dollars and can reach 300. A human specialist doing the same job by hand charges inside that same 25-to-75 window, which means the automated floor undercuts the professional without leaving the professional's band.
Now the figure that explains why anybody buys. Physical staging, meaning actual furniture carried in by actual people, bills the first month of a single listing at 1,500 dollars minimum and 4,000 at the upper end. That is the alternative sitting in the agent's head. Everything in this category is priced against that comparison, never against the cost of a render.
The daily division: at the top of the per-image range, two photos clear the target. At the bottom, six do.
This is the most reachable row here, and not because of picture quality. Listings carry deadlines. Agents hold more than one at a time. A batch of images for a single property is a unit of work that recurs rather than a commission that happens.
Route two: product images, where the margin is comic and the margin is not the point
Numbers first. A usable product image now costs the operator somewhere between ten cents and two dollars. Subscriptions sit at 10 dollars a month up to 50, which resolves to something like five to twenty-five cents an image once volume is real.
Buyers, meanwhile, pay near 25 dollars for a plain listing shot on white and as much as 75 for the same thing done properly, then 100 and upwards, sometimes past 500, for a styled lifestyle frame. Traditional production all in, studio time and retouching included, has been 75 to 150.
Put those beside each other and the spread is almost funny. Fifteen cents of input, forty dollars of invoice.
The daily division: two to four plain listing shots clear 100.
Anybody reading that spread as the opportunity has read it wrong. The gap is a published fact, and the subscription producing it sits on the same pricing page for the person selling the kettle as for the person photographing it. What survives is not the cost difference but whether the seller wants to spend an afternoon learning which prompt yields a shadow that does not float above the surface.
Route three: short-form video, which pays best and pays least often
Creator-style clips for brands. A human creator's standard rate for one short video is 150 dollars at the low end of the usual band and 300 at the high. Entry-level work runs 50 to 150 below that band; premium starts at 300 and climbs past 500. Market-wide, the average deliverable is reported at roughly 198 dollars. A creator who can demonstrate that their videos converted for a previous client is paid 800 for one asset, and up to 2,000.
The base rate is only part of the invoice. Rush turnaround adds a quarter and can add half again. Usage rights add 30 per cent at minimum, half at most, and handing over raw footage carries the same premium. Perpetual rights are the expensive one, adding 100 per cent to the bill and sometimes 150.
Whitelisting is the biggest of the lot. The brand runs its advertising from the creator's own handle, and that access bills monthly - 500 dollars at the bottom, 2,000 at the top, entirely separate from anything produced.
The daily division: one video every second day, at the ordinary rate.
A video every second day is not a Tuesday habit. It is a series of lumps that happen to average out on a spreadsheet. The only line in this entire survey that bills every month is the whitelisting one, and it is paid for access to a person's account rather than for anything rendered.
Two further figures point the same direction. Order a bundle, anywhere from three videos to ten, and the per-unit price falls by a tenth at least and by as much as a quarter against single-video rates - the buyer pays less per unit for precisely the arrangement that makes your month predictable. And once monthly volume clears fifty videos, the render itself costs between 1 and 4 dollars, where a filmed equivalent costs 150 at the very bottom and 600 at the top. Both facts describe one trade: regularity is bought with margin, and the cheaper the render gets the more of the price must sit elsewhere.
Route four: the one that stopped working, in public
Stock photography deserves a section rather than a footnote, because it is the closest thing this category has to a controlled experiment.
The contributor pool is now about 2.5 million people, and between them they add roughly 58 million fresh assets every year. In 2019 that population collectively earned 1.47 billion dollars. By 2026 the same group shares 31 million dollars. The category has lost 98 per cent of what it once paid out.
Nothing about that requires a theory of technology. It requires supply. There is no brief, no deadline and no named buyer at the far end of a stock upload. You are adding to an inventory that already exceeds demand by orders of magnitude, and the same tool letting you add ten thousand files this month lets 2.5 million other contributors do it as well. Price went where price always goes when supply is unbounded and the product is interchangeable.
Hold that against route one. A staged photograph has an addressee: this agent, this property, this Friday. A stock image has no addressee whatsoever, which is why nobody in that market can raise a rate, decline a job or follow up on anything. The tooling in both cases is roughly the same.
So, plainly: any guide telling you to earn by uploading generated images to stock libraries is pointing at a market whose collective payout fell by that 98 per cent while the number of uploaders climbed. We will not publish that recommendation, and I would treat its presence as a signal about the rest of that document.
What actually happened to the wider market
The pattern above is not confined to photographs.
In 2025 the volume of writing work on Upwork came in 32 per cent below the year before, the steepest fall any category on that platform has recorded. Read alone, that is a collapse story. It is not one. Rates for basic and content-mill output dropped somewhere between 15 and 30 per cent over the same stretch, while work described as premium, strategic or humanised rose by 20 per cent at the low end and 40 at the high. Nothing shrank overall. The thing divided, and the halves moved in opposite directions.
Two more figures show where the money went. Measured as a proportion of what companies spend in total, freelance marketplaces went from 0.66 per cent of the budget down to 0.14. Model spending made the opposite trip across those same years: zero at the start, 2.85 per cent by the end. That is not a saving, it is a transfer, out of a line that hired people and into one that does not. Over half the businesses writing cheques to freelance platforms in 2022 had quit the channel entirely by 2025.
The most telling number belongs to the platform itself. Upwork's variable fee is 15 per cent where the supply is commodity - general virtual assistance, basic content - and drops into the 5-to-10 band wherever it is scarce. The intermediary has encoded the divide directly in its own commission, charging most to sell whatever is most abundant. Once the middleman starts discriminating by scarcity, this has stopped being a forecast and become an accounting entry.
Which half you land in is decided before you open the tool
Here is the thesis the figures actually support. The same systems made one half of this market cheaper and the other half more valuable. Which half your work falls into is settled by what the human brings that the model does not.
In staging, that means knowing which rooms sell a property and which are filler, what a buyer in that price bracket expects to see when the photograph loads, and what the disclosure rules covering altered listing images require in the place that agent works. The agent will be asked about that last one. A render has no answer to give.
In product imagery, it means knowing which slot on a listing page does the persuading, and what the marketplace's own upload requirements will reject. Two rejected listings cost a seller more than the whole photography budget.
In video, the 800-and-above tier is not paid for better footage than the 198-dollar average buys. It is paid for a conversion record attached to a name - evidence that this person's work moved a figure for somebody previously. That is a possession rather than a skill, and it accumulates one client at a time.
The pattern across all three is the same. What gets priced is judgment about someone else's business. Rendering is the cheap part everywhere, and it gets cheaper on a schedule you do not control.
Which returns us to the Tuesday. The human contribution is also the thing that makes work repeat. A client who bought judgment has another property next week and rings you about it. A client who bought a render has discovered a subscription and will not be ringing anybody.
What we have earned from any of this
Nothing, and I mean the word literally.
Our own work is automated forecasting agents. No revenue has reached us from virtual staging, product imagery, short-form video or stock libraries, and we run a service business in none of them. Every number above is third-party 2026 material, and it carries that label at each appearance.
The strictness comes from how we handle our own figures. Forecasts get hashed and timestamped before the event they describe, then scored publicly once it resolves, and the ones we got wrong stay on the page beside the ones we got right. Anybody keeping that kind of record cannot then go loose with somebody else's numbers.
100 dollars a day is not a demanding target. Two staged photographs, three listing shots, half a video. What decides it is not whether you can produce those on a Tuesday - you can, cheaply, before lunch - but whether anyone is still asking you for them in March. On the current evidence that question is answered by what you understand about the buyer's business, and never by what you understand about the tool.
Educational content - not financial advice.