Carl, 2026-08-07, on the hedge-and-lawn photo in the About section: *"the image is kinda bad,
i want you to look through 20 of the best prized and highly regarded having a great brand and
look through what they use for images, how they look, feeling, taste etc, and incorporate that
knowledge, to help you gather better stock footage or generate better images."*
He is right about that specific image, and naming why matters more than replacing it: **it is a
photograph of a place, not of work.** A hedge and a mown lawn on a grey afternoon. Nothing is
happening, nobody is present, there is no evidence of skill, and it could be anyone's garden.
It fails the only test that matters on a trade site: *does this show me the thing I am paying
for?*
The strongest single finding from research is that **intentional imperfection now reads as more
trustworthy than polish**. Squareshot's 2026 product-photography trend work states plainly that
"a few rough edges build more trust than a flawless finish", and attributes it to buyers being
tired of imagery that looks too good to be true, with real fabric wrinkles, subtle lighting
variation and natural shadows reading as honest
(Squareshot, Product Photography Trends 2026).
That matches what we found the hard way: our first warm-light generation looked "too perfect"
and read as AI immediately.
Second verified point: art direction, not equipment, is the differentiator. Premium imagery
comes from lighting, styling and post-production aligning under a single direction rather than
from camera settings
(Squareshot),
and natural light is what carries texture and true colour
(Shopify).
Third, and the most useful for us, is Patagonia's actual method: since 1980 they have sourced
photographs from customers and working photographers in the field, and their catalogue gives
roughly half its space to non-selling editorial imagery of people doing the activity
(Patagonia photography retrospective). The pictures are of
the work, taken during the work, by people doing the work. That is exactly the register a tree
surgeon's site needs and it is the opposite of a stock garden.
Everything below this line is my own read of the reference brands rather than something I found
written down. Treated as judgement, not fact.
Grouped by what we can actually steal.
Work as the subject (the core group for us)
Craft and material
Place and atmosphere
Editorial discipline
Nine, in priority order. These now drive both stock selection and generation.
A hedge being cut, or a cut hedge with the tools still leaning against it, is evidence.
This is the rule the flagged image broke.
because a recognisable person implies staff we cannot vouch for.
"f2.8" are render words and produce renders.
Verified as a trend, and it is also what our own testing found.
of new fence board. These make the best feature and texture images.
badly and naming a thing summons it. Describe the swept site instead.
often as golden. A Californian-looking garden is instantly wrong for Gloucestershire.
Every site uses at most three images, and they do different jobs. This is why one weak photo
hurts so much: at three images each carries a third of the visual argument.
| Slot | Job | What earns the slot |
|---|---|---|
| Hero | Prove competence in one frame | Work in progress, human cropped in, depth |
| Feature (About) | Prove a person | Hands, a tool in use, close, warm |
| Texture | Carry a section without competing | Material close-up, near-abstract, no subject |
The image Carl flagged was in the feature slot, which is the "prove a person" slot, and it
had no person and no work in it. It was a texture image doing a feature image's job.
and no human present, which is what let the hedge photo through.
worker, head out of frame, positive description of the finished state, no negation) and gains
the subject rules above: work in progress over finished emptiness, texture close-ups for the
texture slot, English planting.
is hero, feature and texture only, and the footer discloses it.