Culling a newborn session when every frame looks the same
Newborn is quietly the hardest genre to select from, and nobody says so. The baby does not move, the setup does not change, and you shoot forty frames of one pose waiting for the hands to settle. You come home with hundreds of good photographs that are nearly identical. The skill is not filtering junk. It is ranking near identicals, and here is how to do it without losing an evening.
Almost nothing you shot is bad
Every guide to picking photos assumes a chunk of what you shot is junk and your job is to find it. On a wedding that holds. Somebody blinked, somebody walked through the frame, the focus went to the back of a head. Take the failures out and a gallery appears.
A newborn session breaks that. The baby is asleep, which is the entire point. The light has not moved. The camera is on a tripod pointed at the same beanbag it was pointed at ten minutes ago. You shot forty frames of one pose waiting for the hand to unclench and the mouth to settle, and somewhere in there it did. Nothing in that block is bad. Almost every frame is correctly exposed, in focus, and near enough identical to the one before it.
So the standard advice gives you nothing to do. You reach the end with hundreds of good frames and no way in, and the client cannot help, because from their side of the screen those frames are one photograph printed over and over. This is not filtering. It is ranking near identicals, the hardest kind of decision anyone can be asked to make, which is why the general culling playbook stops working the moment you open a newborn folder. Everyone talks about wedding volume. Volume is not the problem here. Sameness is.
Three habits at the session that make the picking easy
Most of the pain of a newborn selection is created during the shoot. Three habits change the shape of the job before you ever open the folder.
Change one variable per setup
If you move the light, swap the wrap and turn the head all at once, you have not made a new setup. You have made a puzzle: two frames that differ in three ways, and no way to tell which difference you care about. Move one thing, shoot it, then move the next. Later you compare one decision at a time, which your eye does quickly and reliably.
Mark the good frame in camera, the second it happens
You see the fingers relax and the mouth go soft, and you fire. You know right then that this is the one. Press the rating or protect button before the moment leaves your head. It costs a second and saves ten minutes at the desk, because instead of hunting for a feeling you had two days ago, you have a shortlist made while you could see it happening.
Shoot a separator frame between setups
Between setups, shoot one deliberately obvious frame. Your hand over the lens, a gray card, a photo of the empty blanket, whatever you will recognize at thumbnail size. That frame turns a flat strip of hundreds of images into visible blocks. Weeks later, on a phone screen, you can still see where one setup ends and the next begins without opening a file. It is the cheapest trick in this genre.
Work setup by setup, not frame by frame
The mistake is to start at frame one and walk forward, judging each image against a vague idea of the finished gallery. Hundreds of sequential yes or no calls on near identical images burns out your judgment, and research suggests the decisions you make late in such a run are worse than the early ones.
Do it in groups instead.
- Break the shoot into setups. Your separator frames do this for you. Each block is one pose, one wrap, one prop, one lighting position.
- Take the best one, at most two, from each block. Nothing else. Do not think about the gallery total or about variety. One block, one winner, move on. You are comparing like with like, the only comparison your eye is genuinely good at.
- Then look at your winners in a row. Only now can you see the shape of the session. This is where you notice you have four wrapped shots and nothing with the parents.
- Go back for thirds. Ask which setups earned a third image. Usually two or three did, because something genuinely different happened. The rest did not.
The order matters. If you decide up front that you owe the client a fixed number of images, you pad weak setups to hit it, and every padded frame is one you pay to retouch. Pick on merit first, count second.
This is the same group first logic that works for family sessions, turned up. In a family session a group is six frames of a pose. In a newborn session it is forty.
What actually separates two near identical newborn frames
When two frames are the same photograph, the tiebreakers are small and specific. These are the checks, roughly in the order they cost you.
| What to check | What you want | Why it decides the frame |
|---|---|---|
| Hands and fingers | Curled soft and naturally, fingers slightly apart | A hand pinched flat, or a finger bent backward, reads as uncomfortable to a parent instantly. Editing cannot fix it. |
| Mouth | Relaxed, or a soft sleep smile | Mid grimace and mid yawn frames look wrong frozen. Drool trails are extra retouching. |
| Eyes | Fully closed and settled | Half open in the strained way, with white showing, reads as discomfort even when the baby was fine. |
| Skin | The calmest skin in the block | Blotching, mottling and peeling change frame to frame inside one setup, and that difference is retouching time. |
| Wraps and props | Edges straight, no gaps, nothing tilting | A slipped wrap or a prop at an odd angle reads as careless, and warping it straight later is slow. |
| Parent and spotter hands | Relaxed fingers, tidy nails, no visible pressure | White knuckles ruin an otherwise lovely frame, and a parent sees their own hand first. |
| Head and neck | Chin off the chest, ear not folded over | Small compressions look uncomfortable frozen, and cannot be corrected later. |
The rule that keeps you moving: pick the frame where the most things are right at once, not the one where a single thing is perfect. A gorgeous hand attached to a strained mouth loses to the frame that is quietly good everywhere.
In newborn work, picking is your editing budget
A UK survey of more than 300 wedding photographers, reported by PetaPixel, found they spent 55 percent of working time editing and 11 percent culling, against 4 percent shooting. That sample was weddings and predates today's AI tools, but it points at something even more lopsided in newborn work: the retouching is the job, and the picking sets its size.
Newborn skin is the most expensive surface in portrait photography. Blotches, dry patches, peeling and scratches all have to be handled by hand, and they change visibly from one frame to the next inside a single block. So choosing the cleanest skin frame out of ten near identical options is not fussiness. It is buying back real time on every image you deliver.
That is the argument for an extra minute inside a block, and against a large delivery, because in this genre where your hours actually go is decided at the picking stage. Pick twenty images and you have bought twenty units of retouching. Pick sixty because you could not decide and you have tripled the bill you send yourself, for a client who cannot tell the extra ones apart.
Composites, spotter hands, and being honest with yourself
Some of the best known newborn poses only exist as composites. A hand supports the baby in every source frame, and that hand is edited out afterward using a second frame of the same scene. That is normal practice, done that way because it is the safe way.
Do not pick a frame that only works as a composite unless composite work is something you actually do. It is easy to fall in love with the frame that would be perfect if the hand were not there. If you are not going to do that work, or pay for it, that frame is not a pick. Be honest at the decision, not halfway through editing.
If a picked shot depends on a composite, both source frames have to stay flagged. The hero is useless without its clean plate, so mark them as a pair. The moment they get separated in your files, you have lost the shot.
Hands in frame are not automatically a failure
Plenty of studios deliver images with a parent's hands visible and always intended to. A hand supporting a head is only a problem if you were pretending it was not there. If the hands belong in the picture, judge them like anything else: relaxed, well positioned, not pulling the eye off the baby.
How big a newborn gallery should actually be
Wedding numbers do not transfer. A wedding day typically produces 2,000 to 4,000 RAW frames and a delivered gallery of 400 to 800 images, and that gallery is big because the day was big: different light, different rooms, different people, hours of things that happened once.
A newborn session is one room, one afternoon, one sleeping subject and a few setups. Size the gallery by counting setups, not photographs. Two or three images per setup feels complete and considered. There is no correct number for every studio, and how many photos to deliver covers the general question.
Be firm about this: more is worse here. Give a client twelve versions of the same wrapped pose and you have handed your job to somebody less equipped to do it. Classic research on choice, from Iyengar and Lepper through to Barry Schwartz on the paradox of choice, points the same way: more options leave people less satisfied and more likely to regret what they picked. Where the options are near identical, that effect is strongest. A tight gallery reads as confident. A bloated one reads as a photographer who could not choose.
The other frames do not stop existing. They stay on your drives, and there is a good argument for keeping the strong runners up organized, which is the subject of what to do with photos you do not pick.
Where AI is useful here, and where it is close to useless
Newborn is the genre where automated selection sits furthest from the marketing. Closed eye detection is almost worthless. It is the headline feature of most culling tools, built for a subject who is supposed to be awake. Yours is asleep on purpose. Every frame is a blink. A tool that marks frames down for closed eyes is marking your whole session down for succeeding.
Face detection struggles too. A newborn face is often turned away, squashed against a hand or half covered by a wrap, so tools that group by recognizing faces have less to work with than at a wedding.
Blur and focus checks still earn their place. Newborns twitch, you shoot wide open, and a soft eyelash is a soft eyelash.
Grouping near identical frames is the feature that matters. Everything above is secondary to putting the forty frames of one pose in front of you together, so you compare them as a set instead of scrolling past them. That is the whole job here, and the same underlying problem as cleaning up duplicate photos, applied to frames you shot deliberately rather than by accident.
Where Kepla fits, stated plainly
Kepla clears the obvious misses and hands you the clean frames. It never deletes, moves or renames a file, so every frame stays where you put it and anything not picked is one tap away. The part that matters for newborn work is the grouping: see a setup as a set, judge it as a set, move on.
The shape of it is phone first, then Mac. A first pass on an iPhone or iPad breaks a session into setups and marks the obvious winners. The close comparisons here, two frames differing only in how one finger sits, are not a phone job, so the same session opens on the Mac and you keep going at full size with everything still there. More on that tradeoff in phone versus computer culling.
Being straight about status: the picking app for iPhone, iPad and Mac is still being built and has not shipped. What is live today is our free booking page, which holds dates and takes deposits, free while we build the rest.
FAQ
How many photos should you deliver from a newborn session?
Count setups, not photographs. Two or three images from each setup you shot gives a gallery that feels complete without repeating itself. A wedding delivers 400 to 800 images because the day was long and varied. A newborn session is one room, one subject and a handful of poses, so a much tighter gallery is not stingy, it is honest about what actually happened.
How do you choose between nearly identical newborn photos?
Compare them as a group rather than one at a time, then use specific tiebreakers: hands curled naturally rather than pinched, a relaxed mouth, eyes fully closed rather than strained half open, the calmest skin, wraps sitting straight, and relaxed parent hands. Pick the frame where the most things are right at once, not the one where a single detail is perfect.
Why is culling a newborn session so hard?
Because almost nothing you shot is bad. The baby is asleep, the light has not moved and the camera is locked off, so you end up with hundreds of technically fine frames that are nearly identical. The usual advice of setting aside the failures gives you nothing to work with. You are ranking near identicals instead of filtering junk, which is a much harder judgment.
Can AI culling software handle newborn sessions?
Partly. Blur and focus checks work fine. Closed eye detection is close to useless because your subject is asleep by design, and face detection has less to work with when a face is turned away or covered by a wrap. The feature that genuinely helps is grouping near identical frames so you can judge a whole setup at once.
How do you organize a newborn shoot so it is easier to select from later?
Change one variable per setup so frames differ in only one way, mark the frame you liked in camera the second it happens, and shoot one obvious separator frame between setups, such as your hand over the lens. That last habit turns a flat strip of hundreds of images into visible blocks you can still recognize at thumbnail size months later.
Should I keep both frames if a shot needs a composite?
Yes. A composite hero frame is worthless without its clean plate, so flag them as a pair and keep them together in your file structure. And if composite work is not something you do or pay for, do not select a frame that only works as one. Decide that at the picking stage rather than discovering it halfway through editing.
Built for the folder where everything looks the same
Kepla for Mac clears the obvious misses from a card, names the reason on every frame it sets aside, and leaves the choosing to you. Nothing is ever deleted, moved or renamed. Free through the private preview · the first hundred photographers keep it at $99 a year.