QC Training
Destage everything. Stage the best. Delete the rest.
Virtual Staging Workflow With Image Cleanup, the full walkthrough. Every section below links to its moment in this video.
What you are delivering
- Every owner occupied home goes out with four kinds of images: the original photos, the virtually destaged (vacant) versions, the virtually staged versions, and virtual twilights for the exteriors. The client gets the full set.
The full set, explained (video starts at 26:28) · open on Loom
- Keep every original. Even when a destaged or staged version is clearly better that the original image, the original image should still stay in the delivery.
- Everything is optional. If a room is giving you too much of a headache, delete the bad versions and move on.
The whole job, in order
Step one: virtually destage every image
- The moment you see a ton of owner belongings in a set, run the destage (virtually vacant) pass on all of the interior images, not just the rooms you plan to stage. You cannot know which angles the AI will nail until you see them, and the vacant versions are a deliverable on their own.
Creating the empty images (video starts at 0:11) · open on Loom
- Do the exteriors too. A backyard with patio clutter or a driveway with cars gets the same pass, and the client gets the original and the cleaned version.
- Destage first, review second, stage third. Staging a photo that still has a mistake in it bakes that mistake into every staged version, so the vacant pass has to be clean before you stage anything.
Why destage everything (video starts at 34:45) · open on Loom
Step two: review the vacant images side by side
- Put each original next to its vacant version and go image by image. This is the slow part, roughly half an hour on a big set the first time, and it gets much faster once you know what to look for.
First pass, TV left behind (video starts at 1:02) · open on Loom
- Check the architecture against the original: walls, doorways, windows, closets, kitchen islands, and ceilings. Anything that moved, disappeared, or appeared is a fail. In the video the AI erased a closet, removed a kitchen island, turned a window into a sliding door, added recessed lights, and painted sky where there was a house.
The closet is gone (video starts at 4:43) · open on Loom
- Check the small stuff too: a smoke detector turned into a light fixture, a TV that stayed behind, vines still on a fence, or a random new light on a wall. These are usually fixable, so tag them instead of deleting them.
- Check the floors. The vacant version has to keep the real flooring. A mat on wood stays a mat on wood. If the AI turned it into full carpet or full marble, that version is a no.
Three buckets: keep, fix, or delete
- Keep: the vacant version matches the original's architecture and flooring, and nothing new appeared. Leave it next to its original and move on.
- Fix: the version is good overall but has one or two small leftovers, like a lamp, a TV, a reflection, or a stray light. Give it a red tag so you can filter to just those later and clean them up by hand.
Red tags for small fixes (video starts at 8:47) · open on Loom
- Delete: the AI changed the room, or the angle was weak to begin with. If the original angle is already bad and the AI messed up on top of it, delete both and move on. If a room came out this bad across every version, delete them all and keep going.
Bad angle plus AI mistake, delete both (video starts at 2:07) · open on Loom
- Some shots do not need a vacant version at all. A plain hallway can ship as the original only. If you have two hallway shots and one vacant version came out great, keep that one and drop the weaker hallway entirely.
Every vacant image lands in one of three buckets
Fix the small leftovers by hand
- Filter to your red tagged images and open them in Lightroom Classic. Use the remove tool to erase what the AI missed, one spot at a time.
- When you remove a lamp or a light fixture, also remove the glow it casts on the wall and ceiling. If the light source stays, the tool reads it as a light and puts a fixture right back.
Remove the light source too (video starts at 14:02) · open on Loom
- When you remove a TV, take its shadow, its shine, and any reflection on the floor with it. Same for a window you are not sure exists: remove the window and the shadow it casts.
- Brush around the furniture you want to keep rather than over it, so the tool understands what is furniture and what is wall.
- If the first result looks wrong, regenerate. Two or three tries is normal. A trick for stubborn spots: draw the selection so it clearly reads as a piece of wall, and the tool tends to fill it as wall.
The draw it as wall trick (video starts at 18:36) · open on Loom
- Know when to stop. If a spot is not bad after a few tries, Delete the image.
- Export the fixed versions, put them back in the folder, and replace the tagged ones.
Step three: sort into rooms and stage
- Before you touch anything else, copy the originals somewhere safe. Move files, never delete them, until the final upload.
- Sort the clean vacant images into their room folders: bedroom, bathroom, living room, dining room, kitchen, office, backyard. The folder you put an image in is the room the AI stages it as, so sorting by hand removes the guesswork. This is the same sorting you already do, only with far more images.
Sorting into rooms (video starts at 23:44) · open on Loom
- Stage generously. Put several angles of the same room into staging, because every extra angle is another chance at a great result and you are only keeping the best one anyway.
Why more angles help (video starts at 36:46) · open on Loom
- Bedrooms and living areas are the must haves. Kitchens, bathrooms, and backyards are nice to have when the angle is good, and you can skip them when it is not.
Step four: pick the winners
- Keep one virtually staged image per room. For living areas keep one, or two if you have two clearly different angles that both came out great. Everything else for that room gets deleted, including good ones. Once a room has its winner, then double check all other images to see if they can beat the winner, if they cant simply delete all the losing images.
One per room, bathroom example (video starts at 29:18) · open on Loom
- Furniture has to sit where furniture would sit. Delete anything blocking a hallway, a doorway, or a walkway, a table pushed into the wall, a couch jammed against the kitchen island, or a rug spilling into the next room.
Blocking the hallway (video starts at 31:41) · open on Loom
- Watch out for fake windows. Staging sometimes paints a window or a door that does not exist, and that is a delete no matter how good the rest looks.
- these are an instant delete: a candle on the floor, a random object in a doorway, a mirror where the door used to be.
Virtual twilights and the final upload
- Every set gets virtual twilights for the exterior shots: the front, the backyard, and any strong side angle. If they are missing, run them. In the video the set had none, and that was the one thing missing from an otherwise solid job.
The missing virtual twilights (video starts at 27:00) · open on Loom
- Do one last pass through the whole folder before uploading: originals, vacant, staged, twilights. Anything that would make you wince if the client saw it gets deleted now.
- Small leftovers you spot at this stage, like a trash can in a backyard shot, get a quick Lightroom fix and a replace, and then you are done.
- Then upload everything. A big set can land at 130 plus images, and that is fine, because the client is getting originals, vacant, staged, and twilights of the same home.
132 images, all good (video starts at 43:55) · open on Loom
Before you upload, check off
- All originals kept, except the ones with people in them.
- Every interior has a vacant version.
- No vacant version with changed architecture or flooring.
- One staged image per room, one or two per living area, no duplicates of the same space.
- No fake windows, blocked doorways, or furniture against the island.
- At least one good Virtual twilight for the exterior.
- Bad images deleted. Do not be afraid to delete bad images..
The whole job in thirty seconds (video starts at 44:46) · open on Loom