Naar de inhoud
CollectAsist

AI culling

AI photo culling guide; starting a cull, the Culled and Suggestions tabs, strictness, target frames, keyboard shortcuts, approval and credit cost.

Status: LiveLaatst bijgewerkt: Functiepagina

De documentatie is geschreven in het Turks en het Engels. Deze pagina wordt in het Engels getoond.

Op deze pagina

AI culling does the first pass before delivery for you: it reviews the event's frames one by one, flags the weak ones with a reason and groups bursts. The final decision is always yours. No frame is ever deleted.

You start a cull from the event's page, review the results on the review page and approve. Making decisions requires the manage events permission; team members without it can only view a cull.

What you can do today

  • Open a cull with Start culling on the AI culling card of an event's page, and see the frame count and the credit quote before it starts.
  • Cull Professional photos or Guest photos.
  • Review the results in the Culled, Suggestions and Kept tabs, and filter by reason, content, shot type and part.
  • Change the Strictness (Light, Medium, Strict) and the Target frames count afterwards.
  • Mark frames one by one or in bulk with Keep, Cull or Leave to AI, and move through enlarged frames with the keyboard.
  • Approve with Approve cull or stop with Cancel culling.
  • See all your culls and their status on the AI culling page in the menu.

Starting a cull

  1. Open the event's page in the panel and find the AI culling card.
  2. Click Start culling.
  3. Under Which photos should be culled? choose Professional photos or Guest photos.
  4. Check the Frames count, the Asist Credits it will use and Your balance in the window.
  5. Choose the Strictness. Medium is the recommended level.
  6. Click the start button, which shows the number of credits.

If the event has nothing to cull, the card says "There are no photos to cull. Upload photos first." An event has one open cull at a time; while one is open the card sends you to it with Go to culling.

The review runs in the background. You can close the page; a notification arrives in the panel when it is done.

How long the review takes depends on the number of frames; a small set is ready within minutes. During this time the cull shows as Reviewing, then moves to Awaiting your approval.

What culling checks

Each frame is measured separately. When measuring ends, the frames are evaluated together in shooting order and each one gets a score from 0 to 100. A flagged frame shows its reason:

ReasonExplanation
Empty / black frameA frame with no content
Flash didn't fireA frame where the flash did not fire or fired late
BlurryThe face, or the sharpest area of the frame, is not sharp enough
Main subject's eyes closed, Main subject's eyes half closedEye state of the event's main people
Someone's eyes closedClosed eyes on a person who is not a main subject
Too dark, OverexposedExposure is far too low or too high
Face cut off, Face coveredThe main subject's face is cut by the edge or behind something
Looking away, Mouth openPassing expression problems in portraits
Noisy (high ISO), Tilted horizonClear noise, or a frame shot slightly tilted
Better frame in burstSurplus copies of nearly identical frames shot within a few seconds
Test / grey cardTest shots and grey card frames
Outside the target countA low-scoring frame culled because of the target frames limit

Main people are worked out from the faces that appear most often at the event. The eye and expression rules are applied to them first. The face data used for this is deleted when the review ends.

The review page

When the review ends, the frames arrive in three tabs:

  • Culled: frames that will be culled. Nothing is deleted; select the ones you want back and click Keep.
  • Suggestions: frames the AI is unsure about. They are kept unless you cull them.
  • Kept: frames with no issue found; best first.

The filters above the tab narrow the frames by reason, by Content label (Ceremony, Rings, Bouquet, Cake, Dance, Details, Decor, Food, Venue), by shot type (Couple, Crowd, Detail, Group, Portrait) and by Part. Frames from a burst carry the Burst label.

There are three ways to decide:

  • One by one: select a frame and choose Keep, Cull or Leave to AI. Your decision takes precedence over the rule; the frame is marked "Kept by you" or "Culled by you".
  • In bulk: select the frames shown and use the same buttons, or choose Cull all suggestions.
  • By reason: when you filter a reason you can cull or restore every frame flagged with it in one step.

Culling is a suggestion. An important moment caught in a single frame, or a frame that is blurry or dark on purpose, can be flagged by mistake. Look through the Culled tab before you approve.

Keyboard shortcuts

When a frame is enlarged you can work with the keyboard:

KeyAction
← →Move between frames
KKeep
XCull
ULeave to AI
EscClose

Strictness and target frames

When you change the Strictness on the review page the results are recalculated; the frames are not measured again and no extra credits are charged.

StrictnessWhat is flaggedKept from a burst
LightOnly obvious problems: empty frame, missed flash, blur, main subject's eyes closedThe best 3 frames
MediumRecommended. Also flags half-closed eyes, open mouth and cut-off framingThe best 2 frames
StrictStrictest thresholds; also flags looking away in portraitsOnly the best frame

If you enter a number under Target frames and click Apply, the lowest-scoring frames are culled too whenever more frames remain than that number, starting with the suggestions. Remove limit takes the target away.

Over time the system adapts to your decisions. If you kept most of the frames flagged for a certain reason in earlier culls, that reason on its own no longer culls or suggests a frame in later jobs; the review page shows this with the note "More lenient based on your past decisions".

Approving and cancelling

  1. Look through the Culled and Suggestions tabs.
  2. Click Approve cull. The confirmation shows how many frames will be kept and how many will be culled.
  3. Confirm. The cull moves to Approved.

Approval marks the culled frames; it deletes no file. Culled frames stay in the event and can be restored. On a standalone CollectAsist account approval only does this marking; no transfer starts.

A cull you do not want to approve can be closed with Cancel culling. The credits spent on a cancelled cull are not refunded.

Credit cost

ItemCost
Culling1 Asist Credit per started 100 frames
Minimum10 Asist Credits per cull

Example: a cull of 3,000 frames costs 30 credits, and a cull of 400 frames costs 10. Credits are deducted when culling starts and match the quote you saw in the start window. If the frames cannot be reviewed, the credits are refunded. Changing the strictness, setting target frames and marking single frames use no extra credits.

If your balance is too low the window says "Your Asist Credit balance is not enough." For packages, see Credits and buying.

Frequently asked questions

Does the AI delete my photos?

No. No frame is deleted. Culling only marks frames; culled frames stay in the event and can be restored.

Where do culled frames go after I approve?

Nowhere. Approval marks them as culled. The files stay in the event.

Do I pay again if I change the strictness?

No. Frames are measured once; changing the strictness or the target frames only recalculates the result.

How long does a cull take?

It depends on the number of frames. A small set is ready within minutes; for large sets you can close the page and wait for the notification.

The card says "AI culling is not available right now". Why?

The feature is not enabled for your account. Contact us.

CloudAsist culling session and transferBèta

For accounts linked to StudioAsist the designed flow starts the cull as a session opened from a CloudAsist folder in StudioAsist, and after approval the kept frames are transferred to CloudAsist. The StudioAsist side of this flow is not ready yet. A transfer starts only on an account linked to StudioAsist; on a standalone CollectAsist account approval only marks frames. See StudioAsist integration for details.