What is the optimal batch size for generating AI video clips to save credits?

Generate video clips in batches of 5. That's the cadence documented across invideo agent productions: small enough to review and lock winners before the next batch, large enough to give you real variation per shot. With typical clip-usage rates of 9–26%, generating 15+ at once burns credits on rejects you'd have caught after clip 5.

Batch 5 clips per shot, review, lock the winner, then move to the next shot. The invideo agent is an agentic video tool that routes each shot to the right video model (Seedance 2.0, Kling, Veo, Runway) and holds your project context across the session — so batching small doesn't cost you setup time between rounds.

The yield math is what makes 5 the right number. Across documented invideo productions, clip-usage rates run wide: 1 of 11 clips used on one UGC ad (9%), 10 of 39 on an outdoor product film (26%), 49 of 103 across a 4-ad run (~48%), and a clean 7-of-7 (100%) on a tightly-locked BTS ad. With a 4-second Seedance minimum and Seedance 2.0 capped at 15 seconds per output, generating 15 at once means burning credits on 8–13 rejects before you spot the drift you'd have caught at clip 5.

"Smaller batches just give me more freedom to iterate quickly and early, and more importantly, they save a ton of credits compared to if I was generating all 15 clips at once and then finding something that's gone off," says Hridaye, invideo's creative director. The credit-per-clip cost doesn't drop when you batch bigger — the savings come from curating early and not animating frames that were never going to work.

A few adjustments make the cadence sharper for your situation:

Lock the still frame before you spend video credits. Iterate the keyframe image cheaply, then animate only the locked frame. Across one campaign this kept video credits flowing only to approved compositions — "I only spent video credits on locked frames," per Hridaye.

Drop to 1–2 per shot once your context is dialed in. Later shots in a session land in the first or second try because the invideo agent has learned your taste — by shot four on one production, a single direction note hit on the first generation. Batch big on shot one, batch small from shot three onward.

Let the invideo agent batch the shot list for you. Seedance 2.0's 15-second output cap means a long shot list gets auto-decomposed into batches — you don't sequence this manually. Ask for an estimated generation breakdown (reference sheets, clip count, UI screens) before committing so you see the credit spend before you spend it.

Run two sub-agents in parallel for independent shot sets. If your shots don't depend on each other, spin up a second sub-agent in the same project — same context, no re-upload — and run two batches of 5 simultaneously. One documented production doubled output and halved time this way.

For cost calibration: a 20-second product film used 25 video clips and 12 final cuts at $75 (300 credits); a 30-second multi-character montage with heavy iteration ran 2,100 credits ($530); per-UGC-ad averages land around 500 credits / ~$125. invideo's generative plan prices credits at ~$0.21 each, so a batch of 5 Seedance 2.0 clips lands in the $15–25 range — a workable review unit before the next round.

Watch some of these to see what works for you:

smaller batches just give me more freedom to iterate quickly and early, and more importantly, they save a ton of credits compared to if I was generating all 15 clips at once and then finding something that's gone off

— Hridaye, invideo's creative director