AI video for finance creators: one analyst character across weekly chart clips

Finance creators on AI video run into the same wall every time. The avatar tools let you lock a presenter but cannot cut the chart overlay into the shot. The generative engines render a chart beautifully but drift the analyst's face by scene three. The articles ranking for "AI video for finance" are listicles of avatar tools and generic workflow blocks, with no credit math on the one thing a channel actually needs: a locked analyst, a chart, and a budget that holds for twelve weeks.

Here is that math on our shelf. One analyst character, one weekly cadence, one plan, the real credit cost of a four-episode month.

The shot inventory, one episode

A weekly finance clip for feed is almost always the same six beats. Numbers vary — the beats don't.

scene	length	content	engine	reason
1 — title + ticker	4s	silent title card over a market room	Veo 3.1 Fast (silent)	4-second ceiling, place-first
2 — analyst opener	8s	locked analyst delivers the week's thesis	Kling O3 Pro (voiced)	element-locked identity, voiced
3 — chart close-up	6s	silent macro, chart overlay in post	Veo 3.1 Fast (silent)	native camera, no analyst in frame
4 — analyst analysis	15s	locked analyst continues, two-shot possible	Kling O3 Pro (voiced)	15s one-generation ceiling, named element
5 — second chart	6s	silent close-up of the second chart	Veo 3.1 Fast (silent)	matched to scene 3
6 — analyst CTA	8s	one line, "see the full breakdown in bio"	Kling O3 Pro (voiced)	element-locked for identity match

Total 47 seconds of finished video, six scenes, two engines on the shelf. The engine routing is not accidental: the shots the analyst is in are Kling because the element lock holds the identity across the three analyst scenes; the shots the chart is in are Veo because the generation returns fast and the compute is lighter on the place-first frame.

The credit math on one episode

step	detail	credits
writer: brief → scene list (convert)	free	0
writer: six scene-card titles and prompts	6 × 5 cr	30
storyboard: six 4-second 480p drafts (Seedance 2.0 Fast)	6 × 4s × 25 cr/s	600
scene 1: Veo 4s silent 720p	4 × 20 cr/s	80
scene 2: Kling 8s voiced 720p	8 × 30 cr/s	240
scene 3: Veo 6s silent 720p	6 × 20 cr/s	120
scene 4: Kling 15s voiced 720p	15 × 30 cr/s	450
scene 5: Veo 6s silent 720p	6 × 20 cr/s	120
scene 6: Kling 8s voiced 720p	8 × 30 cr/s	240
caption: 47s = 1 started minute	1 × 8 cr	8 (free on Creator up)
export: one timeline	flat	20
episode total (Lite / Starter)		1,908
episode total (Creator up, caption free)		1,900

At Lite's one cent a credit that is $19.08 per episode. On Creator's 0.75c a credit, $14.25. One episode a week over a four-episode month is 7,600 credits on Creator, inside the 6,000-credit monthly grant with a 1,600 top-up — or two episodes a week at 15,200 credits, which is a Studio plan at 15,000 credits for $99/month.

The useful anchor: the Creator plan at $45/month covers roughly three weekly finance episodes, which is twelve a quarter. Studio covers eight a month, which is two a week. Agency at $299 for 37,500 credits is twenty episodes.

Why Kling for the analyst

Named element reference. The analyst is @Element1 in every Kling render across every episode. The character sheet is a reference pack of four angles — frontal, three-quarter left, three-quarter right, back — plus the one piece of clothing viewers will recognise. The pack sits in the picker and loads into every analyst scene the show runs. The element binding is the engine's unit of memory, and the identity holds through pans, dollies and whip pans without drifting.

15-second one-generation ceiling. Kling caps a single generation at 15 seconds. Our 30-second Kling renders are two 15-second shots joined; the join is almost always invisible on a locked frame or a conversation and sometimes shows on a long pan. For an analyst speaking to the camera the shot is almost always under 15 seconds in one piece, so the join does not come up.

Voiced 30 credits per second at 720p. The cheapest voiced second on our shelf for a character-locked shot. On a weekly cadence, every credit matters; Kling is the engine where identity, length and price all land at the same row.

Silent-discount for b-roll. If a scene wants the analyst looking at a chart without speaking, Kling at 24 credits a silent second is 20% cheaper. A 10-second silent reaction cut costs 240 credits instead of 300. For a channel that cuts to analyst reactions over a chart in Scene 5 style, the discount compounds.

Why Veo for the chart beats

Native audio even silent. The 20 credits-per-silent-second rate on Veo 3.1 Fast at 720p is the cheapest silent second on the shelf, and the engine renders a place-first shot — a trading room, a monitor wall, a city skyline at open — cleanly. For a 4- to 8-second establishing or chart-adjacent frame, Veo is the row.

4, 6 or 8 seconds, pinned. Veo's duration enum is three values. For the title card at 4 seconds and the chart close-up at 6, Veo quotes exactly those lengths and the provider renders exactly those lengths. On a weekly channel, no second is slop; the enum works in your favour.

1080p at the same price as 720p. If the chart beat is going to a client cutdown or an investor deck at 1080p, Veo quotes the 1080p render at 30 credits a voiced second — the same as 720p. For a feed-first channel 720p is almost always the row, but the 1080p ceiling exists on the same engine without a price hike.

The one thing nobody will show you until you've shipped twelve

Lock the chart overlay in post, not in the generation. We tried both. The first six episodes we rendered with a chart burnt into the prompt; the chart drifted on every third shot, numbers warped, axes shifted, and the brand recognition we were building eroded. From episode seven on we rendered the shot without a chart — a generic analyst-at-screen frame, or a generic room — and composited the real chart in the editor. The analyst stays locked and the chart stays legible. The chart overlay is a design asset, not a generation; the engines are for the frame underneath.

The second rule is adjacent: use the same analyst reference pack for every scene across every episode. The pack of four angles plus the wardrobe close-up is one upload. Save it in the picker and never regenerate the sheet. The identity you lock in week one is the identity viewers recognise in week twelve, and no amount of per-prompt descriptor is a substitute for the same five reference frames across every render.

The budget, over a quarter

Three months, one weekly episode: twelve episodes × ~1,900 credits each = ~22,800 credits. On Creator's 6,000-credit monthly grant across three months you'll run short of 4,800 credits, which is one $50 top-up ($50 for 6,000 credits) — total $185 for the quarter.

Two weekly episodes over a quarter: 24 episodes × ~1,900 = ~45,600 credits. On Studio's 15,000-credit monthly grant across three months you'll run short by 600 — one small top-up — total $307.

Either run is cheaper than hiring a presenter for one shoot in most markets, and the twelfth week looks like the first — same analyst, same voice, same cadence — because the identity sits in a reference pack, not a weekly audition.

Our own 9.7 million views across 30 days on @cevrico came from a serialized character format cousin to this one. Own results — your numbers depend on what you say about the market and how far you push the storytelling. A free account comes with 300 credits, which is enough for one Kling 8-second voiced analyst opener and the start of a reference sheet. Open the shelf and lock a character.