Hampir siap untuk pergi
Memuat...
Mari kita buat sesuatu.

IDE SELANJUTNYA DIMULAI DI SINI

Apa yang akan Anda buat?

Pilih ruang kerja. Buatlah menjadi milik Anda.

Ide Anda bisa dimulai dari mana saja. Merek Anda selalu bersama Anda.

Lewati ke konten

Seedance vs. Kling vs. Google Veo: Which Kept Our Tea Carton on Brand?

Seedance vs. Kling vs. Google Veo: Which Kept Our Tea Carton on Brand?

The perfume bottle looked ready for a product launch. We had asked it to film a carton of tea. We gave three video models the same orange carton and asked for a quiet shot of the label. In one result we could read the blend number. In another, even the packaging material had changed. If you have one product image and one shot to make for a store, start with Kling’s first-frame workflow for a steady packshot. Seedance also kept our carton recognizable and gave the turn more noticeable changes in light. We would check how the reference reaches Veo in Studio before spending another credit on this particular job. Here is what happened across all twelve takes.

The package that made mistakes easy to spot

We drew a fictional carton for MORO Citrus Tea. The front has four things you can check at a glance: the word MORO, an orange dot, a dark green wave and the small line “Blend 07.” A brown side panel gives away any change in angle. Look at that edge when comparing the moving shots below.

The source image for all twelve takes. The right side of the carton is already slightly visible in the illustration.

Follow the brown side panel through the moving shots below.

Orange MORO Citrus Tea carton with a green wave and the words Blend 07.
Original MORO artwork used for all twelve takes.

Two prompts, two takes each

On November 20, 2026, we ran both prompts twice in each of three workflows in Zoviz Pro Studio. We chose 16:9 and 720p in the controls. Seedance 2.5 and Google Veo 3.1 were set to four seconds. Kling 4.0 was set to five. The side-by-side frames come from 0.3, 2.0 and 3.7 seconds, a span shared by all three.

Locked front view. “Keep the product stationary. Locked front view. Soft studio light. Keep the label, spelling, colors and proportions identical to the reference.”

Three-quarter view. “Keep the product stationary. Move the camera slowly from front to a slight three-quarter view. Keep the full product in frame and preserve the label, spelling, colors and proportions.”

The image controls differed. Kling used the uploaded carton as its first frame. For Seedance we attached it as a reference and left the first-frame switch off. For Veo the uploaded thumbnail was visible in Studio. We used the prompts above verbatim, without a numbered image tag. The comparison measures the results of these Studio workflows. We could not inspect what each provider received.

The locked shot: watch the shadow

Kling’s two locked takes came closest to the source illustration. You can follow the orange dot and the right edge without losing them to a lighting change. For a product detail page where the clip sits beside a still, we would choose one of these takes.

Seedance kept the name and the green wave in both takes. The light rose across the scene, and the first take added a shadow that stretched to one side. The carton still reads as MORO. In a loop, though, the change in brightness makes the cut back to the opening frame easier to notice.

Veo’s first take showed an amber perfume bottle. Its second showed a jar with a dark lid and a different label. Both were plausible subjects for a product shoot. Neither showed the carton a customer would receive.

Locked shot · Take 1

Locked shot, take one: Seedance and Kling show MORO while Veo shows an amber bottle.
Locked shot, take one. Read each row left to right. Compare the shadow under Seedance’s carton with Kling’s.

Locked shot · Take 2

Locked shot, take two, with the same reference carton and a different jar in the Veo row.
Locked shot, take two. Repeating the prompt changed the jar but did not recover the reference in the middle row.

The moving shot: follow the right edge

Seedance and Kling both revealed more of the brown right side as the shot progressed. The green wave stayed on the front and MORO remained recognizable. In Seedance’s second take, light swept across the print as the panel widened. Kling’s second take made the turn with less change in exposure. For a catalog sequence we would use Kling. For a social cut where the light change is welcome, we would try the Seedance take.

In Veo’s first moving shot a dark jar turned. In the second it was a white cylinder with a multicolored label. The changing angle gave the shot depth, but rotating the wrong package cannot demonstrate this product. We can see relative motion in these clips. A plain backdrop gives us little evidence about whether the virtual camera or the object moved.

Moving shot · Take 1

First moving take, showing side panels on the Seedance and Kling cartons and a dark jar in Veo.
Three-quarter view, take one. Use the width of the side panel as a simple motion cue.

Moving shot · Take 2

Second moving take: both MORO cartons reveal a wider side panel while Veo turns a white cylinder.
Three-quarter view, take two. Compare the word MORO at two seconds and the width of the brown side at 3.7 seconds.

How many takes kept the product, and what did they cost?

We counted a take if the orange carton, MORO name and green front design were visibly recognizable in the sampled frames. This is a visual identity check. It does not certify every small character in every frame or clear an asset for an ad.

Kling 4.0: 2/2 locked and 2/2 moving takes retained the carton. Four runs at 15 credits each came to 60 credits, or 15 per identity-preserving take.

Seedance 2.5: 2/2 locked and 2/2 moving takes retained the carton. Four runs at 16 credits each came to 64 credits, or 16 per identity-preserving take.

Google Veo 3.1: 0/2 locked and 0/2 moving takes retained the carton. Four runs at 10 credits each came to 40 credits. There was no qualifying take from which to calculate a per-take cost.

The Studio balance fell by 164 credits across all twelve runs. Those prices apply to these settings on the test date. The reviewer could see the model labels. We used one illustrated package and evaluated the visuals only. A photo, a different input mode or a different subject could change the result.

Try the test with your own product

Start with a reference that shows your name, one small line of print and a distinctive shape. Choose one thing that must survive the shot. If that is the packaging, make the locked take first. Pause near the start, middle and end. If the container itself has changed, check the selected input mode and the attachment before paying for another take. Once the still shot works, change only the camera instruction and compare again.

You can start with an image in the Zoviz AI Video Generator. For a walkthrough of the controls, follow How to Make AI Videos in Zoviz Pro Studio.

Dapatkan buletin eksklusif

Berlangganan sekarang untuk mendapatkan pembaruan terbaru langsung ke inbox Anda. Jangan sampai ketinggalan!