BitShovel
中

AI video · Case study

Snow Mountain Fox: how a duck became a story and a store visit

The product became the story's reversal. Selection, rework and audience remixes shaped what the team could make next. The sales gains are maker claims without independent orders or a profit ledger.

Published · Reviewed

BitShovel editorial illustration linking a snow-mountain fox story to prepared duck food in a storyboard; not footage or original production material from Snow Mountain Fox Rescue.Open a related finished video on the brand's Douyin account ↗BitShovel illustration · not film footage
Click the illustration to open a related finished video on the brand account. This is neither a film frame nor the team's actual storyboard; the March 9 Douyin upload is not identified as the Xiaohongshu debut or episode one.Poster source: BitShovel AI editorial illustration, not film footage; the brand's original video is linked in the case · Poster source · Open a related finished video on the brand's Douyin account
Online-store sales · maker statement on April 1Up 70%–75% recently (baseline unspecified)

Ai Linjunji told Yangzi Evening News this figure. A March 26 interview gives separate Tmall month-on-month and promoted-product measures below. The three figures cannot be added or treated as profit. No independent order data, full cost ledger or causal test is public.

Check this result
Read the context and editorial assessment

A sauce-duck seller set out to make routine promotion. Its team borrowed the familiar fox-spirit repayment setup, then brought back the eaten duck for revenge, making the product the punchline. A related finished video is available on the brand's Douyin account, and two reporter interviews record the maker's choices, rework and sales claims. Together they show a path from a short series to product attention, but do not isolate orders, incremental revenue or profit caused by the video.

Follow the projectContents6
01

Verified material

Watch the brand's story before counting its reach

A brand-account video is directly available; the exact debut Xiaohongshu URL remains unverified.

Chuanwa Sauce Duck posted a related Snow Mountain Fox video on Douyin on March 9, 2026. The Hubei Daily interview says the project began as a routine Xiaohongshu update, but we do not have a stable link to that first post. This brand upload is therefore a work entry, not a claimed February 27 debut or episode one.

The maker says the team went on to make five main episodes, one extra, behind-the-scenes material and a music video. The reported 'five billion' refers to discussion and many remixes across the web, not views of this brand video or a count of buyers.

Sources for this section: Chuanwa Sauce Duck: Snow Mountain Fox related video · Hubei Daily interview with maker Ai

02

Public account

Make the product the twist, not a late ad frame

The four-person team changed the character, setting, food and ending so the duck became a story character.

Ai Linjunji says the team saw a predictable AI fox-repayment clip. They changed the farmer to a woodcutter, forest to snow, wild fowl to their own sauce duck, and repayment to the duck's revenge. He chose a recognisable retro wuxia look that he said the tools could generate more smoothly. The product and punchline were designed together, rather than adding a purchase message after the film was finished.

The maker also reports revising the reveal line repeatedly. The transferable decision is to define the audience expectation, what the product does in the story and how the ending breaks that expectation. The original dialogue and character treatment remain the maker's work, not a ready-made template for another brand.

Sources for this section: Hubei Daily interview with maker Ai · Yangzi Evening News interview with Ai Linjunji

03

Public account

Five hours for episode one; generation still needed selection

The maker advises writing the whole story, then splitting it into roughly 15-second parts; no specific failed take is public.

In Hubei Daily's interview, the maker describes writing the whole story, splitting it into roughly 15-second scripts and generating segments. He says AI footage can deviate and needs repeated adjustment and selection to avoid continuity faults; he does not name a failed shot in this title. He mentions Doubao for script breakdown and generation tools for visuals, but no shot-by-shot prompts or version files are public. He told Yangzi Evening News that episode one took about five hours from idea to release. This is the four-person team's recollection, not a reproducible labour or cost record.

We can watch finished footage and read the maker's general selection method, but cannot compare this film's rejected clips against repaired cuts. The reports do not identify a shot that failed in character, prop, motion or continuity terms, so a supposedly successful prompt comparison would be invented.

Documented process

The production chain supported by public material

A finished brand video and two maker interviews; no raw project files or rejected footage.

  1. Rewrite · product becomes the twist

    Changed fox repayment into duck revenge, with a snowy woodcutter and retro wuxia look.

    Check the source for this step

  2. Split · roughly 15-second parts

    Wrote the full story before breaking it into short scripts and generating piece by piece.

    Check the source for this step

  3. Select · guard against continuity faults

    The maker describes repeated adjustment and selection as a method; no rejected shot from this film or exact revised input is confirmed.

    Check the source for this step

  4. Release · respond to viewers

    The team extended the series after episode one; a related finished video remains on the brand's Douyin account.

    Check the source for this step

Evidence limitThis sequence is supported by interviews. It is not a shot-by-shot version history, controlled prompt A/B test or visual comparison of successful and failed takes.

Sources for this section: Hubei Daily interview with maker Ai · Yangzi Evening News interview with Ai Linjunji · Chuanwa Sauce Duck: Snow Mountain Fox related video

04

Public account

The harder part after success: keeping up with remixes

Audience remixes expanded the story and contributed to the original series' pause.

The team told Hubei Daily that viewers took the premise beyond its own ideas, yet it finished five main episodes and an extra. On April 1, Ai told Yangzi Evening News that the original series had stopped: production time and cost were rising, and the team doubted it could top audience remixes. It began another AI short series. The pause is a problem of sustaining quality and ideas, not evidence that video generation itself failed.

A small trial would decide which characters, visual rules and product facts must persist, then use real response to one episode to judge the next investment. That is BitShovel's inference from the case, not the team's recipe for a hit.

Sources for this section: Hubei Daily interview with maker Ai · Yangzi Evening News interview with Ai Linjunji

05

Public account

Read the three sales claims separately

March 26 has Tmall and promoted-product measures; April 1 has an online-store gain without a defined baseline.

In the March 26 Hubei Daily interview, the maker said duck sales at the company's Tmall flagship store were up 10%–15% month on month. Later in the same report he said sales of the duck he promoted had risen by more than 20%. The second statement gives no platform, comparison period or overlap with Tmall, so it is not another distinct block of incremental sales.

On April 1, Ai told Yangzi Evening News that recent 'online-store sales' were up 70%–75%, without identifying the store, comparison period or exact products. All three figures are maker statements recorded by reporters. We have no independent orders, media spend, returns, inventory, labour or full production ledger, and no control test that separates the videos from other causes. His claim that the whole series cost a little over RMB 1,000 also lacks an itemised scope. None of this permits an incremental-revenue or profit calculation.

Sources for this section: Hubei Daily interview with maker Ai · Yangzi Evening News interview with Ai Linjunji

A BitShovel exercise · not the maker’s tutorial

Test a story turn for your own product

BitShovel designed this trial from the team's public account of a product-led twist, short segments and audience response. Make one original episode for a product you may sell, check whether viewers understand and want to learn more, then decide whether a series is worth making. This is not the team's shot-by-shot process or a hit formula.

Before you begin

  • Use a product you sell and can fulfil, with permitted real-product photos, marks and factual information. Create your own characters and story rather than copying the fox, woodcutter, duck dialogue or film.
  • Have access to one store's same-product order and visit measures, with price, promotion and stock notes. Without store data you can still test the story, but cannot report a sales effect. Set a film length and production cap first.

  1. Give the product a real story role

    List three checkable product facts and one ordinary use setting. Choose an original, familiar situation with room for a turn: what viewers first expect, how the product changes the action and how the ending pays it off. Let the product affect the plot rather than appear only in a final ad frame.

    Keep
    A one-page story card and fact list, with support for each product claim.
    Check
    Would the viewer understand why the product matters without the brand name? If removing it leaves the plot unchanged, rewrite the turn. Leave unsupported performance claims out.

  2. Break the whole story into checkable beats

    Write the whole episode first, then split it into expectation, action and reversal beats. Mark person, setting, actual product appearance and prop state for each. The maker discussed roughly fifteen-second segments; choose a length that fits your story and destination.

    Keep
    A three-beat script, still storyboard and product-continuity sheet.
    Check
    Can the same product be recognised across stills? Do packaging, amount and use remain consistent, and does the joke work without an explanatory caption?

  3. Make one episode and keep the rejected takes

    Generate or film the beats using your own product reference, within a preset attempt cap. Save input, tool, version, spend and rejection reason, then edit usable pieces into one episode. Log faults by person, product, motion, sound and continuity; do not attribute imagined errors to the case makers.

    Keep
    One pilot episode, a rejected-take list and a time and tool-spend log.
    Check
    Is the product shown accurately, without misleading claims, and do character and setting connect? If the key turn remains unclear after two rounds, return to script and board rather than generating a series.

  4. Listen to what viewers understood

    Publish this episode on a channel you control, saving the source URL, date, cover and title. Record platform-defined views, completions, comments and visits to the product entry. Separate comments about the story, product recognition, purchase questions and unrelated remixes.

    Keep
    A first-episode response sheet with any actual viewer questions; record 'no relevant feedback yet' when there are none.
    Check
    Are comments responding to your film or someone else's remix? Have you mistaken views or topic heat for buyers? With tiny samples or no product interest, revise before making episode two.

  5. Separate store trends from series decisions

    Compare visits and orders for the same product and store over stated dates, noting promotion, price, stock, media spend and returns. Calculate actual pilot cost. Describe same-period changes separately from any claim that the film caused them, then decide what new viewer question a second episode would answer.

    Keep
    Like-for-like store records, a production cost sheet and a written continue, revise or stop decision.
    Check
    Without comparable periods, independent orders or a control, report observations only, not incremental sales or profit caused by the film. If a sequel repeats the joke while cost rises, stop the original series or change direction.

Make one final choice: Stop if product facts or usage rights are unclear. Revise the script if the stills cannot convey the product and turn. Try another episode only when actual viewer questions and affordable production support it. Use your own response and accounts for a series decision, not other people's remix heat or interview sales claims.

This is a BitShovel exercise for a merchant making its own content, not the Chuanwa team's version history. The three sales statements come from the maker and use different measures; they cannot be added or prove sales growth or profit caused by Snow Mountain Fox.

Sources and limits

Where the material comes from

  • Chuanwa Sauce Duck: Snow Mountain Fox related video
    Finished video on the brand's Douyin accountPublished Checked

    Confirms the brand account, finished imagery and March 9 post; it is not proof of debut or series-wide metrics.

  • Hubei Daily interview with maker Ai
    Firsthand reporter interviewPublished Checked

    Maker statements on the four-person team, five episodes, script splitting, general selection method and two distinct March 26 sales measures.

  • Yangzi Evening News interview with Ai Linjunji
    Original newspaper interview; conducted April 1, published April 2Published Checked

    Confirms the company and maker names, 70%–75% store claim, episode-one time, reported series cost and pause; no sales or cost records attached.