# Open-sourcing the paid version too, for 170 stars in a week: NovaNova Studio and the crowded "AI comic drama" open-source track · GitHub open-source projects

> Source: https://stringers.io/en/github/gh_Swayingleaves__novanova-studio.html (Stringers / 什么值得做, generated 2026-11-08)
> Stringers answers one question every day: what small product can one or two people build in a few months and actually make money from. Numbers come from 15 databases we collect daily, each with source and date.
> For agents: these verdicts and the underlying databases (verified revenue, App Store / Google Play charts, ad libraries, GitHub, Product Hunt…) are queryable live over MCP — tools today / hot / products / directions / direction / data. Setup: https://github.com/komako-workshop/stringers (the user logs in at stringers.io → "配置 agent" to get a personal URL). Site index: https://stringers.io/llms.txt

> On Nov 1 they open-sourced the previously paid enterprise edition in full, switched the license to MIT, and posted on LINUX DO the same day (101 likes). Stars rose about 50 each on days two and three; combined with China's AI micro-drama boom, the whole category grew.

[GitHub open-source project](https://stringers.io/en/github/index.md) · written 11-6

Swayingleaves/novanova-studio · Swayingleaves

# Open-sourcing the paid version too, for 170 stars in a week: NovaNova Studio and the crowded "AI comic drama" open-source track

**What it is** An open-source AI comic drama / short drama creation platform: image generation, video generation, an infinite canvas, novel-to-script conversion, a 3D director stage, and a built-in user and credit billing backend, written in Java.

**Why it caught on** In Nov 1 it open-sourced all of its previously paid enterprise edition and switched the license to MIT, posting about it the same day on LINUX DO (101 likes). Stars rose by roughly 50 on each of the next two or three days. Add the domestic AI comic drama boom, and the whole track was rising.

**C** **How you could follow** Don't build another all-in-one comic drama platform. Build the single-point tools the leading platforms lack, or switch to overseas English audiences or a specific genre. When launching, copy its move: "open-source the paid version + a candid community post."

Grade C: the track is already crowded (leaders at 20k stars, 136 similar projects created this year), and it itself only got a few more than 600 stars in four months with no clear differentiation. What you can learn is the launch approach; copying it or wrapping it in a different shell will still struggle to earn stars. Spreading in small circles

Data (updated daily) · Repo created 2026-07-22 · 695 stars · 7 days +186

GitHub · stars

Data 2026-10-09 → 2026-11-08 (31 days, 24 observations). Points before 10-28 come from deeply / archive backfill, not every day has one

On Nov 1, a little after 2 p.m. Beijing time, an AI comic drama platform called NovaNova Studio posted on LINUX DO under the title "I made a decision that goes against our ancestors": it merged the code of its paid enterprise edition fully into the open-source version and changed the license from GPL3 to MIT. Over the next two days it gained 50 stars on GitHub each day. What's worth looking at is not the project itself, which ranks in the middle-to-low tier of its track, but how much attention the act of "open-sourcing a paid version" can bring, and why it brought relatively little.

### What the post said

The post isn't long. The author said the project, when July open-sourced, used GPL3, with a split between an open-source edition and a paid enterprise edition, and that "the more we built, the more we realized that not being fully open-source gave no advantage and failed to attract community developers (not a single PR). In the AI era there is no code moat, so it's better to go fully open-source." The v1.0.0 released the same day had release notes of only three lines: fully open-source, merge the enterprise code into the open-source edition, change the open-source license to MIT.

The ending is even more candid: "Many people ask, with so many open-source AI comic drama projects out there, what is your advantage? We thought about it for a long time and couldn't name a single one... we might as well build a platform with 'nothing' at all."

By 11-06 the post had 1,200 views, 101 likes, and 21 replies. Most replies were things like "liked first, will read later, support open source" and "already starred." The author self-reported in the post: just over 500 GitHub stars, about 1000 users on the website, and "some friends used the platform to make videos with several million views" (none of this was verified).

### What happened on GitHub

The repo was created in Jul 22, and before this it had grown naturally at three or four stars a day:

| Date | star | fork | What happened |
|---|---|---|---|
| 09-16 | 277 | — | First promotional post on LINUX DO (self-reported by the author) |
| 10-20 | 450 | 56 | Growing three or four a day |
| 11-01 | 509 | 65 | Release v1.0.0, switch to MIT, post |
| 11-02 | 558 | 86 | +49 |
| 11-03 | 608 | 101 | +50 |
| 11-05 | 618 | 104 | Stopped |
| 11-06 | 672 | 105 | +54, source not traced |

(GitHub, daily star / fork counts; later on 11-06 it was 689)

It rose 174 in a week. 11-06 On that day, 54 of them have no matching post. All that's visible is a batch of SEO scraper sites reposting its intro article, possibly forwarded from WeChat Official Accounts or group chats. Unverified.

Forks grew faster than stars: from 65 to 101 in two days. My inference is that this is directly tied to "the enterprise edition is also released": it ships with user management, credits, billing, and OAuth, so you can reskin it and run your own paid AI image and video site, and some people forked it for exactly that.

### What it is

An AI creation platform written in Java (Spring Boot + AgentScope): image generation, video generation, an infinite canvas, novel-to-script, a comic and short drama workbench, a "3D director stage" where you can arrange mannequins and camera angles, plus a complete user and credit backend. The selling point is "Agent throughout": you talk to one main model, and it plans and calls the various generation models for you.

The first screen of the README says "Deployment is a hassle, finding cheap channels is a hassle → click to try online," links to its own live site novanovastudio.cn, and sells cheap video-model API channels (Seedance 2.0 "as low as 4 off", MiniMax at 6 cents per second), with two sponsor slots and a WeChat QR code for "enterprise cluster edition and technical support." The repo is the entry point for these businesses.

### Why it only grew this much

**The track is too crowded.** With the domestic AI comic drama boom in 2026 and video models getting cheaper, individuals and small teams wanting to make comic and short dramas at low cost suddenly multiplied, and open-source projects surged alongside them. This year alone, 136 new GitHub repos with "comic drama / short drama" in their descriptions were created, and 21 of them gained over 100 stars in the last 7 days (GitHub data, 11-06). In one list called awesome-manyu, the "all-in-one platform" category alone includes over 20 projects.

| Project | star | past 7 days | How NovaNova differs |
|---|---|---|---|
| Toonflow-app | 20.8k | +874 | Leader, also an infinite canvas + Agent + storyboarding, ahead by about half a year, and also has an app |
| huobao-drama | 17k | +303 | Go + Vue, fully automatic short drama from one sentence, no commercial use allowed |
| waoowaoo | 14.8k | +80 | Once gained 6.8K stars in 4 days |
| printfilm | 7,400 | +56 | Created in September, also has a built-in billing backend and MIT license, with 10x NovaNova's stars |
| Yingce open-ai-canvas | 1,675 | +193 | An infinite canvas workbench created on the same day as this repo |
| Ronguang ai-fusion-video | 1,710 | +41 | Also Java + AgentScope; NovaNova's author forked it in June |
| alibaba/lumenx | 1,453 | +26 | An Alibaba open-source novel-to-comic-drama pipeline |

(GitHub star counts, 11-06)

**It itself can't articulate what's different.** The only things that can be counted are the Java stack (only Ronguang shares it), conversational Agent, the 3D director stage, and the full operations backend. The first two mean little to ordinary users, and the last two are also being built by others.

**The event only spread in one community.** LINUX DO is a Chinese community that is very friendly to "open source" and "free riding." A post with 101 likes brought in nearly 100 stars, and the conversion was decent. But afterward the same article was only reposted on CSDN, Juejin, SegmentFault, OSChina, and Cnblogs, with nothing on X and no big accounts covering it. The buzz stopped after two days.

### Three things worth learning

1. **"Open-sourcing the paid version" is itself a story.** When open-core isn't working, releasing all the paid parts under the most permissive license, with a post about the journey, can earn more stars than just shipping another version. NovaNova's post title and the candid "couldn't name a single advantage" paragraph were the only highlights of this episode.

2. **A post for every version.** Starting from v0.1.6, the author posted on CSDN with every release, mirrored to Juejin and OSChina, and scraper sites reposted in bulk, so long-tail exposure kept going.

3. **The repo can be the storefront for a business.** A live demo site, cheap model API channels, sponsor slots, enterprise support: stars don't earn money, but these can.

### If it were you

**One-line verdict:** Stop building another "full-pipeline AI comic drama platform." On this path the leader has 20K stars, a new repo appears every week, and NovaNova took four months to reach 600 stars. What you can take from it is its launch playbook, not its topic.

**Who it suits:** developers with a project that's been around for a while and stuck at a few hundred stars, who have a paid tier they can open up; or people in the AI video space willing to build one small piece nobody has done well yet.

**Who it doesn't suit:** anyone wanting to build a feature-complete comic/short-drama platform from scratch. You'd be ranked behind Toonflow, huobao-drama, printfilm and dozens of similar projects. Also not for people with no experience posting in Chinese dev communities who only hope for organic GitHub traffic.

**How to do it:**

1. Don't build a platform, build parts. Everyone in this space is racing to be the "all-in-one" tool. What's missing are single-purpose tools that those platforms can plug in directly: character consistency, storyboard-to-CapCut-draft export, a unified calling layer for a video model, or a solid 3D camera-placement component. Single-purpose tools are easy to explain, and they're more likely to get linked from top projects' READMEs or awesome lists (inferred).

2. Change the audience: NovaNova targets "Java developers," which is too narrow. If you switch, go for an English version aimed at overseas creators, or a tool for a specific genre (say, novel-to-video recaps or children's picture books). At least you'll have something to say that nobody else does.

3. Put a try-it-now online demo and effect videos on the first screen. For things like comic dramas, seeing a finished piece is worth far more than reading a feature list.

**How to get users:** Copy its posting move: publish in a developer community like LINUX DO using the "open-source promotion" format, with a title about a decision or event and a body that tells the truth; mirror each release to CSDN and Juejin. Fill the gaps it left: post an English version on X and Reddit, and get AI video creators to record a finished-video demo. It's an empty field overseas.

**When the window is:** open-source AI comic-drama projects keep popping up every week, and the leaders are still growing fast. The window for building parts is probably the next two or three months. Once the leading platforms ship all these features themselves, nobody will need the parts (estimate).

**What would mean you read it wrong:** two weeks after posting, you're still under a hundred stars, or the feature you built gets built into a leading project like Toonflow. Also watch NovaNova itself: if it drops back to three or four stars a day with no new events, the "open-source with a paid version" move was a one-time spike, not sustained growth.

[GitHub repo ↗](https://github.com/Swayingleaves/novanova-studio)

**Research notes: competitor table, acquisition evidence, full research report**

Competitors

| Name | Where | Launched | Scale | How it differs from this one |
|---|---|---|---|---|
| Toonflow-app | Open source | 2026-01 | 20.8K stars, +874 in 7 days | Leader; same focus on infinite canvas + Agent + storyboarding; Vue; shipped half a year earlier, also has an APP |
| huobao-drama | Open source | 2026-01 | 17.0k stars | Go+Vue, one sentence to a short drama, fully automated; non-commercial license |
| waoowaoo | Open source | 2026-01 | 14.8k stars | Industrial-grade film pipeline; Elastic license |
| printfilm | Open source | 2026-09 | 7.4k stars | Also ships with billing and an admin console; MIT; Python; has 10× its star count |
| ai-fusion-video (Rongguang) | Open source | 2026-03 | 1.7k stars | Another Java + AgentScope agent-driven short-drama platform; the author has forked it 2026-06 |
| open-ai-canvas (Yingce) | Open source | 2026-07-22 | 1.7K stars, +193 in 7 days | Open-source infinite-canvas film workbench created the same day; TypeScript |
| alibaba/lumenx | Open source | 2026-02 | 1.5k stars | Big-company open-source novel-to-comic-drama pipeline |
| LibTV / Xiaoyunque / Jimeng, etc. | Other | Not verified | Not verified | Commercial SaaS; cannot be self-hosted |

Where users come from

- **Verified** **LINUX DO open-source promo post** : 2026-11-01 "Made a decision that goes against our ancestors" 1.2K views, 101 likes; stars rose about 100 on days two and three; 2026-09-16 v0.1.7 post 714 views

- **Verified** **Event marketing: enterprise edition merged into open source, license changed to MIT** : v1.0.0 release (2026-11-01) notes say fully open-sourced / enterprise code merged / license changed to MIT

- **Verified** **Domestic tech blog seeding + SEO scraper reposts** : CSDN qq_32550077 one post per release, plus Juejin, OSChina, SegmentFault, cnblogs; 11-04~06 reposted by scraper sites like yfxerp and jianying88 (Exa)

- **Verified** **Awesome lists and AI directory listings** : awesome-ai-media, ecosyste.ms, AI directory listing notices

- Inference **Source of the +11-06 / +54 stars** : no matching post found; possibly a WeChat Official Account or group chat, unverified

## Why Swayingleaves/novanova-studio (NovaNova Studio) took off

> Research date: 2026-11-06. Channel: GitHub. Data card: 2026-07-22,672 stars / 105 forks at repo creation (card); live on GitHub today: 689 stars / 106 forks.

### Bottom line

It didn't "break out" on its own. This is a Chinese AI comic-drama / short-video / image-to-video creation platform, written in Java + AgentScope, squeezed into an already crowded open-source niche and sitting in the lower-middle of the pack. Over the past week it gained 174 stars, almost all from a single growth move: **2026-11-01 merged the enterprise edition code into the open-source edition and switched the license from GPL3 to MIT (v1.0.0), and the same day posted a promo thread on LINUX DO titled "I made a decision that goes against our ancestors"** (101 likes, 1.2k views). After that it was reposted on CSDN, SegmentFault, OSChina, Juejin and cnblogs, and scraped by a batch of SEO content farms, which kept the buzz going for a few more days. For the author, open source is a customer acquisition channel: the README header links to the hosted site novanovastudio.cn, and they sell low-priced video/image model API access (Seedance 2.0 at as low as 4 off, MiniMax-H3 at 6 cents per second), with sponsors notoken.pro and ll2.cn, plus enterprise/cluster editions and technical support.

### 1. Data

| Date | stars | fork | Notes |
|---|---|---|---|
| 2026-09-16 | 277 | – | The author self-reported in the first LINUX DO promo post (v0.1.7) |
| 2026-10-10 | 401 | 48 | Our library (github repo_day) |
| 2026-10-20 | 450 | 56 | Organic growth of 3–4 stars in a day |
| 2026-11-01 | 509 | 65 | v1.0.0 "fully open source," changed to MIT; posted LINUX DO thread |
| 2026-11-02 | 558 | 86 | +49 stars, +21 forks |
| 2026-11-03 | 608 | 101 | +50 |
| 2026-11-04/05 | 618 | 104 | Stopped |
| 2026-11-06 | 672 | 105 | +54 (cause not verified, see "Uncertain") |

- Author's self-report (LINUX DO post, 2026-11-01): "500+ stars on GitHub, about 1000 website users"; "a friend made a video with several million views using the platform."

- Forks/stars at about 15%, which fits the positioning of "can be self-hosted and resold with a credits system": admin console, credits, billing and OAuth are all there.

### II. Competitors (open-source AI comic-drama / short-drama platforms; star counts from our GitHub library 2026-11-06)

This track is very crowded. In our database, 2026 repos created this year with "漫剧/短剧" (AI micro-drama / short drama) in their descriptions number 136, and 21 of them gained ≥100 stars in the past 7 days. The awesome-manyu list (updated 2026-11-03) alone catalogs over 20 all-in-one platforms.

| Project | Repo created | stars | up 7 days | How it differs from NovaNova |
|---|---|---|---|---|
| HBAI-Ltd/Toonflow-app | 2026-01 | 20,763 | +874 | The category leader, same focus on infinite canvas + Agent + script storyboarding; Vue; MIT; shipped half a year earlier, and also has an APP |
| chatfire-AI/huobao-drama | 2026-01 | 17,030 | +303 | Go+Vue, one sentence to a short drama, fully automated; non-commercial license |
| waooAI/waoowaoo | 2026-01 | 14,789 | +80 | Once "4K stars in 6.8 days"; Elastic license |
| dramaclaw/dramaclaw | 2026-03 | 7,250 | +677 | General-purpose AIGC video engine; Python |
| yi1108/printfilm | 2026-09 | 7,427 | +56 | Template-driven, with built-in billing and admin console; MIT; the closest positioning, but its star count is 10× this one |
| ArcReel / Jellyfish / moyin-creator | 2026-02~03 | 4.8k–7.1k | – | Pipeline from novel → storyboard → video → CapCut drafts |
| Stonewuu/ai-fusion-video (Rongguang) | 2026-03 | 1,710 | +41 | **Also Java + AgentScope, a full-pipeline agent-based short-drama platform** ; the author forked it 2026-06, and the technical approach clearly draws on this project |
| ddcat-ai/open-ai-canvas (Yingce) | 2026-07-22 | 1,675 | +193 | Created the same day; open-source infinite-canvas film workbench; 2.5× its star count |
| alibaba/lumenx | 2026-02 | 1,453 | +26 | Big-vendor open source |
| tianjiangqiji/nova-image-studio | 2026-06 | 487 | +17 | A self-hosted image/video workbench with a similar name |
| Commercial products | – | – | – | LibTV, Xiaoyunque, Jimeng, etc. (scale not checked one by one) |

**Differentiation** : the author says in the promo post: "When people ask what our advantage is, I thought hard and couldn't name one... so I might as well build a platform with 'nothing' in it." The only differences they can actually name are: ① Java/Spring Boot stack, aimed at Java developers who want enterprise-grade scaling (only Rongguang shares this stack); ② conversational creation with "Agent throughout," likened to Codex/Claude Code; ③ a 3D director's stage on the canvas (v0.1.9: posing mannequins, arranging cameras, motion paths); ④ MIT plus a full user/credits/billing setup, so you can reskin it and run your own paid site. None of these explain the star growth; that came from the "fully open source" event.

### 3. Growth tactics (evidence)

1. **LINUX DO "open-source promotion" thread (confirmed, main channel)**
    - 2026-09-16 "[Open-source promo] Agent-driven image generation, video generation, comic/short drama, infinite canvas project v0.1.7": 714 views, 29 likes. [linux.do/t/topic/2910998](https://linux.do/t/topic/2910998)
    - 2026-11-01 "[Open-source promo] Made a decision that goes against our ancestors...": 1,200 views, 101 likes, 21 replies. Replies are mostly "liked first, will read later," "already starred," "great sir." The nearly 100 stars gained on days two and three line up with this post. [linux.do/t/topic/3104246](https://linux.do/t/topic/3104246)

2. **"Fully open source" event marketing (confirmed)** : the v1.0.0 (2026-11-01) release notes are three lines: "Fully open source / enterprise code merged into open source / license changed to MIT." The author's words: "Not a single PR... In the AI era there's no code moat, so it's better to go fully open."

3. **Domestic tech blog seeding (confirmed)** : CSDN (qq_32550077, many posts, one per version from v0.1.6 on), Juejin (from 2026-08-02), OSChina, SegmentFault (2026-11-03), cnblogs, DevPress. Later reposted in bulk by SEO scraper sites such as yfxerp, mjmagnets, jianying88, nxmr and mwwr (2026-11-04~06, Exa search).

4. **Listing inclusions (confirmed)** : awesome-ai-media, the ecosyste.ms awesome lists, plus "listing updated" notices from AI directory sites.

5. **Monetization hook (confirmed)** : README hero says "Deployment is a pain, finding cheap channels is a pain → click here to try it online," with low-cost Seedance/MiniMax/gpt-image channels. Sponsors notoken.pro and ll2.cn (with invite-code links). "Enterprise edition, cluster edition, tech support" points to a WeChat QR code and a group chat.

6. X / TikTok: x-radar found 0 mentions of "novanova" in the last 60 days; no overseas spread found.

### 4. Why it (modestly) took off

- It rode the 2026 wave of China's "AI micro-drama" boom: price cuts on video models like Seedance 2.0 and MiniMax drew individuals and small teams wanting cheap micro-drama/short-drama production, and the whole category's repos are growing.

- In communities like LINUX DO, where "free riding is welcome," the pitch "we're dropping the paid enterprise edition and switching to MIT" hit a sentiment of supporting open source, and drew a wave of stars.

- The feature list is long (images, video, canvas, micro-drama, novel-to-script, 3D director's desk, credits backend) and the screenshots look good. A high fork rate suggests people are modifying it to run their own operations.

### V. Visibility

It's spreading in small circles, only through LINUX DO and CSDN. No big accounts have written about it, and no revenue has been disclosed. But the category it sits in is already crowded: the leader Toonflow has 20K stars, and new similar repos appear every week.

### VI. Copycats

No copycats found. It's a follower in its own category: Toonflow, Rongguang and others launched half a year earlier.

### VII. What to take away

- Go open-core first; if it doesn't take off, "open-source everything + switch to MIT," framed as an event posted to LINUX DO, which can bring a concentrated wave of stars and forks.

- Use the open-source repo as a lead-capture page: online demo site + selling low-cost model channels + sponsor slots + enterprise-edition WeChat. Money comes back from the API price spread, sponsorships and custom work.

- Write a post on CSDN and Juejin for each release; scraper sites repost them, capturing long-tail SEO.

### 8. Not sure

- I couldn't find where the +54 stars on 2026-11-06 came from (only scraper-site reposts were visible that day; WeChat official accounts or group forwards are possible, unverified). The GitHub API was rate-limited that day, so I couldn't pull stargazer details to check for star-botting.

- "Around 1000 users on the site" and "videos with millions of plays" are self-reported by the author, unverified.

- I don't know how much revenue the API channel price spread brings in.

### Sources

- GitHub repo and releases: [github.com/Swayingleaves/novanova-studio](https://github.com/Swayingleaves/novanova-studio)

- LINUX DO:[linux.do/t/topic/3104246](https://linux.do/t/topic/3104246) ;[linux.do/t/topic/2910998](https://linux.do/t/topic/2910998)

- CSDN: [blog.csdn.net/qq_32550077/article/details/169927799](https://blog.csdn.net/qq_32550077/article/details/169927799); Juejin: [juejin.cn/post/7668692837796003875](https://juejin.cn/post/7668692837796003875); OSChina: [my.oschina.net/u/3483138/blog/19785763](https://my.oschina.net/u/3483138/blog/19785763); SegmentFault: [segmentfault.com/a/1190000048383792](https://segmentfault.com/a/1190000048383792)

- Category list: [github.com/TWO-ICE/awesome-manyu](https://github.com/TWO-ICE/awesome-manyu)

- Star counts: our GitHub data (repo_profile / repo_day, 2026-11-06)

This article was written by AI from public data and the data we collect daily. Numbers are sourced, and inferences are labeled.