# Update to Papermorph: All I wanted was a few math worksheets for kids. Opus 5.5 made it wild. — retained record

> Retained reddit record from reddit:ClaudeAI.

Canonical: https://highsignal.app/data/records/8f447f2bc72a5650

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:ClaudeAI)  published 2026-10-24 09:38 ingested 2026-10-25 02:36

[original source ↗](https://www.reddit.com/r/ClaudeAI/comments/1xe4e6c/update_to_papermorph_all_i_wanted_was_a_few_math/)

## retained text



```
First of all, thank you all for the support. [About three weeks ago I posted my project here](https://www.reddit.com/r/ClaudeAI/comments/1wwhlgb/opus_55_can_oneshot_a_video_so_i_pushed_it_a/), and it got way more attention and upvotes than I ever imagined. What I didn’t mention in that post: this whole thing started as me trying to make a few math worksheets for elementary school kids. As for the roadmap I shared back then, honestly, I had no idea if I could actually pull any of it off.

Well, three weeks later, I have to say Opus 5.5 blew my mind. Even without any image models, its motion design is seriously impressive.

https://reddit.com/link/1xe4e6c/video/1fgve8r3ydxh1/player

GitHub: [https://github.com/DozenTwelve/Papermorph](https://github.com/DozenTwelve/Papermorph)  
Live bookshelf: [https://papermorph.diamonddoge.org/](https://papermorph.diamonddoge.org/)

As you can see in the video, there are two big updates:

1. **Physics textbook.** I uploaded a few chapters of a physics textbook with much more complex animations (first half of the video). Since the animations are fairly simple, I just had Opus find public, copyright-free assets.
2. **Picture books.** The picture book feature I promised is done. I tried a few different styles and ended up liking this South Park-style cutout puppet look the most (second half of the video). At this point, I felt it was time to bring in an image model. That said, if you insist on having Opus draw everything in SVG, that works too. The backgrounds just end up looking a bit thin.

Since it’s fully open source, feel free to hack on it however you like. Locally I also tried a shadow puppet style, which turned out pretty nice, but I didn’t put it on the site. And if you bring in three.js, you can get some really crazy backgrounds. It’s not on the level of real video models, but it’s still pretty wild. For voiceover, it really depends on your budget. Local options like epubTTS or KokoroTTS are good enough for most people, but if you want the best quality, ElevenLabs v4 is the way to go (it’s not cheap, though).

**One word of warning, though.** I looked at a lot of other people’s projects along the way, and doing motion design with Opus still burns a *lot* of tokens. I wouldn’t recommend doing this through the API, because it gets really expensive. I also tried Sonnet, GPT (Sol), GLM, DeepSeek and others for motion design. They’re definitely much cheaper than Opus, but the results just weren’t as good.

**One current limitation:** if you want to use different models for different steps (handing the simple tasks to Haiku or Sonnet and leaving the actual motion design to Opus), you’ll need to tweak the Skill a bit yourself. In the next version, I’ll fully split out each step so you can choose which model handles what. That should cut token usage a lot, and the quality holds up well. I actually mix Opus and Sonnet myself, by letting Sonnet do the checking frames job.

**What’s next:** I’m going to finish the last row on the bookshelf, the documentary-style videos. You can see I’ve already added [Foucault’s Madness and Civilization](https://papermorph.diamonddoge.org/madness/), but the Skill for it isn’t ready yet. Also, a friend asked for [vertical short-video](https://youtube.com/shorts/QlpQ-E936qw?si=KqCQpD8FmrZq6Gqu) support, so that’s going in as a little easter egg, haha. The plan is to finish all three tracks first, then go back and polish the original textbooks.

Thanks again for all the support!
```



document fetched 2026-10-25 00:22 · canonical https://reddit.com/r/ClaudeAI/comments/1xe4e6c/update_to_papermorph_all_i_wanted_was_a_few_math
