Differences

This shows you the differences between two versions of the page.

--- newsletters:2026-01 [2026/02/10 02:48] – [OCR experiments] osnr
+++ newsletters:2026-01 [2026/02/11 04:19] (current) – [January 2026 newsletter] osnr
@@ Line 1: / Line 1: @@
 ====== January 2026 newsletter ======
-Our next [[https://luma.com/ptzbza54|Folk open house is in the evening on Thursday, February 26th, in East Williamsburg, Brooklyn.]]
+{{htmlmetatags>metatag-og:title=(Folk Computer January 2026 newsletter)
+metatag-media-og:image=(newsletters:619449636_18568814476029757_6219918909947160428_n.jpg)
+metatag-og:description=(Receipt printer & camera fixes; dual-tag quads experiment; China trip; shape library work; TrOCR handwriting experiment; new intro booklet)
+}}
+Our next Folk open house will be on the evening on Thursday, February 26th, in East Williamsburg, Brooklyn. RSVP here:
+[[{}https://luma.com/ptzbza54|RSVP for the Folk open house on Feb 26]]
+Last month, we officially [[newsletters/2025-12#nonprofit-501-c-3|formed a nonprofit for Folk.]] We’re trying to get more serious about fundraising — we'd love for you to sponsor us so we can keep doing this work:
+[[{}https://github.com/sponsors/FolkComputer|**Sponsor Folk on GitHub Sponsors**]]
 ===== What we've been up to =====
-  * [[https://danielpipk.in|Daniel Pipkin]] [[https://github.com/FolkComputer/folk/commit/8551cff340862ded01e57ec6e67d790294f5e823|fixed our ESC/POS receipt printer support for folk2]]: now it uses fn and Subscribe:/Notify: instead of globals
+  * [[https://danielpipk.in|Daniel Pipkin]] and [[https://github.com/smj-edison|Mason Jones]] paired in person in Utah on dual-tag-derived quads that work in 3D:
+    * {{newsletters:pxl_20260103_231220991.mp4?200px}} {{newsletters:pxl_20260103_231501474.mp4?200px}}
+    * Omar: sort of a modernized version of the dual-tags that [[https://x.com/naveenma|Naveen Michaud-Agrawal]] [[newsletters/2023-10#what-we-ve-been-up-to|had a couple of years ago in 2D]]
+  * Daniel [[https://github.com/FolkComputer/folk/commit/8551cff340862ded01e57ec6e67d790294f5e823|fixed our ESC/POS receipt printer support for folk2]]: now it uses fn and Subscribe:/Notify: instead of globals
   * [[https://github.com/sorig|Esben Sørig]] contributed some camera/setup fixes:
     * [[https://github.com/FolkComputer/folk/pull/241|Extend camera load timeout to 2 seconds and add retry mechanism]], which seems to fix the Logitech C270 camera
@@ Line 33: / Line 47: @@
 {{newsletters:img_1463.jpeg?250px}}
-It runs really fast on folk-hex when CUDA is on (maybe 10x faster, multiple frames per second). That liveness is fun to play with. Excited to show more demos around this next month.
+[[https://github.com/FolkComputer/folk/blob/7a62fc2de4599b72ad7f6a612c69655e952a43b8/builtin-programs/trocr.folk|Here's a draft trocr.folk]] that exports a callable function (more verbose / AI-y than it probably should be, but it's fully self-contained, which is great).
+Easy invocation on a static on-disk image (to use live, put it in an infinite loop and query for camera slice):
+<code>
+When the image library is / imageLib/ &\
+     the TrOCR function is /TrOCR/ {
+  fn TrOCR
+  set im [$imageLib loadJpeg "a01-122-02.jpg"]
+  puts [TrOCR $im]
+}
+</code>
+It runs really fast on folk-hex when CUDA is on (maybe 10x faster, multiple frames per second). That liveness is fun to play with.
+Excited to show more demos around this next month.
+A lot of this is built on [[https://docs.astral.sh/uv/guides/tools/|uvx]], which I just learned about this month. uvx gives us the ability to run Python with arbitrary dependencies in 'immediate mode', which aligns a lot better with Folk (wanting self-contained declarative Folk programs, wanting to be able to change them to do anything at any time, etc) than a traditional project/env-oriented Python workflow would. You just run ''uvx'' with all dependencies as arguments and then can eval any Python you want, and it automatically caches the dependencies so it doesn't have to reinstall from scratch every time.
 Some issues to resolve:
-  * TrOCR really can't cope with any padding around the handwriting; it needs a tight box around the text, so we probably need a separate text detection model to maximize usability (otherwise you have to write your word to exactly fill the camera slice, which is sort of annoying, especially if the slice is reasonably large)
+  * TrOCR really can't cope with any padding around the handwriting; it needs a tight box around the text, so we need a separate text detection model to maximize usability (otherwise you have to write your word to exactly fill the camera slice, which is sort of annoying, especially if the slice is reasonably large)
   * I think we'll need to push harder on finding and keeping a really good calibration if we want to be able to project onto/alongside handwriting
   * I also want to figure out a camera slice change detection system so we don't have to run the OCR model continuously
+=== Text detection with CRAFT ===
+I started [[https://github.com/osnr/craft-text-detector|experimenting with]] the [[https://arxiv.org/abs/1904.01941|CRAFT]] [[https://github.com/fcakyon/craft-text-detector|text detector]], which seems to be what people [[https://medium.com/@harshanck/trocr-craft-multi-line-handwritten-text-recognition-htr-db1235bc6e0f|generally]] pair with TrOCR.
+{{.:pasted:20260210-031918.png?300px}}
+I tuned it to use the GPU and to be more aggressive in linking regions, so we capture the whole "wow" in one box here:
+{{.:pasted:20260210-031834.png?300px}}
+{{.:pasted:20260210-031820.png?400px}}
+[[https://github.com/FolkComputer/folk/blob/d8c9d4d1402455b9a745a994894917f50d5b1c49/builtin-programs/trocr.folk|Here's the draft code that does both CRAFT and TrOCR]], which I haven't tested on a table (I'm now working on a more general Python FFI so that the OCR pipeline isn't one big monolith, and so we can try other models easily).
 ==== Stuff from China ====
-Omar:
+Omar: I was in China for a few weeks (for some workshops and conferences in Shanghai and Shenzhen). I prioritized getting stuff I couldn't find in the US (often these are not even on Aliexpress, only on Taobao or in person). Got a few interesting Folk-related gadgets:
-{{newsletters:img_1439.jpeg?400px}}
+  * [[https://item.taobao.com/item.htm?id=961137090756|"4K" "X10 MAX" projector]] (actually 1080p)
+    * {{newsletters:img_1439.jpeg?0x200px}} {{newsletters:img_1432.jpeg?0x200px}} {{.:pasted:20260210-011824.png?0x200px}}
+      * I like the automatic focus mechanism. The projector is pretty bright
+    * Got this for 888RMB ($126) on Taobao which is good for a good 1080p projector, but it's not 4K
+    * It's really hard to find a genuine 4K projector. A lot of projectors advertise 4K, but it's just 4K input and they downscale on output to 1080p (pretty useless for us, since we want the actual resolution for text editing). Salespeople and product pages often don't know this distinction and will mislead you
+  * [[https://www.gtc-technology.com/products/hachi-k1-觸控智能投影機|Hachi K1 tabletop projector/touchscreen]]
+    * I got this for 1400RMB ($200) at a random store in HQB -- way cheaper than I've seen anything comparable online (I mostly see the M1 online; not sure what the difference is), or I think its original retail price
+      * I [[https://www.youtube.com/results?search_query=hachi+projector|heard about the Hachi devices]] years ago, but $600-1000 wasn't worth it to me to import, but $200 is easily worth it
+      * {{.:pasted:20260210-190828.jpeg?250px}} {{newsletters:img_0227.mp4?150px}}
+    * I think the K1 is the last model they put out and is China-only? It seems mostly similar to the older models I've seen online
+    * It's basically an Android tablet. Trying to root it and see what the sensor inputs are (cameras, laser line), maybe run Folk on it. It's a really nice form factor -- could also just copy it and make our own device
+  * Deli barcode scanners: the [[https://item.taobao.com/item.htm?id=631744239452|AA601]] tabletop & especially the [[https://www.delioa.com/products/es228wb-1/|ES228WB]] handheld
+    * I just liked the industrial design of these:
+      * {{newsletters:aa601.png?0x250px}} {{.:pasted:20260210-190411.png?0x250px}}
+    * The handheld ES228WB especially is very inspiring for a Folk gadget form factor and looks so much more modern than most barcode scanners I've seen
+    * (this makes sense, since barcode scanner is the equivalent of a cash register there -- they're very cheap, $10-20, very common, very consumer-facing)
+  * [[https://item.taobao.com/item.htm?id=977623174365|Guangzhou X-Dream X1200W portable A4/letter printer]]
+    * I got this on Taobao for 800RMB ($115) -- way cheaper than the Canon or Epson portable printers we've used in the past
+    * This is a totally viable printer for Folk if only we could get good drivers for it (I can only drive it from their iOS app right now, even their PC Linux drivers don't work for me)
-{{newsletters:img_1432.jpeg?400px}}
+In general, my hope was that we could find competitively-priced devices in these categories (that maybe are only available in China), but was only kind of successful. Let us know if you know of anything:
-{{.:pasted:20260210-011824.png?400px}}
+  * 4K projector ($800 or cheaper, and/or less than 5 pounds) like the AAXA 4K1
+  * Pico projector ($400 or cheaper) like the Ultimems AnyBeam
+  * Portable printer ($300 or cheaper) like the Canon TR150 or Epson EC-C110
-TODO: table one
+Briefly showed off the [[newsletters/2025-12#new-gadget1|new pink gadget1]] I made last month:
+{{newsletters:619449636_18568814476029757_6219918909947160428_n.jpg?250px}}
+Also started looking into PCB design for a side project. As a result, I'm thinking very seriously about making a smaller gadget that fits in a barcode scanner-like chassis (something like the Deli one above). Would need a custom PCB and heat dissipation strategy, but we could just use the bare Ultimems projector board (no heat sink), and we could use just one 18650 battery. That would go a long way in reducing weight and volume.
 ==== Outreach ====
@@ Line 60: / Line 132: @@
       * {{newsletters:img_3827-2-medium.jpeg?400px}}
-===== What we'll be up to in January =====
+===== What we'll be up to in February =====
   * Our next [[https://luma.com/ptzbza54|Folk open house is in the evening on Thursday, February 26th, in East Williamsburg, Brooklyn.]]
   * Andrés: Finishing shape library
   * Andrés: Working on restoring and making demos with the dot detector capability
+  * Omar: Make and post some CRAFT+TrOCR demos; publish Python FFI; maybe start playing with segmentation
+  * Omar: Fix some bugs with folk2 (texture blinking, leaks)
+  * Omar: Maybe try and start on stereo calibration and/or calibration refinement
+  * Omar: New gadget design?
 ===== Links we've enjoyed =====