Glyph selection in scramble/decrypt effects — curated vs. range-based

[squelch-zero]

I've been building text noise transforms — functions that corrupt, decay, and resolve text at the character level using Unicode glyph substitution. Different problem from TTE (I'm working with pure text transforms, not terminal rendering), but I ran into the same design question and I'm curious how you think about it.

Looking at the decrypt effect source, the scramble pool is built from contiguous Unicode ranges: ASCII printable, block elements, box drawing, and Latin Extended — roughly 540 characters. That produces a specific texture: dense, computery, varied in visual weight.

I went a different direction — a curated set of about 80 glyphs chosen individually for visual properties. Characters like ░ ▒ ▓ ╳ ◌ ◍ ◎ ⊘ ⊗ ∅ ≋ ≈. The goal was a noise texture that feels textural rather than symbolic — closer to static than to code.

The difference is noticeable. Range-based selection reads as encrypted data (which is exactly what decrypt is going for). Curated selection reads more as material degradation — corruption of the medium itself rather than substitution of one symbol set for another.

I'm curious whether you've experimented with the glyph pool composition in TTE's effects. A few specific questions:

1. Was the decrypt character set a deliberate aesthetic choice, or more of a practical "grab a wide range" decision?
2. Have you noticed that different glyph pools change the feel of the same animation timing? I found that heavier glyphs (blocks, fills) make corruption feel slower even at the same frame rate.
3. The errorcorrect and unstable effects seem like they'd benefit from different scramble textures than decrypt. Do they share the same pool, or do they each have their own?

For reference, my library is at squelch-zero/noise — TypeScript, six functions, zero dependencies. The glyph set is in src/glyphs.ts. Not suggesting TTE should use it — just sharing because the design space overlaps and I'm interested in how someone with more experience in this space thinks about character selection.

[squelch-zero]

Following up with findings from my own experiment on this.

I expanded my noise library with glyph pool utilities — range() to expand Unicode blocks, pool() to combine them, plus preset pools (box-drawing, block elements, braille, geometric shapes). Then ran the same text through noise transforms with each pool and compared.

The key variable is weight consistency.

• Curated pools (glyphs chosen for uniform visual weight): noise reads as texture. Signal pops out by contrast. The feeling is degradation — something was whole and is breaking down.
• Range-based pools (contiguous Unicode blocks, varied weight): noise reads as content. Signal competes with noise rather than emerging from it. The feeling is corruption or encryption — something is being hidden, not falling apart.

Within those categories, weight also sets mood:

• Light pools (thin box-drawing, sparse braille) → dust, static, erosion
• Heavy pools (full blocks, dense geometric) → censorship, redaction, obliteration

The _DecryptChars approach in TTE uses four Unicode ranges (~540 chars) which produces that "encrypted data" feeling — consistent with the intended decrypt metaphor. My curated set (~80 glyphs) was chosen for the opposite: visual uniformity so the noise recedes and the signal comes forward.

Neither approach is better. They are tools for different moods. The choice of glyph pool is a narrative decision as much as an aesthetic one.

Source if useful: https://github.com/squelch-zero/noise (the pool utilities are range(), pool(), and the GLYPHS_* constants)