Speed-independent scope trail persistence (XY plots)#240
Speed-independent scope trail persistence (XY plots)#240esaruoho wants to merge 1 commit intopfalstad:devfrom
Conversation
esaruoho
force-pushed
the
v3-mosfet-and-scope-fixes
branch
from
9c4971c to
4c6eaad
Compare
|
It still seems too persistent even at the lowest setting. Maybe you should vary alpha instead of the number of draw calls between fade outs? It would also be nice if it looks the same regardless of the simulation speed setting (as much as practical). The MOSFET looks fine but I would call it "Lambda" in the dialog (since "Beta" is there) and use "la" as the xml dump attr. You don't need to add it to the StringTokenizer constructor. |
…stence MOSFET lambda: - Rename dialog label to "Lambda" (matching "Beta" style) - Change XML dump attribute from "lambda" to "la" - Remove lambda from StringTokenizer constructor (XML-only) Scope trail persistence: - Replace draw-call-counting with alpha-based fade approach - Vary alpha value (trailAlpha) instead of skipping draw calls - Normalize by elapsed wall-clock time for speed independence - Scrollbar controls alpha 0-0.100 (0 = max persistence) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
esaruoho
force-pushed
the
v3-mosfet-and-scope-fixes
branch
from
4c6eaad to
0f6995d
Compare
…nstant model Replaces the draw-call-counter scope fade with a wall-clock-based exponential decay. The user setting now means a literal trail persistence in milliseconds, not an opaque alpha number. Per @pfalstad on PR pfalstad#240: - "vary alpha instead of the number of draw calls between fade outs" - "should look the same regardless of the simulation speed setting" - "It still seems too persistent even at the lowest setting" Implementation: alpha = 1 - exp(-elapsed_ms / trailPersistence) This is the analytic solution for an exponential decay with time constant `trailPersistence`. Same fade rate at 30fps or 144fps, because elapsed wall-clock time appears in the exponent. The user sees and sets the time constant directly (default 200 ms). Range 0..2000 ms. Setting trailPersistence = 0 maps to alpha = 1 (instant erase, no trail at all) so "lowest setting" really is no trail, addressing the "still too persistent" complaint. Persisted per-scope as XML attribute "tp"; omitted when at the default to keep file size small. The MOSFET lambda part of the original PR pfalstad#240 has been superseded by PR pfalstad#313 (MosfetModel class), so this PR is now scope-fade-only. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
esaruoho
force-pushed
the
v3-mosfet-and-scope-fixes
branch
from
0f6995d to
52c36f7
Compare
esaruoho
changed the title
|
@pfalstad — major rework based on your feedback. Three changes:
Force-pushed onto the rebased branch. |
…nstant model Replaces the draw-call-counter scope fade with a wall-clock-based exponential decay. The user setting now means a literal trail persistence in milliseconds, not an opaque alpha number. Per @pfalstad on PR pfalstad#240: - "vary alpha instead of the number of draw calls between fade outs" - "should look the same regardless of the simulation speed setting" - "It still seems too persistent even at the lowest setting" Implementation: alpha = 1 - exp(-elapsed_ms / trailPersistence) This is the analytic solution for an exponential decay with time constant `trailPersistence`. Same fade rate at 30fps or 144fps, because elapsed wall-clock time appears in the exponent. The user sees and sets the time constant directly (default 200 ms). Range 0..2000 ms. Setting trailPersistence = 0 maps to alpha = 1 (instant erase, no trail at all) so "lowest setting" really is no trail, addressing the "still too persistent" complaint. Persisted per-scope as XML attribute "tp"; omitted when at the default to keep file size small. The MOSFET lambda part of the original PR pfalstad#240 has been superseded by PR pfalstad#313 (MosfetModel class), so this PR is now scope-fade-only. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
esaruoho
force-pushed
the
v3-mosfet-and-scope-fixes
branch
from
52c36f7 to
8382a64
Compare
2 participants
Summary
Addresses all of @pfalstad's feedback on this PR. MOSFET lambda part removed — superseded by #313 (MosfetModel class). This PR is now scope-fade-only.
Scope trail persistence — three things you asked for
"Vary alpha instead of the number of draw calls" — replaced the
alphaCounter > 2draw-call counter with a real per-frame fade alpha."Should look the same regardless of simulation speed setting" — the alpha is computed from elapsed wall-clock time, not frame count, using the analytic exponential-decay formula:
Same trail at 30fps or 144fps because elapsed time appears in the exponent, not the multiplier.
"Still too persistent even at the lowest setting" — the user setting is now a literal time constant in milliseconds (default 200 ms, range 0..2000 ms). Setting it to 0 sets alpha = 1 → instant erase → genuinely no trail at all. So "lowest persistence" actually means no trail.
UX
The Scope Properties dialog now shows a single slider labelled "Trail Persistence (ms)" with a numeric readout next to it (
200 ms,500 ms, …, ornoneat zero). No more opaque alpha number.Persistence
Saved per-scope as XML attr
tp. Omitted when at the default (200) to keep file size small.Test plan
0→ no trail at all (each frame fully overwrites).2000 ms→ long trail, slow fade.tpround-trips for non-default values; default circuits unchanged.🤖 Generated with Claude Code