itohio · foxis · Aug 17, 2026 · Aug 17, 2026
diff --git a/blog/content/colorimetry/laser-power-meter/index.md b/blog/content/colorimetry/laser-power-meter/index.md
@@ -137,8 +137,6 @@ $$
 P_{cal} = \frac{V^2}{R}
 $$
 
-<!-- TODO: Add board.jpg - photo of RP2040 board with wires and trim potentiometers -->
-
 ### The Circuit
 
 I use a [Wheatstone bridge](https://en.wikipedia.org/wiki/Wheatstone_bridge) configuration with the two thermistors as variable resistors. The bridge converts the resistance difference (temperature difference) into a voltage difference.
@@ -259,7 +257,6 @@ Where $s$ is the measured slope. The quadratic term accounts for nonlinearities
 
 I connected the RP2040 to my computer via USB serial. A Go application displays the differential signal in real-time and lets me trigger calibration measurements.
 
-<!-- TODO: Add software-ui.jpg - screenshot of Go application interface -->
 ![Software interface showing the measurement signal](software-ui.jpg)
 
 The software:

diff --git a/blog/content/fpv/betaflight-vtx-indexing-trap/index.lt.md b/blog/content/fpv/betaflight-vtx-indexing-trap/index.lt.md
@@ -187,7 +187,7 @@ Perkrauk, ir ratukas daro tai, ką siūlo fizinis pasukimas: apačia = žema, vi
 
 Sutvarkyk indeksavimą, ir 400 mW *turėtų* pasirodyti ratuko viršuje. Jei vis tiek nepasirodo, gali būti, kad juos laiko pats VTX. Daugelis SmartAudio VTX — tarp jų ir TH3 — parduodami su **apribotu (regionui užrakintu) režimu**, ribojančiu išvestį dėl reglamentų. Tokioje būsenoje aukštos galios įrašai tavo `vtxtable` yra apibrėžti teisingai, išrenkami teisingai — ir tiesiog atmetami aparatinės įrangos.
 
-Apribotas režimas gyvena VTX'e, ne Betaflight'e, tad jokie CLI pakeitimai jo nepašalins. Jį išjungi ant paties įrenginio — paprastai palaikant VTX mygtuką (arba atitinkama SmartAudio „unlock" komanda), kad išeitum iš regionui užrakinto profilio. [TODO: prieš publikaciją patikrinti tikslią TH3 V02 atrakinimo seką.]
+Apribotas režimas gyvena VTX'e, ne Betaflight'e, tad jokie CLI pakeitimai jo nepašalins. Jį išjungi ant paties įrenginio — paprastai palaikant VTX mygtuką (arba atitinkama SmartAudio „unlock" komanda), kad išeitum iš regionui užrakinto profilio. Tikslios atblokavimo sekos savo TH3 V02 nepatikrinau, tad mygtuko palaikymą vertink kaip bendrą principą, o ne kaip receptą, ir konkrečią seką tikrink savo VTX dokumentacijoje.
 
 Požymis diagnostinis: jei žema ir vidutinė galia išrenkamos teisingai, bet vien viršutinis lygis lieka miręs po indekso pataisymo, įtark apribojimą, ne konfigūraciją.
 

diff --git a/blog/content/fpv/betaflight-vtx-indexing-trap/index.md b/blog/content/fpv/betaflight-vtx-indexing-trap/index.md
@@ -187,7 +187,7 @@ Reload, and the wheel does what the physical rotation implies: bottom = low, mid
 
 Fix the indexing and 400 mW *should* appear at the top of the wheel. If it still doesn't, the VTX itself may be holding it back. Many SmartAudio VTX — the TH3 included — ship with a **restricted (region-locked) mode** that caps output for regulatory compliance. In that state the high-power entries in your `vtxtable` are defined correctly, selected correctly, and simply refused by the hardware.
 
-Restricted mode lives on the VTX, not in Betaflight, so no amount of CLI editing clears it. You disable it on the unit itself — typically a button hold on the VTX (or the matching SmartAudio "unlock") to drop out of the region-locked profile. [TODO: confirm the exact TH3 V02 unlock sequence before publish.]
+Restricted mode lives on the VTX, not in Betaflight, so no amount of CLI editing clears it. You disable it on the unit itself — typically a button hold on the VTX (or the matching SmartAudio "unlock") to drop out of the region-locked profile. I have not confirmed the exact unlock sequence on my own TH3 V02, so treat the button-hold as the general pattern rather than a recipe, and check your VTX documentation for the specific one.
 
 The tell is diagnostic: if low and mid power select correctly but the top level alone stays dead after the index fix, suspect restriction, not configuration.
 

diff --git a/blog/content/fpv/drone-detector-nn/index.lt.md b/blog/content/fpv/drone-detector-nn/index.lt.md
@@ -186,10 +186,6 @@ pos_weight = (n_neg / n_pos).clamp(min=NUM_CLASSES, max=30.0)
 
 ## Rezultatai ir kiekvienos klasės analizė
 
-*[TODO: Mokymo kreivė — nuostoliai ir makro F1 per ciklus]*
-
-*[TODO: 10×10 painiavos matrica]*
-
 ### Kiekvienos klasės F1
 
 | Klasė | F1 | Pastaba |

diff --git a/blog/content/fpv/drone-detector-nn/index.md b/blog/content/fpv/drone-detector-nn/index.md
@@ -204,12 +204,6 @@ The minimum clamp at `NUM_CLASSES` (10) prevents near-zero weights for majority
 
 ## Results and Per-Class Analysis
 
-<!-- IMAGE: Chart.js training curve — loss and macro F1 across staggered P1↔P2 cycles, phase transitions annotated -->
-*[TODO: Training curve — loss and macro F1 across cycles, phase transitions annotated]*
-
-<!-- IMAGE: 10×10 confusion matrix heatmap -->
-*[TODO: 10×10 confusion matrix]*
-
 ### Per-Class F1
 
 | Class | F1 | Note |

diff --git a/blog/content/fpv/meteor75-pro-ii-fighting-resonances/index.lt.md b/blog/content/fpv/meteor75-pro-ii-fighting-resonances/index.lt.md
@@ -19,28 +19,6 @@ series:
 thumbnail: "meteor75-pro-vs-pro-ii.jpg"
 ---
 
-<!--
-VERTIMO PASTABA / DRAFT NOTE — pašalinti prieš publikavimą.
-Tai vertimo juodraštis. Betaflight parametrų pavadinimai, FPV terminai ir
-matavimų žymėjimai palikti angliškai — jie tokie ir naudojami (sąmoningas
-sprendimas, ne apsileidimas).
-
-ANDRIUI PERŽIŪRĖTI — penki terminai, kurių negaliu patvirtinti:
-  1. "vibracijos gaubtinė"          <- vibration envelope
-  2. "struktūrinė moda"             <- structural mode
-  3. "atsako dozė"                  <- dose-response
-  4. "prisukta masė"                <- sprung mass
-  5. "struktūrai fiksuota ypatybė"  <- structure-fixed feature
-Jei kuris nors iš jų lietuviškame FPV/virpesių žargone skamba ne taip —
-pakeisk, ir aš atnaujinsiu visus pasikartojimus abiejose versijose.
-
-series ir thumbnail dabar sutampa su anglų versija: [FPV Builds] ir
-meteor75-pro-vs-pro-ii.jpg (anksčiau čia buvo [Snake] ir jokio thumbnail).
-
-Skaičiai, kurie tekste pasakyti dviem skirtingais būdais, surašyti anglų
-versijos DRAFT NOTES bloke (index.md) — nesprendžiau nė vieno, palikta tau.
--->
-
 Craft name **Snake**. Pradžioje tai buvo Meteor75 Pro, dabar — Meteor75 Pro II: rėmas ir
 gaubtas iš AliExpress, viskas, kas kainuoja, perkelta be pakeitimų. Tas pats
 **Matrix 1S 3-in-1 FC**. Tas pats **narrow-FOV DJI O4** air unit. Naujas kiautas, seni vidūriai,

diff --git a/blog/content/fpv/meteor75-pro-ii-fighting-resonances/index.md b/blog/content/fpv/meteor75-pro-ii-fighting-resonances/index.md
@@ -20,44 +20,6 @@ series:
 thumbnail: "meteor75-pro-vs-pro-ii.jpg"
 ---
 
-<!--
-DRAFT NOTES FOR ANDRIUS — delete this comment block before publishing.
-
-1. TITLE / DATE — still yours to confirm. series is [FPV Builds] to match the
-   other FPV posts, thumbnail is meteor75-pro-vs-pro-ii.jpg. Both language
-   files match on series and thumbnail.
-2. LITHUANIAN TERMINOLOGY — needs your review. Five coinages I could not
-   validate, all in index.lt.md:
-     - "vibracijos gaubtinė"          (vibration envelope)
-     - "struktūrinė moda"             (structural mode)
-     - "atsako dozė"                  (dose-response)
-     - "prisukta masė"                (sprung mass)
-     - "struktūrai fiksuota ypatybė"  (structure-fixed feature)
-   Betaflight parameter names and metric labels are deliberately left in
-   English throughout — intentional, not an oversight.
-3. NUMBERS STATED TWO WAYS — RESOLVED. Rather than silently picking one value,
-   the opening of "Method notes worth keeping" now names the two analysis
-   windows and explains every apparent disagreement:
-     - 34.8/24.6 vs 34.5/24.5, +66% vs +65%, 0.789% vs 0.812%, 0.04% vs 0.076%
-       -> different impact-exclusion pad or a different log, not a correction
-     - foam post-filter roll 0.50 vs 0.67 deg/s -> two different metrics
-       (Welch above 60 Hz vs an 80-780 Hz bandpass RMS)
-     - the two five-mount sets (37.7/26.2/33.0/30.1/41.0 and
-       38.3/26.0/31.0/26.6/42.5) -> outdoor RPM-binned vs steady-flight
-       throttle-banded windows
-   The intro now says "seven of them, itemised later" and matches the seven H3s
-   under "Everything I got wrong". The stray "wrong for the fifth time" is gone.
-4. chart8 in snake-chart-data.json (outside this folder) had a stale
-   "Peak 48.8 -> 28.0 deg/s (-43%)" annotation. Already corrected there; the
-   retired figure survives only in a _retracted metadata field as an audit
-   trail, so regenerating from the JSON will not reintroduce it.
-5. EXIF — GPS IFD removed from all five photos, remaining tags kept per the
-   GPS-only rule. Verified before committing.
-6. TONE — all four photographs were rebuilt from the originals with anchored
-   monotone L-channel curves, not brightness multiplication. Every one passes
-   the verification gate. The screenshot is deliberately uncurved.
--->
-
 Craft name **Snake**. It started life as a Meteor75 Pro, and it is now a Meteor75 Pro II —
 frame and canopy ordered off AliExpress, everything expensive carried straight over. Same
 **Matrix 1S 3-in-1 FC**. Same **narrow-FOV DJI O4** air unit. New shell, old guts, and by the
@@ -2744,7 +2706,6 @@ impact-exclusion pad or the log being different, not a correction. Post-filter r
 as both 0.50 and 0.67 °/s for the foam because one is a Welch integration above 60 Hz and the
 other an 80–780 Hz bandpass RMS; they are two metrics, not two answers.
 
-
 Practices that repeatedly changed the conclusion — not general advice, things that actually
 flipped an answer in this specific week: