このプラグインは 3 つのエフェクトフィルタを提供します。
| フィルタ | 日本語名 | 内部 ID |
|---|---|---|
| LED Strip Wall | LEDテープウォール | led_strip_wall |
| Transparent LED Screen | 透明LEDスクリーン | led_transparent_screen |
| HUB75 LED Panel | HUB75 LEDパネル | led_hub75_panel |
「設定キー」は obs-websocket(SetSourceFilterSettings)やシーンコレクション JSON で
使う名前です。「対象」は T = LEDテープウォール、S = 透明LEDスクリーン、H = HUB75 LEDパネル。
特記のない「T / S」の共通項目は H でも同様に使えます(電源・にじみ・明るさ・ちらつき・
恒久ドット抜け/点灯・通信途切れ・3D飛び出し・確認用拡大・デバッグ等)。
対象 T / S ・ 型 bool ・ 初期値 ON
OFF にすると全 LED が消灯します(約 0.13 秒のフェード付き)。消灯中もテープ・電極・ LED パッケージ/ガラス基板は残るので、「電源が落ちた LED ウォール」の見え方になります。 ホットキー・ドック・obs-websocket からも操作できます。
対象 T / S ・ 型 int ・ 範囲 8〜1024 ・ 初期値 T: 160 × 90、S: 96 × 54
パネルの LED 密度(=実効解像度)です。元画像はこの格子でサンプリングされ、 シリアルデータもこの個数分だけ生成されます。
- 密度を上げるほど元映像に近づき、下げるほど「粗いLEDウォール」になります
- 1 LED あたり約 4 出力画素を下回ると構造を描けないため、自動的に面積平均の 簡易表示へフェードします(モアレ防止)。目安として 1080p 出力なら 480×270 程度、 4K 出力なら 960×540 程度が「構造が見える上限」です
- LED 数の上限は 1024×1024(約 105 万)。ストリームテクスチャは最大 1024 texel 幅に 折り畳んで格納します
対象 T / S ・ 型 int ・ 選択肢 0=ソースに合わせる / 1=1920×1080 / 2=3840×2160 ・ 初期値 0
フィルタが出力する解像度です。固定モードはソース解像度に関係なくその解像度で テープ構造を描き直すので、1080p ソースでも「4K」を選べば 4K キャンバス上で構造が くっきり出ます(拡大してボケるのではなく、パネル自体を 4K で生成します)。逆に 4K ソースを 1080p キャンバスで使うなら「1080p」で表示段の負荷を 1/4 にできます。
- ソースと縦横比が異なる場合は引き伸ばされます
- 色のサンプリング(シリアルデータ)は LED 密度依存なので、どのモードでも エンコード内容・劣化の挙動は変わりません
対象 T / S ・ 型 int ・ 初期値 T: 0、S: 2
LED をシリアルチェーンに並べる順序です。断線やビット化けの伝播方向がこれで決まります。
| 値 | 対象 | 内容 |
|---|---|---|
| 0 | T | 横折り返し(蛇行): 1234 / 8765 / 9ABC — 行末で折り返して次の行へ |
| 1 | T | 縦折り返し(蛇行): 167C / 258B / 349A — 列末で折り返して次の列へ |
| 2 | S | くし形・横: 各行が独立チャンネルで、全行 左→右 |
| 3 | S | くし形・縦: 各列が独立チャンネルで、全列 上→下 |
蛇行(0/1)は 1 本の長いテープを折り返して貼った構造なので、断線するとそれ以降の パネル全体が死にます。くし形(2/3)は駆動 IC が縁にあり各ラインへ並列給電する 実物の ITO 構造なので、故障はそのライン内に閉じます。
対象 T / S ・ 型 float ・ 範囲 0.0〜2.0 ・ 初期値 0.8
発光の広がり(グロー)です。自分の LED と上下左右 4 個の隣接 LED のガウス分布を 加算合成しています。上げるほど「にじんで明るい LED ウォール」、0 にすると くっきりした点光源になります。透過型パネルでは隙間の上にも加算で乗るため、 上げると「光のカーテン」感が出ます。
対象 T / S ・ 型 float ・ 範囲 0.1〜2.0 ・ 初期値 1.0
LED の輝度倍率です。1.0 で元映像相当。上げると白飛び側にレンズのハイライトが乗り、 下げると「輝度を絞った LED パネル」になります。
対象 T ・ 型 int ・ 選択肢 0=黒テープ / 1=白テープ ・ 初期値 0
テープ基材(FPC)の色です。黒は暗所でコントラストが高く、白は照明下で 「白いテープが見える」実物らしさが出ます。
対象 T ・ 型 int ・ 選択肢 1=白パッケージ / 0=黒パッケージ ・ 初期値 1
LED の樹脂ボディの色です。白は一般的な WS2812B、黒は黒枠タイプ(WS2812B-B 等)で、 消灯時のレンズもやや暗くなり、点灯時はレンズの発光だけが浮かび上がります。
対象 T ・ 型 float ・ 範囲 5〜100 ・ 初期値 62
LED ピッチに対するテープの太さです。100% 未満にするとテープの隙間がアルファ透過に なり、シーン上でこのソースの下にあるソースが隙間から見えます(透過型パネル)。 100% にするとテープ同士が密着し、完全不透明になります(継ぎ目の半透明も出ません)。
対象 T ・ 型 float ・ 範囲 10〜100 ・ 初期値 55
テープ上の LED 実装間隔に対するパッケージの大きさです。小さくすると LED 同士の 間にテープ面が広く見え、疎な実装のテープになります。
対象 T ・ 型 float ・ 範囲 0〜100 ・ 初期値 50
ハンダパッド(電極)の茶色の濃さ。0% で明るい銅色、100% で暗い焦げ茶です。 電極は実物と同じく 3 接点(+5V / DATA / GND)に分かれており、横蛇行では「三」の字、 縦配線では「川」の字に見えます(分割方向は常にテープと直交)。
対象 S ・ 型 float ・ 範囲 50〜98 ・ 初期値 85
ガラス/アクリル基板の透過率です。実物の透明 LED スクリーンは 70〜95% 程度。 ここで決まる不透明度(1 − 透過率)が全面に乗り、その分だけ背後のソースが見えます。 下げるほど基板のティントが濃くなります。
対象 S ・ 型 float ・ 範囲 5〜60 ・ 初期値 18
LED ドットの直径(ピッチ比)です。実物は極小のドットがガラスに浮いて見えるので、 既定は小さめ。消灯時はうっすら灰色の粒として残ります。
対象 H ・ 型 int(値 = 横×1000+縦)・ 選択肢 64×64 / 64×32 / 32×32 / 32×16 ・ 初期値 64×64
LED モジュール(パネル 1 枚)の LED 数です。実物の P2.5(64×64)や P4/P5(64×32)に 相当します。
対象 H ・ 型 int ・ 範囲 1〜16 ・ 初期値 6 × 4
モジュールを何枚並べてウォールを作るか。総 LED 数 = モジュール解像度 × 枚数
(片軸 1024 LED を超える場合は自動的に枚数が制限されます)。LED 数の直接指定
(led_cols/led_rows)は HUB75 では使いません。
対象 H ・ 型 float ・ 範囲 0〜100% ・ 初期値 60
モジュール同士の継ぎ目の見え方。大型ビジョンで格子状にうっすら見えるあの線です。 0% で消えます。
対象 H ・ 型 float ・ 範囲 20〜95% ・ 初期値 70
角丸四角の SMD パッケージの大きさ(ピッチ比)。実物の高密度パネルは 60〜80% 程度。
対象 H ・ 型 int ・ 選択肢 8(1/8)/ 16(1/16)/ 32(1/32)・ 初期値 16
走査駆動の分割数です。走査縞の本数と、アドレス線故障の見え方の前提になります。
対象 H ・ 型 float ・ 範囲 0〜100% / 0.1〜10 Hz ・ 初期値 0(OFF)/ 2
カメラで撮ったときに見える、流れる暗い横縞(ローリングシャッターとリフレッシュの 干渉)です。肉眼では見えない現象なので既定は OFF。「カメラで撮った LED ビジョン」の 演出をしたいときに上げてください。
対象 H ・ 型 float + int ・ 範囲 0〜100%(0.1 刻み)/ 0=消灯・1=フリーズ・2=ビット化け ・ 初期値 0 / 消灯
故障しているモジュール(パネル 1 枚単位)の割合です。「1 枚だけ黒い/固まっている/ 砂嵐のように化けている」大型ビジョンの定番故障。位置はハッシュ固定ランダムで、 時間で変化しません。フリーズは故障発生時点の映像を保持し続けます(ラッチ機構による 物理的に正しい挙動)。
対象 H ・ 型 float ・ 範囲 0〜100% ・ 初期値 0 / 50
経年劣化で赤茶けたモジュールの割合と、その色ずれの強さです。青・緑 LED(InGaN)は 赤 LED(AlGaInP)より輝度劣化が速いため、古いモジュールほど B > G > R の順で暗くなり赤寄りの暖色にずれます(実装: B×(1−0.50s)、G×(1−0.35s)、 R×(1−0.10s)。s はモジュールごとにランダムなばらつき付き)。ロット違い・交換モジュール 混在の「まだらなパネル」が再現できます。俗に「バーンイン」と呼ばれることもありますが、 正式には**経年劣化による色度シフト(色ムラ)**です(バーンイン=画像残留は別現象)。
対象 H ・ 型 bool + int(1〜256)+ int ・ 初期値 OFF / 1 / フリーズ
モジュール間をつなぐ HUB75 ケーブルの断線です。連結順(モジュールを横蛇行: 1段目 左→右、2段目 右→左、…)で数えた指定番号のモジュール以降すべてが フリーズ(最後の映像を保持)または消灯します。
対象 H ・ 型 float ・ 範囲 0〜100%(0.1 刻み)・ 初期値 0
死んだドライバ IC の割合です。1 個死ぬと、その IC が受け持つ 16 列 × モジュール半分のブロックが消灯します。
対象 H ・ 型 bool ・ 初期値 OFF
行選択(アドレス)線の故障です。走査行が半分選択されなくなり、モジュール内で 1 行おきに暗い縞が出ます。
3D Effect(exeldro 製)などで ソースを 3 次元的に傾けたとき、LED だけが基材から浮いて見える視差表現です。 傾きゼロのときは従来と完全に同じ絵になります。
対象 T / S ・ 型 float ・ 範囲 -100〜100 ・ 初期値 5
LED が基材からどれだけ浮いているか。視差量 = 高さ × tan(傾き) です。
- 0% で機能無効
- 飛び出す方向が逆に見える場合はマイナス値にすると反転します
- 上げるほど強調されますが、視差は 0.6 ピッチでクランプされます
対象 T / S ・ 型 int ・ 選択肢 0=自動 / 1=手動 ・ 初期値 0
自動では、同じソースに付いている 3D Effect フィルタ(内部 ID 3d_effect_filter)の
回転角 rot_x / rot_y / rot_z を毎フレーム読み取って自動追従します(0.25 秒ごとに
フィルタチェーンを再走査)。3D Effect 側のスライダーを動かすだけで LED の浮きが
ついてきます。見つからない場合は視差なしです。
手動では下の傾きスライダーで直接指定します。正射影近似のため、3D Effect の FOV (遠近感)が強いときは手動で微調整してください。
対象 T / S ・ 型 float ・ 範囲 -85〜85 度 ・ 初期値 0
手動モードでの傾き。X は上下(ピッチ)、Y は左右(ヨー)方向です。
対象 T / S ・ 型 bool(グループのチェックボックス)・ 初期値 OFF
グループ全体の有効/無効です。OFF のときは完全に無劣化の絵に戻ります
(グリッチ発動ホットキー/glitch_pulse だけは OFF でも効きます)。
対象 T / S ・ 型 float ・ 範囲 0〜100(0.1 刻み)・ 初期値 5
経年劣化してちらつく LED の割合です。該当 LED は個別のリズムで暗く(25%)なったり 完全に消えたりします。低密度パネルなら 0.1%=十数個なので、「ごく数個だけチラつく」 繊細な表現もできます。
対象 T / S ・ 型 float ・ 範囲 0.1〜30 Hz ・ 初期値 8
ちらつきの基準周波数です。LED ごとに 0.5〜1.5 倍の個体差が付くので、 全体が同期してパタパタすることはありません。
対象 T / S ・ 型 float ・ 範囲 0〜100(0.01 刻み)・ 初期値 0.5
常時故障している LED の割合です。時間で変化しない固定の故障で、内訳は:
- 約 40%: 完全消灯(灰色の死にドット)
- 約 45%: 1 チャンネル死亡の変色ドット(白を表示させると、緑が死んだ LED は マゼンタに見える)
- 約 15%: 2 チャンネル死亡(赤だけ・緑だけ・青だけの単色ドット)
実物の許容基準は 0.01% 程度なので、0.01 刻みで実機相当の値も設定できます。 消えた LED はグロー(にじみ)にも寄与しません。
対象 T / S ・ 型 float ・ 範囲 0〜100(0.01 刻み)・ 初期値 0.2
常時フル点灯に固着した LED の割合です。ドライバがフル出力でラッチした状態の 再現なので、映像内容に関係なく白く光り続け、パネルの明るさ設定も無視します。 断線やドロップアウトの影響も受けません(データと無関係に光るため)。ただし 電源系(パネル電源 OFF・テープ1本消灯・バス断線)では消えます。
恒久ドット抜けと同じ LED で重なった場合、死んだチャンネルの分だけ白から色が 欠けます: 1 チャンネル死亡ならマゼンタ/シアン/黄、2 チャンネル死亡なら 赤だけ・緑だけ・青だけの点灯になります(両方の率を上げるほど重なりが増えます)。
対象 T ・ 型 bool + int(1〜1024)・ 初期値 OFF / 1
電源コネクタが抜けたテープを 1 本再現します。指定番号(横配線なら上から、 縦配線なら左から数える)のテープの LED が全消灯し、テープ自体は残って見えます。 LED 数を超える番号を指定した場合は何も起きません。
対象 S ・ 型 bool + int(1〜1024)・ 初期値 OFF / 1
画面の縁を走る給電・データバスの断線です。指定したライン以降のすべてのラインが 消灯します(大型ビジョンで「下半分が真っ暗」になるあの故障に相当)。
対象 T / S ・ 型 bool ・ 初期値 OFF
配線の断線を再現します。伝播範囲は配線方式で変わります。
- 蛇行(T): 断線位置以降のパネル全体にデータが届かなくなる
- くし形(S): 指定した1 ラインだけが途中から死ぬ(他のラインは正常)
対象 T / S ・ 型 int ・ 選択肢 0=フリーズ / 1=消灯 ・ 初期値 0
- フリーズ(データ線切断): 断線した瞬間の映像を保持し続ける。実物の WS2812B はデータが来なければ最後の PWM 値を出し続けるので、これが物理的に正解
- 消灯(電源切断): 完全に暗くなる
対象 T / S ・ 型 float ・ 範囲 0〜100% ・ 初期値 50
蛇行ではシリアルチェーン全体に対する位置、くし形ではライン内での位置(給電側から)です。
対象 S ・ 型 int ・ 範囲 1〜1024 ・ 初期値 1
くし形配線で、どのラインが断線しているか(上から/左から数えた番号)。
対象 T / S ・ 型 float ・ 範囲 0〜10 回/秒 ・ 初期値 0.5
一瞬データが途切れるイベントの発生頻度(1 秒あたりの期待回数)です。0 で発生しません。 イベント時は下の 2 種類のどちらかがランダムに起きます:
- フレーム保持型(確率 1/3): パケットが 1 フレーム丸ごと届かず、全 LED が 直前の映像を保持 → 映像が一瞬「カクッ」と止まる
- ビット化け型(確率 2/3): ランダムな位置から下流(くし形では 1 ライン内)が でたらめな GRB 値を受信 → 彩度の高いランダム色が明滅するスパークル (うち約 35% は黒なのでザラつく)
対象 T / S ・ 型 float ・ 範囲 20〜1000 ms ・ 初期値 120
1 イベントの持続時間です。実際は ±50% のジッタが付きます。イベント終了後は 次のフレームで正常データが再ラッチされ、残留なく復帰します。
テープ構造を画面上で確認するための拡大機能です。表示のみで、エンコードされる シリアルデータや劣化の挙動には一切影響しません。
フィルタ設定画面を閉じると自動的に 1x に戻ります(配信・録画に拡大が残らない ようにするため)。拡大中心は保持されるので、次に開いたときすぐ同じ場所を見られます。
対象 T / S ・ 型 float ・ 範囲 1.0〜16.0 ・ 初期値 1.0(無効)
対象 T / S ・ 型 float ・ 範囲 0〜100% ・ 初期値 50
拡大窓がパネル外へはみ出さないよう自動でクランプされます。
対象 T / S ・ 型 bool ・ 初期値 OFF
パネルがデコードしているシリアルストリームテクスチャ(チェーン順の GRB パケット)を そのまま表示します。配線モードを切り替えると並びが変わるので、蛇行/くし形の エンコード順を目で確認できます。断線や通信途切れがストリーム上でどう見えるかも 分かります。
型 int ・ 初期値 0
前回と違う整数を書き込むと、ビット化けグリッチが 1 回だけ発火します (劣化グループが OFF でも有効)。ドックの「グリッチ」ボタンと同じ仕組みです。
| 症状 | 原因と対処 |
|---|---|
| テープ構造が見えず、ぼんやりした映像になる | LED 密度が出力解像度に対して高すぎます(1 LED あたり 4 画素未満)。密度を下げるか「パネル描画解像度」を 4K にしてください |
| 隙間から後ろが見えない | ① テープ幅が 100%(=隙間なし)。② シーン上でこのソースの下に別のソースが無い。③ フィルタ設定画面のプレビューで確認している(プレビューはソース単体を黒背景に描くため後ろに何も無い)。シーン側のプレビューで確認してください |
| 断線・劣化が効かない | 「信号劣化シミュレーション」グループのチェックが OFF です(グループ全体の有効化が必要) |
| 3D飛び出しが効かない | ① LED高さが 0%。② 「自動」なのに同じソースに 3D Effect フィルタが無い、または回転が 0。③ 縮小表示で構造フォールバック中(視差も一緒にフェードします) |
| 飛び出す方向が逆に見える | LED高さをマイナス値にすると反転します |
| 拡大が勝手に戻る | 仕様です。確認用のため、設定画面を閉じると 1x に戻ります |
| ドックで操作できない | ドックはフィルタ種別ごとに 2 つあります(「LEDテープウォール」「透明LEDスクリーン」)。対象のフィルタに対応したドックを使い、「更新」でソース一覧を再取得してください |
This plugin provides three effect filters:
| Filter | Internal ID |
|---|---|
| LED Strip Wall | led_strip_wall |
| Transparent LED Screen | led_transparent_screen |
| HUB75 LED Panel | led_hub75_panel |
"Key" is the setting name used by obs-websocket (SetSourceFilterSettings) and in scene
collection JSON. "For" is T = LED Strip Wall, S = Transparent LED Screen, H = HUB75 LED
Panel. Items marked "T / S" that aren't otherwise noted work identically on H too (power,
glow, brightness, flicker, dead/stuck dots, dropouts, 3D pop-out, inspection zoom, debug).
For T / S ・ Type bool ・ Default ON
Turns every LED off with a short (~0.13 s) fade. The tape, pads, packages and glass substrate stay visible, so it reads as a powered-down LED wall rather than a hidden source. Also driveable from the hotkey, the dock and obs-websocket.
For T / S ・ Type int ・ Range 8–1024 ・ Default T: 160 × 90, S: 96 × 54
The LED density — the panel's effective resolution. The source is sampled on this grid and exactly this many serial packets are generated.
- Higher density looks closer to the original image; lower density gives a coarse wall
- Below roughly 4 output pixels per LED the structure cannot be resolved, so the shader fades to an area-averaged panel (moiré protection). Practical ceilings for visible structure: ~480×270 at 1080p output, ~960×540 at 4K
- The maximum is 1024×1024 (~1.05 M LEDs); the stream texture folds at 1024 texels wide
For T / S ・ Type int ・ Options 0 = follow source / 1 = 1920×1080 / 2 = 3840×2160 ・ Default 0
The resolution the filter outputs. A fixed mode redraws the strip structure at that resolution regardless of the source size, so a 1080p source set to "4K" produces crisp structure on a 4K canvas (the panel is generated at 4K rather than upscaled). Going the other way, a 4K source on a 1080p canvas can use "1080p" to cut the display pass cost to a quarter.
- Sources with a different aspect ratio are stretched
- Colour sampling (the serial data) depends only on LED density, so encoding and degradation behave identically in every mode
For T / S ・ Type int ・ Default T: 0, S: 2
The order LEDs are chained in. This decides how breaks and bit-garbage propagate.
| Value | For | Meaning |
|---|---|---|
| 0 | T | Horizontal serpentine: 1234 / 8765 / 9ABC — folds at the end of each row |
| 1 | T | Vertical serpentine: 167C / 258B / 349A — folds at the end of each column |
| 2 | S | Comb, horizontal: every row is its own channel, all fed left → right |
| 3 | S | Comb, vertical: every column is its own channel, all fed top → bottom |
Serpentine (0/1) models one long tape folded back and forth, so a break kills the whole downstream panel. Comb (2/3) models real ITO screens — driver ICs on the edge feeding each line in parallel — so faults stay inside one line.
For T / S ・ Type float ・ Range 0.0–2.0 ・ Default 0.8
Spread of the emitted light: Gaussian contributions from the LED itself plus its four neighbours are summed. Higher values give a bloomed, bright wall; 0 gives crisp point sources. On see-through panels the glow is also added over the gaps, so raising it produces a "curtain of light" look.
For T / S ・ Type float ・ Range 0.1–2.0 ・ Default 1.0
Luminance multiplier. 1.0 matches the source; higher values push lens highlights toward clipping, lower values read as a panel run at reduced output.
For T ・ Type int ・ Options 0 = black / 1 = white ・ Default 0
Colour of the flexible substrate. Black maximises contrast in the dark; white reads as a real white-FPC strip under room light.
For T ・ Type int ・ Options 1 = white / 0 = black ・ Default 1
Colour of the plastic body. White is the common WS2812B; black is the black-framed variant (WS2812B-B and similar) — its unlit lens is darker too, so when lit only the lens emission stands out.
For T ・ Type float ・ Range 5–100 ・ Default 62
Width of the strip relative to the LED pitch. Below 100% the gaps become alpha transparent, so sources placed under this one in the scene show through (see-through panel). At 100% the tapes butt together and the panel is fully opaque — no half- transparent seam is left on the cell boundary either.
For T ・ Type float ・ Range 10–100 ・ Default 55
Size of the package relative to the LED spacing along the tape. Smaller values leave more bare tape visible between LEDs, like a sparsely populated strip.
For T ・ Type float ・ Range 0–100 ・ Default 50
Brown tone of the solder pads: 0% bright copper, 100% dark brown. As on the real part the pads are split into three contacts (+5V / DATA / GND). The split always runs across the strip, so on horizontal wiring you get three short bars stacked one above the other, and on vertical wiring three bars standing side by side:
horizontal wiring vertical wiring
--- | | |
--- | | |
--- | | |
For S ・ Type float ・ Range 50–98 ・ Default 85
Transmittance of the glass/acrylic substrate; real transparent screens reach roughly 70–95%. The resulting opacity (1 − transmittance) covers the whole area, and whatever is behind shows through by the same amount. Lower values deepen the substrate tint.
For S ・ Type float ・ Range 5–60 ・ Default 18
Diameter of the LED dot. Real screens show tiny specks floating in the glass, hence the small default. Unlit dots remain as faint grey flecks.
For H ・ Type int (value = width×1000 + height) ・ Options 64×64 / 64×32 / 32×32 / 32×16 ・ Default 64×64
LEDs per module (one physical tile) — matching real P2.5 (64×64) or P4/P5 (64×32) parts.
For H ・ Type int ・ Range 1–16 ・ Default 6 × 4
How many modules are tiled into the wall. Total LEDs = module resolution × count (the
count is limited automatically if one axis would exceed 1024 LEDs). The direct LED-count
keys (led_cols/led_rows) are not used by the HUB75 filter.
For H ・ Type float ・ Range 0–100% ・ Default 60
Visibility of the joints between modules — the faint grid on large video walls. 0% removes it.
For H ・ Type float ・ Range 20–95% ・ Default 70
Size of the rounded-square SMD package relative to the pitch; real high-density panels sit around 60–80%.
For H ・ Type int ・ Options 8 (1/8) / 16 (1/16) / 32 (1/32) ・ Default 16
Multiplexing ratio of the drive. Sets the number of camera scan bands and underlies the address-line fault.
For H ・ Type float ・ Range 0–100% / 0.1–10 Hz ・ Default 0 (off) / 2
The dark bands that roll through the picture when a camera films the panel (rolling shutter beating against the refresh). Invisible to the naked eye on real panels, hence off by default — raise it for a "filmed LED wall" look.
For H ・ Type float + int ・ Range 0–100% (0.1 steps) / 0 = dark, 1 = freeze, 2 = garbage ・ Default 0 / dark
Share of failed modules (whole tiles): the classic "one tile is black / frozen / snowed with garbage". Positions are hash-random but static over time. Freeze holds the frame that was on the tile when it failed — physically correct via the latch stage.
For H ・ Type float ・ Range 0–100% ・ Default 0 / 50
Share of modules aged into a reddish tint, and how strong the shift is. Blue and green LEDs (InGaN) lose brightness faster than red ones (AlGaInP), so old modules darken in the order B > G > R and drift warm (implementation: B×(1−0.50s), G×(1−0.35s), R×(1−0.10s), with s randomised per module). Reproduces the patchy look of mixed-batch or partially-replaced walls. Loosely called burn-in, but properly this is ageing colour shift — burn-in (image retention) is a different phenomenon.
For H ・ Type bool + int (1–256) + int ・ Default OFF / 1 / freeze
A broken HUB75 cable between modules. Counting along the chain (modules snake: row 1 left→right, row 2 right→left, …), the given module and everything after it freezes (holds its last picture) or goes dark.
For H ・ Type float ・ Range 0–100% (0.1 steps) ・ Default 0
Share of dead driver ICs. Each one takes out the 16-column block in one half of its module that it drives.
For H ・ Type bool ・ Default OFF
A broken row-address line: half the scan rows are never selected, leaving every other row dark inside each module.
When the source is tilted in 3D with a plugin such as 3D Effect (by exeldro), this makes only the LEDs appear raised above the substrate. With zero tilt the render is pixel-identical to having the feature off.
For T / S ・ Type float ・ Range -100–100 ・ Default 5
How far the LEDs sit above the substrate; parallax = height × tan(tilt).
- 0% disables the effect
- If the pop-out direction looks inverted, use a negative value to flip it
- Larger values exaggerate the effect, but parallax is clamped to 0.6 pitch
For T / S ・ Type int ・ Options 0 = automatic / 1 = manual ・ Default 0
Automatic reads rot_x / rot_y / rot_z from a 3D Effect filter (internal id
3d_effect_filter) on the same source every frame, rescanning the filter chain every
0.25 s. Moving the 3D Effect sliders is enough — the LEDs follow. Without such a filter
no parallax is applied.
Manual uses the tilt sliders below. The approximation is orthographic, so with a strong FOV in 3D Effect, fine-tune manually.
For T / S ・ Type float ・ Range -85–85 degrees ・ Default 0
Manual tilt: X pitches up/down, Y turns left/right.
For T / S ・ Type bool (group checkbox) ・ Default OFF
Master switch for the group. While OFF the picture is completely undegraded (the
glitch hotkey and glitch_pulse still fire, by design).
For T / S ・ Type float ・ Range 0–100 (0.1 steps) ・ Default 5
Share of LEDs that have aged into flickering. Affected LEDs dim (to 25%) or go dark on their own rhythm. On a low-density panel 0.1% is a dozen or so LEDs, so "just a few specks flickering" is achievable.
For T / S ・ Type float ・ Range 0.1–30 Hz ・ Default 8
Base flicker frequency. Each LED gets a 0.5–1.5× individual variation, so the panel never blinks in lockstep.
For T / S ・ Type float ・ Range 0–100 (0.01 steps) ・ Default 0.5
Share of permanently failed LEDs — a static fault that does not change over time:
- ~40%: fully dark (grey dead dot)
- ~45%: one dead channel — an LED with a dead green channel shows magenta where white was intended
- ~15%: two dead channels — a pure red, green or blue dot
Industry tolerance is around 0.01%, and the 0.01 step lets you dial in exactly that. Dead LEDs also stop contributing to the glow.
For T / S ・ Type float ・ Range 0–100 (0.01 steps) ・ Default 0.2
Share of LEDs latched at full output: the driver is stuck at full duty, so they burn white regardless of the picture — and regardless of the panel brightness setting. Breaks and dropouts don't affect them either (they ignore the data), but the power paths still do (panel power off, one strip powered off, broken edge bus).
Where a stuck dot overlaps a permanent dead dot, the dead channels drop out of the white: one dead channel leaves magenta/cyan/yellow, two dead channels leave a pure red, green or blue glow. Raising both rates makes such overlaps more frequent.
For T ・ Type bool + int (1–1024) ・ Default OFF / 1
Simulates one tape with a loose power connector: every LED on the numbered strip (counted from the top, or from the left for vertical wiring) goes dark while the tape itself stays visible. Numbers beyond the LED count do nothing.
For S ・ Type bool + int (1–1024) ・ Default OFF / 1
A break in the power/data bus running along the screen edge: the given line and every line after it go dark — the "bottom half is black" failure seen on large video walls.
For T / S ・ Type bool ・ Default OFF
A broken trace. How far it propagates depends on the wiring:
- Serpentine (T): everything after the break point — the whole downstream panel
- Comb (S): only the selected single line, from the break point on
For T / S ・ Type int ・ Options 0 = freeze / 1 = dark ・ Default 0
- Freeze (data line cut): the affected LEDs hold the frame they latched at the moment of the break. This is the physically correct behaviour — a real WS2812B keeps outputting its last PWM values when no data arrives
- Dark (power cut): they go completely black
For T / S ・ Type float ・ Range 0–100% ・ Default 50
Position along the whole serial chain (serpentine) or within the line, counted from the feed side (comb).
For S ・ Type int ・ Range 1–1024 ・ Default 1
Which line carries the break in comb wiring (counted from the top, or from the left).
For T / S ・ Type float ・ Range 0–10 per second ・ Default 0.5
Expected number of momentary data-loss events per second; 0 disables them. Each event is randomly one of:
- Frame hold (1 in 3): no packet arrives at all, so every LED holds the previous frame — the picture hitches for an instant
- Bit garbage (2 in 3): from a random point downstream (within one line in comb mode) LEDs latch nonsense GRB values — a saturated random-colour sparkle that re-rolls every frame (about 35% of them land on black, so it reads as grain rather than a solid wash)
For T / S ・ Type float ・ Range 20–1000 ms ・ Default 120
Duration of one event, with ±50% jitter applied. When it ends the next frame latches valid data again, so recovery is instant with no residue.
A magnifier for checking the strip structure on screen. Display only — the encoded serial data and the degradation behaviour are not affected at all.
It snaps back to 1x when the filter settings dialog closes, so nothing magnified is ever left on the stream or recording. The zoom centre is kept, so reopening the dialog lands you on the same spot.
For T / S ・ Type float ・ Range 1.0–16.0 ・ Default 1.0 (off)
For T / S ・ Type float ・ Range 0–100% ・ Default 50
The magnified window is clamped so it never leaves the panel.
For T / S ・ Type bool ・ Default OFF
Displays the raw serial stream texture (GRB packets in chain order) that the panel decodes. Switching wiring modes visibly rearranges it, so you can verify the serpentine and comb encode order by eye — and see what a break or dropout does to the stream itself.
Type int ・ Default 0
Writing an integer different from the last one fires a single bit-garbage glitch burst (works even with the degradation group off). This is the same mechanism the dock's Glitch button uses.
| Symptom | Cause and fix |
|---|---|
| No visible strip structure, just a soft image | LED density is too high for the output resolution (under 4 pixels per LED). Lower the density or set "Panel render resolution" to 4K |
| Nothing shows through the gaps | (1) Tape width is 100% (no gaps). (2) No source is placed under this one in the scene. (3) You are looking at the filter dialog's preview, which draws the source alone on black — check the scene preview instead |
| Breaks / degradation do nothing | The "Signal degradation" group checkbox is off; the group itself must be enabled |
| 3D pop-out does nothing | (1) LED height is 0%. (2) Mode is automatic but there is no 3D Effect filter on the same source, or its rotation is 0. (3) The panel is minified into the structure fallback — parallax fades out with it |
| Pop-out direction looks inverted | Use a negative LED height to flip it |
| The zoom keeps resetting itself | By design: it is an inspection aid and returns to 1x when the dialog closes |
| The dock won't control my filter | There are two docks, one per filter type ("LED Strip Wall" and "Transparent LED Screen"). Use the matching one and press Refresh to rebuild the source list |