# VapourWiki — Audion VS Engine preset reference (EN)

**Contents**

- [Before you start: install the engine](#before-you-start-install-the-engine)
- [Decision tree — which preset for which task](#decision-tree--which-preset-for-which-task)
- [PRECISION palette — technical pipeline (8 presets)](#precision-palette--technical-pipeline-8-presets)
- [FILM_LOOKS palette — film emulations (5 presets)](#filmlooks-palette--film-emulations-5-presets)
- [RETRO palette — analog character (2 presets)](#retro-palette--analog-character-2-presets)
- [RESTORATION palette — heavy guns (7 presets, Phase 18.B + NNEDI3 2x)](#restoration-palette--heavy-guns-7-presets-phase-18b--nnedi3-2x)
- [Smoke / bench status](#smoke--bench-status)
- [Encoders — my patterns](#encoders--my-patterns)
- [More — full reference](#more--full-reference)

Full English reference for all 22 presets across 4 palettes. Starts with a **decision tree** ("which preset for which task"). Followed by detailed per-preset description: what it does, on what footage, key parameters, recommended encoder.

> Per-preset docstrings live inside each `.vpy` file in `system_core/presets/<palette>/`. This document is the navigation-friendly companion.

---

## Before you start: install the engine

The build ships without VapourSynth and its plugins — some
seventy modules, each under its own licence.
The program installs them itself, from their authors.

Until that is done, not one of the presets described below will run.

Run `builder_main.cmd` and pick, in order: `VAPOURSYNTH` (10), `VS PLUGINS`
(11). The order matters: plugins
need the engine already in place.

## Decision tree — which preset for which task

Read top to bottom: the first matching case is your preset.

| Footage symptom | Preset | Palette |
|---|---|---|
| Interlaced legacy (DV, HDV, VHS digitization, broadcast TS) | **`qtgmc_deinterlace`** | restoration |
| NTSC 29.97 fps with 3:2 pulldown (telecine, film → video) | **`tivtc_ivtc`** | restoration |
| High-ISO noise, low-light digital, want to keep detail | **`mvtools_mcdegrain`** | restoration |
| Rainbow / dot crawl on VHS-rip / composite capture | **`derainbow_decross`** | restoration |
| Over-compressed H.264/MPEG (YouTube-rip, WhatsApp, SD broadcast) | **`deblock_h264_artefacts`** | restoration |
| Haze / low contrast / want to "lift" the image | **`dehaze_local_contrast`** | restoration |
| Digital noise only in shadows ("low-light" digital footage) | **`shadow_denoise_sota`** ⭐ | precision |
| Mild uniform noise, hardware-agnostic (no CUDA/OpenCL) | `mild_denoise` | precision |
| Chroma noise only (dirty blue channel) | `chroma_cleanup` | precision |
| Banding in gradients (sky, walls) | `deband_safe` or `deband_fine_grain` | precision |
| Want one preset "everything" for filmic material | **`filmic_rebuild`** ⭐ | precision |
| Pre-grade prep for DaVinci | `pregrade_prep` | precision |
| Clean archival master, no grain | `archive_clean` | precision |
| Subtle filmic, safe default | `cinematic` | film_looks |
| Specific Kodak 35mm stock (250D / 500T / 50D) | `film_35mm` | film_looks |
| 16mm organic, lifted blacks | `film_16mm` | film_looks |
| Super 8, faded 70s, heavy grain | `super8` | film_looks |
| High contrast desaturated (Se7en, Saving Private Ryan) | `bleach_bypass` | film_looks |
| VHS / CRT aesthetic | `vhs_crt` | retro |
| 90s amateur camcorder | `camcorder_90s` | retro |

**Pipeline scenarios** (multiple presets chained via `apply-profile-batch` or manually):

| Scenario | Steps |
|---|---|
| Night timelapse → film | `shadow_denoise_sota` → `filmic_rebuild` → `cinematic` |
| VHS archive → restored master | `qtgmc_deinterlace` → `derainbow_decross` → `mvtools_mcdegrain` → `archive_clean` |
| Old YouTube-rip → project-usable | `deblock_h264_artefacts` → `dehaze_local_contrast` → `pregrade_prep` |
| Telecined NTSC DVD → 24p master | `tivtc_ivtc` → `archive_clean` |
| Modern digital → film look | `mild_denoise` → `film_35mm` |

---

## PRECISION palette — technical pipeline (8 presets)

Menu order = pipeline order: Stage 1 (denoise) → Stage 2 (deband) → Stage 3 (compositions).

### Stage 1 — denoise

#### `mild_denoise` — gentle universal denoise

DFTTest spectral denoiser. Pure CPU, hardware-agnostic — works the same on any processor. Recommended as a **first pick** when you don't know much about the footage: light noise, small touch-up, low risk.

- Parameters: `strength` = light (sigma=4.0) / medium (8.0) / strong (14.0)
- Encoder: `h264_crf17` or `h264_crf14` for archival quality
- Adobe/DaVinci equivalent: only Temporal NR / Noise Reduction, much coarser result

#### `shadow_denoise_sota` ⭐ — flagship shadow denoiser

BM3D with float32 pipeline + smooth luma-mask "shadows only". Noise is suppressed **only** in zones below `shadow_threshold`, the rest of the image is untouched. Chroma planes are always denoised (chroma noise is ugly everywhere). Optional micro-grain return via `grain_back` so shadows don't look "plastic".

- Parameters: `sigma=2.5` (BM3D luma), `use_cuda=0/1`, `grain_back=0.6`, `shadow_threshold=0.20`, `transition=0.10`
- On NVIDIA: `--use-cuda 1` gives 25–35% wall-time gain on 4K
- Encoder: `h264_crf17` / `prores_lt`
- Not in NLE: zone-targeted denoise with smooth blend, no cheap "threshold mask"

#### `chroma_cleanup` — chroma-only cleanup

DFTTest on planes=[1,2]. Luma untouched. Use when "blue channel is dirty" (typical issue of old compact cameras, low-bitrate AVCHD).

- Parameters: `strength` = light / medium / strong
- Encoder: `h264_crf17`
- Not in NLE: pure chroma-only DFTTest with no luma side-effects

### Stage 2 — deband

#### `deband_safe` — safe deband

neo_f3kdb with light settings, no grain. Removes banding in gradients (sky, walls, night lighting) without losing sharpness.

- Parameters: `range=15`, `y=64`, `cb=64`, `cr=64`
- Encoder: `h264_crf17` / `h265_crf17`

#### `deband_fine_grain` — deband + fine grain restore

Same neo_f3kdb + AddGrain. After debanding a thin monotonic grain is added — prevents the "too clean" look (signature of cheap compression).

- Parameters: `range=15`, `grain_var=1.5`
- Encoder: `prores_lt` (for downstream grading)

### Stage 3 — compositions

#### `filmic_rebuild` ⭐ — flagship composition

Full chain: BM3D denoise → neo_f3kdb deband → 3-zone luma grain (shadows / midtones / highlights). Grain is zoned — heavier in shadows (like real film), lighter in highlights. Recommended as a "one preset for everything" on filmic material.

- Parameters: `sigma=2.0`, `use_cuda=0/1`, `deband_range=14`, `grain_shadow=1.5`, `grain_mid=0.9`, `grain_high=0.4`, `shadow_threshold=0.30`, `high_threshold=0.65`
- Encoder: `prores_lt` / `h265_crf17`

#### `archive_clean` — neutral archival master

DFTTest + neo_f3kdb. **No grain.** Goal — maximally clean "flat" picture for archival. Don't use if grading is planned (grading works better on grainy material).

- Parameters: `denoise=medium`, `range=15`
- Encoder: `prores_422hq` / `prores_422hq_mxf`

#### `pregrade_prep` — DaVinci handoff

Minimum touch: only light DFTTest. No deband, no grain. Goal — give Resolve the cleanest source so the colorist doesn't grade on top of compression artefacts.

- Parameters: `strength=light`
- Encoder: `prores_lt_mxf` (for Adobe / Avid round-trip)

---

## FILM_LOOKS palette — film emulations (5 presets)

Each look uses the shared `audion_lib` library (MTF softening, halation bloom, zone grain, gamma curve, black-lift, desaturation).

### `cinematic` — universal subtle filmic

Safe default. Light MTF softening + halation + micro-grain. Not tied to any specific stock. Works on any modern digital footage.

- Parameters: `intensity=1.0` (range 0..2)
- Encoder: `prores_lt` / `h264_crf17`

### `film_35mm` — Kodak 35mm with stock variants

Emulates real Kodak stocks: 250D (daylight), 500T (tungsten), 50D (fine grain). Each stock has its own gamma curve, balance, grain density.

- Parameters: `stock=250D|500T|50D`, `intensity=1.0`
- Encoder: `prores_lt` / `prores_lt_mxf`

### `film_16mm` — organic, more grainy

Lifted blacks, more pronounced grain, softer than 35mm. Suits documentary / arthouse aesthetic.

- Parameters: `intensity=1.0`
- Encoder: `prores_lt`

### `super8` — strongest look

Heaviest grain, softest optics, faded 70s. For music video / stylized inserts.

- Parameters: `intensity=1.0`
- Encoder: `h264_crf17` (grain already baked in, aggressive compression OK)

### `bleach_bypass` — high contrast desaturated

"Se7en" / "Saving Private Ryan" look. High contrast, silver greys, hot highlights. Strong stylistic choice.

- Parameters: `intensity=1.0`
- Encoder: `prores_lt`

---

## RETRO palette — analog character (2 presets)

### `vhs_crt` — VHS / CRT

Chroma bleed (horizontal color spread), soft optics, analog noise. Emulates VHS tape playback on a CRT TV.

- Parameters: `intensity=1.2`
- Encoder: `h264_crf17`

### `camcorder_90s` — amateur camcorder

Softer than VHS, slight overexposure, minimal chroma bleed. 90s home video aesthetic.

- Parameters: `intensity=1.0`
- Encoder: `h264_crf17`

---

## RESTORATION palette — heavy guns (7 presets, Phase 18.B + NNEDI3 2x)

The set that is **missing or poorly implemented** in Adobe Premiere / DaVinci Resolve.

### `qtgmc_deinterlace` — reference deinterlace

havsfunc.QTGMC (NNEDI3 + MVTools motion estimation). Reconstructs progressive frames from interlaced source via neural-network upscaling of each field + motion-compensated temporal smoothing. **Better than any NLE out-of-the-box.**

- Parameters: `field_order=tff|bff`, `qtgmc-preset=Faster|Fast|Medium|Slow|Slower|Placebo`, `output_fps=single|double`
- When to pick: DV, HDV, VHS digitization, broadcast TS, S-VHS / U-matic transfers
- Encoder: `prores_lt` (for downstream work) or `h264_crf17` (final master)

### `tivtc_ivtc` — inverse telecine

TIVTC.TFM (field matching) + TIVTC.TDecimate (drop duplicate). Reference-quality 3:2 pulldown removal: NTSC 29.97i → 23.976p. Restores the original 24p film master from a telecined source.

- Parameters: `pp=6` (TFM post-processor), `cycle=5`, `rdrop=1` (standard NTSC pattern)
- When to pick: NTSC DVD, broadcast prints, digitized film transfers
- Encoder: `prores_422` / `prores_422hq_mxf`

### `mvtools_mcdegrain` — motion-compensated denoise

MVTools motion estimation → MDeGrain temporal averaging along motion vectors. **Removes noise WITHOUT detail loss** — what Topaz Video Enhance AI and Neat Video do internally. DaVinci Temporal NR is a coarse implementation of the same; this is reference-grade.

- Parameters: `radius=2` (5-frame window), `thsad=200`, `blksize=16` (HD) / `8` (4K detail)
- When to pick: high-ISO low-light footage, preserving skin texture / fabric detail
- Encoder: `prores_lt` / `h265_crf17`

### `derainbow_decross` — NTSC composite cleanup

MVTools-based motion-compensated chroma-only smoothing. Removes rainbow / dot crawl / cross-color on composite-captured footage (VHS-rip, U-matic, BetaSP, S-Video → SDI). Luma untouched.

- Parameters: `strength=0.6`, `blksize=16`
- When to pick: digitized VHS, color "running" on thin lines, mosquito noise on B&W edges
- Encoder: `prores_lt` / `h264_crf17`

### `deblock_h264_artefacts` — over-compressed rescue

havsfunc.Deblock_QED — edge-aware deblocker for H.264 / MPEG-2 / MPEG-4. Smooths 8x8 block boundaries, ringing around edges, mosquito noise. Edge-aware — does not touch real detail.

- Parameters: `quant1=24`, `quant2=26`, `aoffset=1`, `boffset=1`
- When to pick: old YouTube-rips, WhatsApp/Telegram re-encodes, low-bitrate SD broadcast TS
- Encoder: `h264_crf17` (after deblock the material is cleaner, can re-encode at higher CRF)

### `dehaze_local_contrast` — clarity without halos

Luma-only local contrast via high-pass + zone-weighted MaskedMerge. Zone-mask (parabolic, peaks in midtones) prevents highlight burnout and crushed shadows. No color shift (luma-only). 16-bit math.

- Parameters: `strength=1.0`, `radius=8` (clarity) / `12-16` (haze removal)
- When to pick: hazy footage, flat low-contrast material, want to "wake up" the picture without destroying highlights
- Not in NLE: Resolve "Dehaze" burns highlights and shifts color; this one has no halos
- Encoder: `prores_lt` (handoff to colorist) or `h264_crf17` (final)

### `nnedi3_upscale_2x` — deterministic non-ML 2x upscale

ZNEDI3/NNEDI3-based enlargement through `nnedi3_resample`. This is the Lite answer for clean, predictable upscaling without MLRT or ONNX models.

- Parameters: `quality=fast|balanced|best`, `chroma=spline36|nnedi3`
- When to pick: SD/HD archive material that needs a clean 2x resize without hallucinated ML texture
- Encoder: `prores_lt` for handoff or `h264_crf14` / `h265_crf14` for final

---

## Smoke / bench status

`install\Bench-AllPresets.{cmd,ps1}` walks every `.vpy` preset through `vspipe -c y4m --end <Frames-1> | ffmpeg -f null` without writing output files. It checks preset parsing, plugin namespaces, frame delivery, and the CPU/CUDA BM3D path when requested.

Current local references for Lite:
- CPU fallback: **22/22 PASS** with `Frames=1`, `Cuda=off` (2026-05-16).
- RTX 5070 validation: **30/30 PASS** with `Frames=1`, `Cuda=sweep` (2026-05-27). The 30 rows are 22 presets plus the extra CPU/CUDA Precision sweep.

```cmd
system_core\powershell\pwsh.exe -NoLogo -NoProfile -ExecutionPolicy Bypass -File install\Bench-AllPresets.ps1 -ProjectRoot . -InputFile <video> -Frames 1 -Cuda sweep
```

---

## Encoders — my patterns

(Audion default workflow per discussion 2026-04-28)

**Quality ladder** (same 14 / 17 / 21 tiers across software, NVENC, QuickSync, and AMF): **14 → 17 → 21**, three distinguishable steps without overlap.

| Profile | When to use |
|---|---|
| `h264_crf14` / `h265_crf14` ⭐ default | Semi-lossless archive, master for downstream work. **Default Audion since 2026-04-28.** |
| `h264_crf17` / `h265_crf17` | "Almost-invisible" lossy, final master |
| `h264_crf21` / `h265_crf21` | Web preview / proxy |
| `prores_lt` ⭐ | Audion default ProRes — no keying, for grading and round-trip |
| `prores_lt_mxf` ⭐ | ProRes LT in MXF wrapper — for Adobe Premiere / Avid round-trip |
| `prores_422` | When you need a bit more bitrate than LT |
| `prores_422_mxf` | ProRes 422 in MXF wrapper for Adobe / Avid handoff |
| `prores_422hq` | Only if keying is planned |
| `prores_422hq_mxf` | HQ + MXF for broadcast / Avid finishing |
| `dnxhr_lb` / `_sq` / `_hq` / `_hqx` | Avid-style DNxHR (LB low / SQ standard / HQ high / HQX 10-bit) |
| **`h264_nvenc_q14` / `_q17` / `_q21`** | NVENC hardware H.264 on NVIDIA. Removes CPU bottleneck from libx264 — VS filtering doesn't fight encode. **CQ — analog of CRF**. 8-bit yuv420p. |
| **`h265_nvenc_q14` / `_q17` / `_q21`** | NVENC hardware HEVC, 10-bit p010le. Ideal for full pipeline on NVIDIA: VS filtering on CPU/CUDA + encode on NVENC = zero CPU stall. |
| **`h264_qsv_q14` / `_q17` / `_q21`** | Intel QuickSync H.264 for Intel iGPU batch/proxy work. |
| **`h265_qsv_q14` / `_q17` / `_q21`** | Intel QuickSync HEVC, p010le where supported. |
| **`h264_amf_q14` / `_q17` / `_q21`** | AMD AMF H.264 via CQP. |
| **`h265_amf_q14` / `_q17` / `_q21`** | AMD AMF HEVC, p010le output. |

**Rule**: for any non-final processing → `prores_lt` / `prores_lt_mxf` or DNxHR if the receiver prefers it. For final delivery → `h264_crf14` / `h264_crf17` (software, best density per bit) OR the matching hardware ladder for the host (`nvenc`, `qsv`, `amf`) when wall-time matters.

### When to use hardware encode vs software

| Scenario | Encoder |
|---|---|
| Heavy VS pipeline (filmic_rebuild, mvtools_mcdegrain) on NVIDIA | **NVENC** — CPU is freed for VS filter, total wall-time drops 30-50% |
| Final delivery where bit density matters most | **libx264/libx265 CRF** — software encoder is still slightly more accurate at low CRF |
| Intel iGPU host / laptop batch | **QuickSync** — good throughput with low CPU pressure |
| AMD/Radeon host | **AMF** — same 14/17/21 CQP ladder, avoids CPU encode bottlenecks |
| Large batch of hundreds of files | **NVENC / QSV / AMF** — savings accumulate |
| No working hardware encoder | Software only (`h264_crf*` / `h265_crf*`) |

NVENC requires: NVIDIA driver R525+. Quality: NVENC on Ada/Blackwell (RTX 40/50) is **comparable** to libx264 medium at the same CQ; on older generations (Pascal/Turing) software CRF wins slightly on bitrate efficiency at the same quality, but NVENC is still many times faster.

---

## More — full reference

- English per-preset reference: docstring in `system_core/presets/<palette>/<preset>.vpy`
- CLI flags: `runtime\python.exe system_core\main.py run --help`
- Profiles (cross-palette combinations): `runtime\python.exe system_core\main.py list-profiles`
- Recursive batch with mirror folder structure: `apply-profile-batch --recursive`
- Stack health check: `runtime\python.exe system_core\doctor.py`

---

**Last updated**: 2026-05-27 (Phase 18.B + NNEDI3 2x — 22 presets across 4 palettes; CPU fallback smoke `22/22 PASS`, `Frames=1`, `Cuda=off`; RTX 5070 CUDA smoke `30/30 PASS`, `Frames=1`, `Cuda=sweep`)