debanders ¶
Functions:
-
f3k_deband
–Debanding filter wrapper using the
neo_f3kdb
plugin. -
mdb_bilateral
–Multi stage debanding, bilateral-esque filter.
-
pfdeband
–Prefilter and deband a clip.
-
placebo_deband
–Debanding wrapper around the
placebo.Deband
filter from the VapourSynthvs-placebo
plugin.
F3KDeband ¶
F3KDeband(f3k_deband: Callable[P, R])
Bases: Generic[P, R]
Class decorator that wraps the f3k_deband function and extends its functionality.
It is not meant to be used directly.
Classes:
-
RandomAlgo
–Random number generation algorithm used for reference positions or grain patterns.
-
SampleMode
–Enum that determines how reference pixels are sampled for debanding.
Methods:
-
__call__
–
Source code in vsdeband/debanders.py
43 44 |
|
RandomAlgo ¶
Bases: CustomIntEnum
Random number generation algorithm used for reference positions or grain patterns.
Methods:
-
__call__
–Configure the standard deviation for the GAUSSIAN algorithm.
Attributes:
-
GAUSSIAN
–Gaussian distribution. Supports custom standard deviation via
__call__
. -
OLD
–Legacy algorithm from older versions.
-
UNIFORM
–Uniform distribution for randomization.
-
sigma
(float | None
) –Standard deviation value used only for GAUSSIAN.
Source code in vsdeband/debanders.py
319 320 321 |
|
GAUSSIAN class-attribute
instance-attribute
¶
GAUSSIAN = 2
Gaussian distribution. Supports custom standard deviation via __call__
.
sigma instance-attribute
¶
sigma: float | None = sigma
Standard deviation value used only for GAUSSIAN.
__call__ ¶
Configure the standard deviation for the GAUSSIAN algorithm.
Only values in the range [-1.0, 1.0]
are considered valid and used as multipliers. Values outside this range are ignored.
Parameters:
Returns:
Source code in vsdeband/debanders.py
338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 |
|
SampleMode ¶
Bases: CustomIntEnum
Enum that determines how reference pixels are sampled for debanding.
Methods:
-
__call__
–Configure
MEAN_DIFF
with either individual values or sequences.
Attributes:
-
COLUMN
–Take 2 reference pixels from the same column as the current pixel.
-
COL_ROW_MEAN
–Take the arithmetic mean of COLUMN and ROW sampling patterns.
-
MEAN_DIFF_FLOAT
–Similar to COL_ROW_MEAN, but includes configurable maximum and median
-
MEAN_DIFF_GRADIENT
–An extension of MEAN_DIFF_FLOAT that adds a gradient angle check for smarter detail preservation.
-
MEAN_DIFF_INT
–Similar to COL_ROW_MEAN, but includes configurable maximum and median
-
ROW
–Take 2 reference pixels from the same row as the current pixel.
-
SQUARE
–Take 4 reference pixels forming a square pattern around the current pixel.
-
kwargs
(dict[str, Any]
) –Additional keyword arguments.
Source code in vsdeband/debanders.py
57 58 59 |
|
COLUMN class-attribute
instance-attribute
¶
COLUMN = 1
Take 2 reference pixels from the same column as the current pixel.
COL_ROW_MEAN class-attribute
instance-attribute
¶
COL_ROW_MEAN = 4
Take the arithmetic mean of COLUMN and ROW sampling patterns. Reference pixels are randomly selected from both configurations.
MEAN_DIFF_FLOAT class-attribute
instance-attribute
¶
MEAN_DIFF_FLOAT = 6
Similar to COL_ROW_MEAN, but includes configurable maximum and median difference thresholds for finer control and detail preservation.
MEAN_DIFF_GRADIENT class-attribute
instance-attribute
¶
MEAN_DIFF_GRADIENT = 7
An extension of MEAN_DIFF_FLOAT that adds a gradient angle check for smarter detail preservation.
MEAN_DIFF_INT class-attribute
instance-attribute
¶
MEAN_DIFF_INT = 5
Similar to COL_ROW_MEAN, but includes configurable maximum and median difference thresholds for finer control and detail preservation.
ROW class-attribute
instance-attribute
¶
ROW = 3
Take 2 reference pixels from the same row as the current pixel.
SQUARE class-attribute
instance-attribute
¶
SQUARE = 2
Take 4 reference pixels forming a square pattern around the current pixel.
__call__ ¶
__call__(
y1: int | None = None,
cb1: int | None = None,
cr1: int | None = None,
y2: int | None = None,
cb2: int | None = None,
cr2: int | None = None,
) -> Literal[MEAN_DIFF_INT]
__call__(
thr_max: Sequence[int | None] | None = None,
thr_mid: Sequence[int | None] | None = None,
) -> Literal[MEAN_DIFF_INT]
__call__(
y1: int | None = None,
cb1: int | None = None,
cr1: int | None = None,
y2: int | None = None,
cb2: int | None = None,
cr2: int | None = None,
) -> Literal[MEAN_DIFF_FLOAT]
__call__(
thr_max: Sequence[int | None] | None = None,
thr_mid: Sequence[int | None] | None = None,
) -> Literal[MEAN_DIFF_FLOAT]
__call__(
y1: int | None = None,
cb1: int | None = None,
cr1: int | None = None,
y2: int | None = None,
cb2: int | None = None,
cr2: int | None = None,
/,
angle_boost: float | None = None,
max_angle: float | None = None,
) -> Literal[MEAN_DIFF_GRADIENT]
__call__(
thr_max: Sequence[int | None] | None = None,
thr_mid: Sequence[int | None] | None = None,
/,
*,
angle_boost: float | None = None,
max_angle: float | None = None,
) -> Literal[MEAN_DIFF_GRADIENT]
__call__(
y1_or_thr_max: int | None | Sequence[int | None] = None,
cb1_or_thr_mid: int | None | Sequence[int | None] = None,
cr1: int | None = None,
y2: int | None = None,
cb2: int | None = None,
cr2: int | None = None,
angle_boost: float | None = None,
max_angle: float | None = None,
) -> Any
Configure MEAN_DIFF
with either individual values or sequences.
You can use either: - Individual thresholds: y1, cb1, cr1, y2, cb2, cr2
- Sequences: thr_max
and thr_mid
y1 / cb1 / cr1¶
These are detail protection thresholds (max difference).
This threshold applies to the maxDif check. maxDif is the largest absolute difference found between the current pixel and any of its four individual cross-shaped reference pixels. If this maxDif is greater than or equal to y1 / cb1 / cr1, the pixel is considered detail.
It helps protect sharp edges and fine details from being blurred by the debanding process.
y2 / cb2 / cr2¶
These are gradient / texture protection threshold (mid-pair difference).
This threshold applies to the midDif checks. midDif measures how much the current pixel deviates from the midpoint of a pair of opposing reference pixels (one check for the vertical pair, one for the horizontal pair). If the current pixel is far from this midpoint (i.e., midDif is greater than or equal to y2 / cb2 / cr2), it might indicate a texture.
This helps distinguish true banding in gradients from textured areas or complex details.
Parameters:
-
y1_or_thr_max
¶int | None | Sequence[int | None]
, default:None
) –maxDif threshold for luma (Y) or sequence of max thresholds.
-
cb1_or_thr_mid
¶int | None | Sequence[int | None]
, default:None
) –maxDif threshold for blue-difference chroma (U) or sequence of mid thresholds.
-
cr1
¶int | None
, default:None
) –maxDif threshold for red-difference chroma (V).
-
y2
¶int | None
, default:None
) –midDif threshold for luma (Y).
-
cb2
¶int | None
, default:None
) –midDif threshold for blue-difference chroma (U).
-
cr2
¶int | None
, default:None
) –midDif threshold for red-difference chroma (V).
-
angle_boost
¶float | None
, default:None
) –Multiplier to increase the debanding strength on consistent gradients.
-
max_angle
¶float | None
, default:None
) –Threshold for the gradient angle check.
Returns:
-
Any
–The configured enum.
Source code in vsdeband/debanders.py
240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 |
|
__call__ ¶
__call__(*args: args, **kwargs: kwargs) -> R
Source code in vsdeband/debanders.py
46 47 |
|
f3k_deband ¶
f3k_deband(
clip: VideoNode,
radius: int = 16,
thr: int | Sequence[int] = 96,
grain: float | Sequence[float] = 0.0,
planes: PlanesT = None,
*,
sample_mode: SampleMode = SQUARE,
dynamic_grain: bool = False,
blur_first: bool = True,
seed: int | None = None,
random: RandomAlgo | tuple[RandomAlgo, RandomAlgo] = UNIFORM,
**kwargs: Any
) -> VideoNode
Debanding filter wrapper using the neo_f3kdb
plugin.
More informations can be found in the plugin documentation: https://github.com/HomeOfAviSynthPlusEvolution/neo_f3kdb
Parameters:
-
clip
¶VideoNode
) –Input clip.
-
radius
¶int
, default:16
) –Radius used for banding detection. Valid range is [1, 255].
-
thr
¶int | Sequence[int]
, default:96
) –Banding detection threshold(s) for each plane (Y, Cb, Cr). A pixel is considered banded if the difference with its reference pixel(s) is less than the corresponding threshold.
-
grain
¶float | Sequence[float]
, default:0.0
) –Amount of grain to add after debanding. Accepts a float or sequence of floats.
-
planes
¶PlanesT
, default:None
) –Specifies which planes to process. Default is all planes.
-
sample_mode
¶SampleMode
, default:SQUARE
) –Strategy used to sample reference pixels. See SampleMode.
-
dynamic_grain
¶bool
, default:False
) –If True, generates a unique grain pattern for each frame.
-
blur_first
¶bool
, default:True
) –If True, compares current pixel to the mean of surrounding pixels. If False, compares directly to all reference pixels. A pixel is marked as banded only if all pixel-wise differences are below threshold.
-
seed
¶int | None
, default:None
) –Random seed for grain generation.
-
random
¶RandomAlgo | tuple[RandomAlgo, RandomAlgo]
, default:UNIFORM
) –Random number generation strategy. Can be a single value for both reference and grain, or a tuple specifying separate algorithms. See RandomAlgo.
-
**kwargs
¶Any
, default:{}
) –Additional keyword arguments passed directly to the
neo_f3kdb.Deband
plugin.
Raises:
-
CustomValueError
–If inconsistent grain parameters are provided across chroma planes.
-
InvalidColorFamilyError
–If input clip is not YUV or GRAY.
Returns:
-
VideoNode
–Debanded and optionally grained clip.
Source code in vsdeband/debanders.py
364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 |
|
mdb_bilateral ¶
mdb_bilateral(
clip: VideoNode,
radius: int = 16,
thr: float = 260,
debander: _DebanderFunc[Any] = f3k_deband,
dark_thr: float | Sequence[float] = 0.6,
bright_thr: float | Sequence[float] = 0.6,
elast: float | Sequence[float] = 3.0,
planes: PlanesT = None,
) -> VideoNode
Multi stage debanding, bilateral-esque filter.
This function is more of a last resort for extreme banding.
Example usage:
from vsdeband import mdb_bilateral, f3k_deband
from functools import partial
debanded = mdb_bilateral(clip, 17, 320)
Parameters:
-
clip
¶VideoNode
) –Input clip.
-
radius
¶int
, default:16
) –Banding detection range.
-
thr
¶float
, default:260
) –Banding detection thr(s) for planes.
-
debander
¶_DebanderFunc[Any]
, default:f3k_deband
) –Specifies what debander callable to use.
-
dark_thr
¶float | Sequence[float]
, default:0.6
) –limit_filter parameter. Threshold (8-bit scale) to limit dark filtering diff.
-
bright_thr
¶float | Sequence[float]
, default:0.6
) –limit_filter parameter. Threshold (8-bit scale) to limit bright filtering diff.
-
elast
¶float | Sequence[float]
, default:3.0
) –limit_filter parameter. Elasticity of the soft threshold.
-
planes
¶PlanesT
, default:None
) –Which planes to process.
Returns:
-
VideoNode
–Debanded clip.
Source code in vsdeband/debanders.py
544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 |
|
pfdeband ¶
pfdeband(
clip: VideoNode,
radius: int = 16,
thr: float | Sequence[float] = 96,
prefilter: PrefilterLike | _SupportPlanesParam = gauss_blur,
debander: _DebanderFunc[Any] = f3k_deband,
ref: VideoNode | None = None,
dark_thr: float | Sequence[float] = 0.3,
bright_thr: float | Sequence[float] = 0.3,
elast: float | Sequence[float] = 2.5,
planes: PlanesT = None,
) -> VideoNode
Prefilter and deband a clip.
Parameters:
-
clip
¶VideoNode
) –Input clip.
-
radius
¶int
, default:16
) –Banding detection range.
-
thr
¶float | Sequence[float]
, default:96
) –Banding detection thr(s) for planes.
-
prefilter
¶PrefilterLike | _SupportPlanesParam
, default:gauss_blur
) –Prefilter used to blur the clip before debanding.
-
debander
¶_DebanderFunc[Any]
, default:f3k_deband
) –Specifies what debander callable to use.
-
planes
¶PlanesT
, default:None
) –Planes to process
-
ref
¶VideoNode | None
, default:None
) –limit_filter parameter. Reference clip, to compute the weight to be applied on filtering diff.
-
dark_thr
¶float | Sequence[float]
, default:0.3
) –limit_filter parameter. Threshold (8-bit scale) to limit dark filtering diff.
-
bright_thr
¶float | Sequence[float]
, default:0.3
) –limit_filter parameter. Threshold (8-bit scale) to limit bright filtering diff.
-
elast
¶float | Sequence[float]
, default:2.5
) –limit_filter parameter. Elasticity of the soft threshold.
Returns:
-
VideoNode
–Debanded clip.
Source code in vsdeband/debanders.py
606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 |
|
placebo_deband ¶
placebo_deband(
clip: VideoNode,
radius: float = 16.0,
thr: float | Sequence[float] = 3.0,
grain: float | Sequence[float] = 0.0,
planes: PlanesT = None,
*,
iterations: int = 4,
**kwargs: Any
) -> VideoNode
Debanding wrapper around the placebo.Deband
filter from the VapourSynth vs-placebo
plugin.
For full plugin documentation, see: https://github.com/sgt0/vs-placebo?tab=readme-ov-file#deband
Parameters:
-
clip
¶VideoNode
) –Input clip.
-
radius
¶float
, default:16.0
) –Initial debanding radius. The radius increases linearly with each iteration. A higher radius will find more gradients, but a lower radius will smooth more aggressively.
-
thr
¶float | Sequence[float]
, default:3.0
) –Cut-off threshold(s) for each plane. Higher values increase debanding strength but may remove fine details. Accepts a single float or a sequence per plane.
-
grain
¶float | Sequence[float]
, default:0.0
) –Amount of grain/noise to add after debanding. Helps mask residual artifacts. Accepts a float or a sequence per plane. Note: For HDR content, grain can significantly affect brightness. Consider reducing or disabling.
-
planes
¶PlanesT
, default:None
) –Which planes to process. Defaults to all planes.
-
iterations
¶int
, default:4
) –Number of debanding steps to perform per sample. More iterations yield stronger effect but quickly lose efficiency beyond 4.
Returns:
-
VideoNode
–Debanded and optionally grained clip.
Source code in vsdeband/debanders.py
451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 |
|