ColorBars / ColorBarsHD¶
The ColorBars. ColorBarsHD and `ColorBarsUHD`_ filters generate a video clip containing SMPTE color bars scaled to any image size.
You can find `ColorBarsUHD`_ in a separate document, here we describe the first two version. Both filters also generate audio, see the audio section for details.
ColorBars¶
ColorBars produces a video clip containing SMPTE color bars (Rec. ITU-R BT.801-1) scaled to any image size. By default, a 640x480, RGB32, TV range, 29.97 fps, 1 hour long clip is produced.
The color values are computed from ground-truth linear RGB primaries using the
BT.601 matrix (AVS_MATRIX_ST170_M). This applies to all YUV output formats.
RGB output formats use the same primaries directly in studio-swing encoding.
Syntax and Parameters
ColorBars (int "width", int "height", string "pixel_type", bool "staticframes")
- width, height
Set size of the returned clip.
Default: 640, 480
- pixel_type
Set color format of the returned clip. May be any of the following: "RGB24", "RGB32", "RGB48", "RGB64", "YUY2", "YV12", "YV16" "YV24", "YV411", or any planar RGBPx, RGBAPx, YUV4xxPx, YUVA4xxPx format.
Default: "RGB32"
- staticframes
If set to false, generate all frames. Default true (one static frame is served).
Default: true
ColorBarsHD¶
ColorBarsHD produces a video clip containing SMPTE color bars (Rec. ITU-R BT.709 / ARIB STD-B28 v1.0) scaled to any image size. By default, a 1288x720, YV24, TV range, 29.97 fps, 1 hour long clip is produced.
The color values are computed from ground-truth linear RGB primaries using the
BT.709 matrix (AVS_MATRIX_BT709). Output is always a YUV 4:4:4 format.
Syntax and Parameters
ColorBarsHD (int "width", int "height", string "pixel_type", bool "staticframes")
- width, height
Set size of the returned clip.
Default: 1288, 720
- pixel_type
Set color format of the returned clip. Must be "YV24" or any YUV444Px / YUVA444Px format.
Default: "YV24" (identical to "YUV444P8")
- staticframes
If set to false, generate all frames. Default true (one static frame is served).
Default: true
Audio¶
For both filters, an audio tone is also generated. The tone is a 440Hz sine at 48KHz sample rate, 32 bit (Float), stereo. The tone pulses in the right speaker, being turned on and off once every second. Level is 0 dBFS.
You can use Amplify to set a softer level (0dB can be a little deafening!)
ColorBarsHD
AmplifyDB(-20)
Broadcasting organizations usually specify an "alignment tone" accompanying colorbars at anywhere from -12 to -20 dBFS; if sending materials to another party, be sure to get their preferred alignment tone level. The exact level doesn't matter as long as all parties agree to it.
A note on notation¶
This page adopts the ITU style when discussing video levels which might be represented at different bit depths:
" To avoid confusion between 8-bit and 10-bit representations, the eight most-significant bits are considered to be an integer part while the two additional bits, if present, are considered to be fractional part. For example, the bit pattern
10010001would be expressed as 145d, whereas the pattern1001000101would be expressed as 145.25d. "
Video levels shown below with the subscript "d" are assumed to be scaled by 2^(bit depth-8). For example, 235d at bit depth 10 becomes 235 × 2^(10-8) = 235 × 4 = 940.
see Deep Color
TV range¶
For both filters, in all color formats, luminance levels are TV (limited) range, where black=16d and white=235d, within a total possible range of 0-255d.
The table below shows the TV-range values ColorBarsHD generates, and those same values as they should be after converting to full range.
\(\mathtt{Y_\text{full} = (Y_\text{tv}-16_\text{d}) × 255_\text{d}/(235_\text{d}-16_\text{d})}\) // (for R, G, B, Y)
\(\mathtt{U_\text{full} = (U_\text{tv}-128_\text{d}) × 255_\text{d}/(240_\text{d}-16_\text{d}) + 128_\text{d}}\) // (for U, V)
\(\mathtt{Y_\text{tv} = Y_\text{full} × (235_\text{d}-16_\text{d})/255_\text{d} + 16_\text{d}} \quad\) // (for R, G, B, Y)
\(\mathtt{U_\text{tv} = (U_\text{full}-128_\text{d}) × (240_\text{d}-16_\text{d})/255_\text{d} + 128_\text{d}}\) // (for U, V)
The table below shows the 8-bit TV-range values ColorBarsHD generates, and those same values as they should be after converting to full range. At higher bit depths, values are computed directly from linear RGB primaries via the BT.709 matrix and are not simple left-shifts of these 8-bit codes.
Color bar |
TV range output (8-bit) |
(expanded to full range, 8-bit) |
|||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
R |
G |
B |
Y |
U |
V |
R |
G |
B |
Y |
U |
V |
||||
75% White |
180 |
180 |
180 |
180 |
128 |
128 |
191 |
191 |
191 |
191 |
128 |
128 |
|||
75% Yellow |
180 |
180 |
16 |
168 |
44 |
136 |
191 |
191 |
0 |
177 |
32 |
137 |
|||
75% Cyan |
16 |
180 |
180 |
145 |
147 |
44 |
0 |
191 |
191 |
150 |
149 |
32 |
|||
75% Green |
16 |
180 |
16 |
133 |
63 |
52 |
0 |
191 |
0 |
136 |
54 |
41 |
|||
75% Magenta |
180 |
16 |
180 |
63 |
193 |
204 |
191 |
0 |
191 |
55 |
201 |
214 |
|||
75% Red |
180 |
16 |
16 |
51 |
109 |
212 |
191 |
0 |
0 |
41 |
106 |
223 |
|||
75% Blue |
16 |
16 |
180 |
28 |
212 |
120 |
0 |
0 |
191 |
14 |
223 |
118 |
|||
These colors are at "75%" of maximum, per common broadcast practice. You may occasionally see "100%" color bars.[1] They are rather useless, as you cannot detect gain or saturation that is too high on a signal that is already at maximum.
PLUGE¶
The lower part of the frame is called the PLUGE (also lowercase: "pluge") signal. From left to right it consists of: -I, white, +Q, then a series of black and near-black bars: 0, -4, 0, +4 and 0 IRE relative to black.
Note 'IRE' is used here to mean 'percent luminance', on a scale from 0 (black) to 100 (white), ignoring the varying broadcast standards where black might be 0 IRE or 7.5 IRE depending on the country.
This section documents the ColorBars pluge only; ColorBarsHD's pluge is similar, but dispenses with -I and +Q.
The table below shows the 8-bit TV-range values ColorBars generates, and those same values as they should be after converting to full range. At higher bit depths, values are computed directly from linear RGB primaries via the BT.601 (SMPTE 170M) matrix and are not simple left-shifts of these 8-bit codes.
PLUGE Element |
TV range output (8-bit) |
(expanded to full range, 8-bit) |
|||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
R |
G |
B |
Y |
U |
V |
R |
G |
B |
Y |
U |
V |
||||
-I |
16 |
90 |
130 |
16 |
158 |
95 |
0 |
86 |
130 |
0 |
162 |
90 |
|||
100% White |
235 |
235 |
235 |
235 |
128 |
128 |
255 |
255 |
255 |
255 |
128 |
128 |
|||
+Q |
92 |
16 |
143 |
16 |
174 |
149 |
88 |
0 |
148 |
0 |
180 |
151 |
|||
Black |
16 |
16 |
16 |
16 |
128 |
128 |
0 |
0 |
0 |
0 |
128 |
128 |
|||
-4 IRE |
7 |
7 |
7 |
7 |
128 |
128 |
-10 |
-10 |
-10 |
-10 |
128 |
128 |
|||
Black |
16 |
16 |
16 |
16 |
128 |
128 |
0 |
0 |
0 |
0 |
128 |
128 |
|||
+4 IRE |
25 |
25 |
25 |
25 |
128 |
128 |
10 |
10 |
10 |
10 |
128 |
128 |
|||
Black |
16 |
16 |
16 |
16 |
128 |
128 |
0 |
0 |
0 |
0 |
128 |
128 |
|||
(negative values will be clipped to 0) |
|||||||||||||||
Important note on -I and +Q values: The RGB and YUV values shown above do not convert to each other via standard BT.601 matrix conversion. This is intentional and reflects the dual-specification nature of these legacy NTSC test signals:
RGB output uses luma-corrected values where the most negative component is lifted to studio black (code 16), ensuring all RGB codes are broadcast-safe (≥16).
YUV output uses zero-luma pure chroma-axis values (Y=16) that preserve the theoretical definition but would require super-black RGB components if converted back.
The -I and +Q bars are vestigial artifacts of NTSC analog TV and are not really used any more.
The -4, 0 and +4 IRE bars can be used to set your monitor brightness – assuming your playback chain expands TV range (16-235) to full-range (0-255) as shown in the images above. The -4 IRE and 0 IRE bars should have the same apparent brightness (they should be as dark as the monitor can display), and the +4 should be a little brighter. If you can see the -4 bar, your monitor brightness is set too high; if you cannot see the +4 bar, your monitor brightness is set too low.[2]
Note that the PLUGE signal intentionally includes super-black values in the -4 IRE bar (code 7, below studio black at code 16). These test the full capability of the signal chain and cannot be accurately represented when converting to full-range 0-255.
More information about the colorbars and the PLUGE can be found on the color bars theory page.
Miscellaneous¶
Note, that for example
ColorBars(pixel_type="YUV444P8")
...is equivalent to
ColorBars(pixel_type="RGB32")
ConvertToYUV444(matrix="PC.601")
# "PC.601" / "PC.709" / "PC.2020" don't scale the luma range
When directly generating YUV format data, the color transitions are arranged to occur on a chroma-aligned boundary.
Advanced: color value derivation¶
This section documents the precise mathematical derivation of the color values used by ColorBars and ColorBarsHD, including the handling of the -I and +Q PLUGE components.
ColorBarsHD derives all color values from ground-truth linear RGB primaries using double-precision arithmetic and the BT.709 RGB-to-YUV matrix:
Y = 0.2126·R + 0.7152·G + 0.0722·B
Cb = (B − Y) / 1.8556
Cr = (R − Y) / 1.5748
ColorBars uses the same approach with the BT.601 (SMPTE 170M) matrix:
Y = 0.299·R + 0.587·G + 0.114·B
Cb = (B − Y) / 1.772
Cr = (R − Y) / 1.402
For integer output at any bit depth n, the encoding is:
Y_encoded = (Y × 219 + 16 ) × 2^(n−8) (luma, limited range)
Cb_encoded = (Cb × 224 + 128) × 2^(n−8) (chroma, limited range)
Cr_encoded = (Cr × 224 + 128) × 2^(n−8) (chroma, limited range)
For 32-bit float output, the AviSynth limited-range float convention is used, derived from the same centralized conversion constants as all other AviSynth filters:
Y_float = Y × (219/255) + (16/255)
Cb_float = Cb × (224/255)
Cr_float = Cr × (224/255)
All RGB output formats use studio-swing encoding with the same ground-truth
primaries. The helper function studio_rgb_to_integer encodes a normalized
double value v as:
code = (int)(v × 219 × 2^(n−8) + 16 × 2^(n−8) + 0.5)
where v = 0.0 maps to studio black (code 16) and v = 1.0 maps to studio white (code 235). Convention: normalized values represent limited-range positions, where 0.0 = studio black (not absolute black).
-I and +Q derivation
The -I and +Q signals are defined in the YIQ colour space as pure chroma-axis signals with zero luma and 20 IRE saturation (0.2162 normalized):
-I: I = −0.2162, Q = 0
+Q: I = 0, Q = +0.2162
Converting via the BT.601 UV rotation (Poynton eq. 33, with UV swap):
-I raw RGB (Y=0): R = −0.2067, G = +0.0588, B = +0.2394
+Q raw RGB (Y=0): R = +0.1343, G = −0.1400, B = +0.3685
Both signals contain out-of-range (negative) RGB components. Three interpretations exist in the literature:
Option 1 — Zero-luma, studio black hack (legacy AviSynth): Y = 16 (studio black). This is a HACK applied differently for RGB vs YUV:
YUV output: Uses raw zero-luma values (Y=16, Cb=158, Cr=95 for -I)
RGB output: Individually adjusted/clipped components to avoid super-blacks
Result: Two incompatible specifications that don't convert via standard matrices.
Option 2 — Luma-corrected to absolute black (valid alternative, not used): Luma raised until most negative component reaches code 0 (absolute black). Gives colorimetric consistency but uses super-black range (codes 0-15).
Option 3 — Luma-corrected to studio black (current RGB implementation): Luma raised until most negative component reaches code 16 (studio black).
For -I: Y_lift = 0.2067 - 16/219 = 0.13364 For +Q: Y_lift = 0.1400 - 16/219 = 0.06694
After lifting (RGB output):
-I: R = 16 (studio black), G = 90, B = 130, Y ≈ 77 +Q: R = 92, G = 16 (studio black), B = 143, Y ≈ 63
Dual specification implementation
The current implementation maintains two separate ground-truth tables:
RGB-native values (Option 3): Used for RGB output formats
All components >= 16 (broadcast-safe)
Colorimetrically consistent RGB↔YUV conversion
-I: RGB(16, 90, 130) → Y≈77 after BT.601 conversion
+Q: RGB(92, 16, 143) → Y≈63 after BT.601 conversion
YUV-targeted values (Option 1): Used for YUV output formats
Produces exact legacy values: -I Y=16, +Q Y=16 (zero-luma)
Contains out-of-range RGB components if converted back
Preserves theoretical chroma-axis purity
This dual-table approach acknowledges historical reality: the original AviSynth ColorBars had two independent specifications that don't convert to each other. The -I and +Q signals were analog broadcast test signals (voltage levels), not digital RGB/YUV values, and their digital representation requires compromises.
Higher bit-depth output is computed directly from the double-precision ground-truth values via the full encoding formula — not by bit-shifting 8-bit codes — giving greater accuracy at 10, 12, 14, 16-bit and float.
Changelog¶
Version |
Changes |
|---|---|
AviSynth+ 3.7.6 |
ColorBarsHD: fixed high bit-depth output (values now
computed from linear RGB primaries via BT.709 matrix
instead of upscaling 8-bit table entries).
ColorBarsHD: fixed ramp in pattern 3
ColorBars: fixed high bit-depth output for all YUV
and RGB formats (same ground-truth RGB approach).
ColorBars: fixed reported frame property
_matrix
to AVS_MATRIX_ST170_M (BT.601); was incorrectly
reporting AVS_MATRIX_BT709.ColorBars: -I and +Q values kept from legacy, but
the explanation is included in this documentation.
Fix: "staticframes"=false parameter copied U instead
of A for alpha plane.
|
AviSynth+ 3.4.0 |
ColorBars: add support for all YUV(A)422 formats and
RGB24, RGB48, YV411.
|
AviSynth+ r2487 |
ColorBars: add support for all YUV(A)444/420, planar
RGB(A) formats and RGB64.
ColorBarsHD: add support for all YUV(A)444 formats.
|
AviSynth 2.6.0 |
Added pixel_type="YV24" to ColorBars.
Initial release of ColorBarsHD.
|
AviSynth 2.5.6 |
Added
pixel_type parameter.Added "YUY2" and "YV12" pixel types.
|
AviSynth 2.5.5 |
Width and height parameters are now named and optional. |
$Date: 2026/02/19 09:57:00 $
