X264 command line parameter explanation

This document corresponds to the x264 command line mode, and the VFW mode also uses the same parameters, but it is just a graphical interface, you can find the corresponding English.

Format: x264 Default Option-O output file input file [length x Width]
Supported input formats: raw, y4m, Avi, and AVS (optional during compilation)
Supported output formats: 264/MKV/MP4 (optional during compilation)

Many parameters of x264 can have-/-two input methods. I don't know why. The following equivalent parameters are represented by "parameter 1/parameter 2 <required value format>", and are recommended at the end of the parameter.

-H/-- Help help

Frame Type options:

-I/-- keyint <integer> maximum IDR frame spacing, 250 by default
-I/-- Min-keyint <integer> Minimum IDR frame spacing. The default value is 25.
-- Scenecut <integer> the threshold for Dynamic screen changes. When the value is exceeded, the I frame is inserted. The default value is 40.
-B/-- bframes <integer> maximum number of B frames that can be inserted between IP frames, ranging from 0 ~ 16. The default value is 0.
-- No-B-adapt disables adaptive B-frame determination (available when-B is set to 1, others are not recommended)
-- B-bias <integer> controls the determination of inserting B frames. The value range is-100 ~ + 100. The higher the value, the easier it is to insert B frames. The default value is 0.
-- B-pyramid allows B frames as reference frames
-- No-cabac disable content adaptive binary arithmetic coding (cabac, efficient entropy encoding) (which increases the speed but seriously affects the quality)
-R/-- ref <integer> maximum number of reference frames, ranging from 0 ~ 16. The default value is 1.
-- NF Closed Loop filtering (a mosaic Removal Algorithm)
-F/-- filter <ALPHA: Beta>: Set the alphac and beta parameters of loop filter. The value range is-6-6. The default value is 0.

Bit Rate Control Options:

-Q/-- QP <integer> fixed quantization mode and set the used quantization value. The value range is 0 ~ . 0 indicates lossless compression. The default value is 26.
-B/-- bitrate <integer> sets the average bit rate.
-- CRF <integer> quality mode. The quantitative value is dynamically variable (currently immature, it is better to set a fixed quantitative value for quality)
-- Qpmin <integer> sets the minimum quantization value, ranging from 0 ~ 51. The default value is 10.
-- Qpmax <integer> sets the maximum quantization value, ranging from 0 ~ 51. The default value is 51.
-- Qpstep <integer> sets the quantization value difference between adjacent frames, ranging from 0 ~ 50. The default value is 4.
-- Ratetol <decimal> the deviation ratio of the instantaneous bit rate in the average bit rate mode. The value range is 0.1 ~ 100.0. The default value is 1.0.
-- Vbv-maxrate <integer> maximum instantaneous bit rate in average bit rate mode. The default value is 0 (same as-B)
-- Vbv-bufsize <integer> bit rate controls the buffer size. Unit: kbit. The default value is 0.
-- Vbv-init <decimal> bit rate controls the ratio of the maximum data retained in the buffer zone to the buffer zone size, ranging from 0 ~ 1.0. The default value is 0.9.

-- Ipratio <decimal> quantization coefficient between frames I and P. The default value is 1.40.
-- Pbratio <decimal> quantization coefficient between P frames and B frames. The default value is 1.30.
-- Color-QP-offset <integer> quantization difference between the color and the brightness, range:-12 ~ + 12. The default value is 0.

-P/-- pass <1 | 2 | 3> multiple compression Bit Rate Control
1: Create a statistical file for the first Compression
2: compress and output the data according to the created Statistical file, without overwriting the Statistical File,
3: compress the created Statistical File to optimize the Statistical File
-- Stats <string> The Name Of The Statistics file. The default value is "x1__2pass.log"
-- Rceq <string> rate control formula. The default value is "blurcplx ^ (1-qcomp )"
-- Qcomp <decimal> linear quantization control. The value 0.0 indicates a fixed bit rate and the value 1.0 indicates a fixed quantization value. The default value is 0.6. It is only used for 2-pass and quality modes.
-- Cplxblur <decimal number> the maximum value based on the ratio of the smooth quantization value of adjacent frames. The value range is 0 ~ 99.9. The default value is 20.0. It is only used for 2-pass and quality modes.
-- Qblur <decimal> the maximum ratio of the smooth quantization value of the Statistical File results, ranging from 0 ~ 99.9. The default value is 0.5. It is only used for 2-pass.

-- Zones <z0>/<Z1> /... Segment quantization, in the format of <start frame>, <End frame>, <option>, and optional values: q = <integer> (quantization value) or B = <decimal number> (bitrate multiple)

Analysis options:

-A/-- analyze <string> Dynamic Block division method. The default value is "p8x8, b8x8, i8x8, i4x4 ". Optional: p8x8/p4x4/b8x8/i8x8/i4x4; none/All (p4x4 requires p8x8. i8x8 -- 8x8dct)
-- Direct <string> dynamic prediction method. The default value is "spatial ". Optional: NONE/spatial/temporal/auto
-W/-- weightb allows B-frame weighted prediction (which can reduce the impact of low quality of adjacent B-frames)
-- Me <string> specifies the dynamic prediction search method for all pixel blocks. The default value is "hex". Optional:
Dia: Diamond Search, radius 1 (FAST)
HEX: hexagonal search, radius 2
Umh: hexagonal search with variable radius
ESA: comprehensive search (very slow and almost identical to umh)
-- Merange <integer> -- the search radius when me is umh/ESA. The maximum value is 64. The default value is 16.
-M/-- subme <integer>: Dynamic Prediction and partitioning method. Option 1 ~ 7. The default value is 5 (closely related to the compression quality and time, and 1 is more than four times the speed of 7)
1: use full pixel blocks for dynamic search, and then use the quick mode to perform a precise search of 1/4 pixel blocks.
2: Use half-pixel blocks for dynamic search, and then use the quick mode to perform a precise search of 1/4 pixel blocks.
3: Use half-pixel blocks for dynamic search, and use the quality mode to perform a precise search of 1/4 pixel blocks for each block.
4: exact search of 1/4 pixel blocks in Quick Mode
5: precise search of 1/4 pixels in quality mode
6: optimizing the rate distortion of pixel blocks of I and P frames (rdo)
7: optimizing I and P frame motion vectors and intra-block rate distortion (best quality)
-- B-rdo B frame also performs rdo. The value of -- subme is more than 6.
-- Mixed-Refs can use different reference frames within one frame
-- No-chroma-me: do not perform dynamic color Prediction
-- Bime can average the motion vector of the Reference Block of B frames.
-8/-- 8x8 DCT can use 8x8 discrete cosine transform (DCT)
-T/-- trellis <integer> trellis quantization. Find the appropriate quantization value for each 8x8 block. cabac is required. The default value is 0.
0: Disable
1: used only for the final Encoding
2: Always Use
-- No-fast-pskip disable fast p-frame skip Detection
-- No-DCT-decimate disable p-Frame Joint encoding (details can be added, but the size will also increase)
-- Nr <integer> Noise Removal, range: 0 ~ 100000, 0 by default

-- CQM <string> sets the format of the external quantization matrix. The default value is "flat". Optional values: JVT/flat.
-- Cqmfile <string> reads the external quantization matrix file in JM format and automatically ignores other -- CQM * options.
-- Cqm4 <list> sets the 4x4 quantization matrix separated by commas (,). The value range is 1 ~ 16 Integers of 255
-- Cqm8 <list> sets the 8x8 quantization matrix separated by commas (,). The value range is 1 ~ 64 integers in 255
-- Cqm4i/-- cqm4p/-- cqm8i/-- cqm8p sets different quantization matrices for I and P frames.
-- Cqm4iy/-- cqm4ic/-- cqm4py/-- cqm4pc sets the quantization matrix with Different Brightness and color.

Video standardization options:
These options have nothing to do with encoding, but if you want to use a player like MP4, you can set them at your own risk.

-- SAR width: Height: Set the aspect ratio.
-- Overscan <string> overscan line. The default value is "UNDEF" (not set). The options are show (View)/crop (delete)
-- Videoformat <string> video format. The default value is "UNDEF". Optional values: component/PAL/NTSC/SECAM/MAC/UNDEF.
-- Fullrange <string> specify full range samples setting, default "off", option: Off/on)
-- Colorprim <string> original color format. The default value is "UNDEF". Optional values: UNDEF, bt709, bt470m, bt470bg, smpte170m, smpte240m, and film.
-- Transfer <string> conversion method. The default value is "UNDEF". Options include UNDEF, bt709, bt470m, bt470bg, linear, log100, log316, smpte170m, and smpte240m.
-- Colormatrix <string> color matrix setting. The default value is "UNDEF", UNDEF/bt709/FCC/bt470bg, smpte170m/smpte240m/GBR/ycgco.
-- Chromaloc <integer> color sample specified, range: 0 ~ 5. The default value is 0.

Input and Output options:

-- Level <string> set the level (as defined by Annex A) (do not understand, please advise)
-- FPS <decimal> set Frame Rate
-- Seek <integer> specifies the start frame.
-- Frames <integer> maximum number of encoded Frames
-O/-- output specifies the output file

-- Threads <integer> encoding thread (using sharding Technology)
-- Thread-input: Run avisynth In the encoding thread
-- No-ASM disable all CPU optimization commands
-- No-SNR disable SNR Calculation
-- Quiet Mode
-V/-- verbose displays the information of each frame.
-- Progress: display the Encoding Process
-- Visualize: Display motion vectors
-- SPS-id <integer> specifies the ID of SPs and PPS. The default value is 0.
-- Aud uses data units to define symbols