Man / CODEC固有オプション / -lavcopts

Man / CODEC固有オプション / -lavcopts

• Prev
• Next
• Man/CODEC固有オプション

---

acodec=<value> †

audio codec (default: mp2)

abitrate=<value> †

audio bitrate in kbps (default: 224)

atag=<value> †

Use the specified Windows audio format tag (e.g. atag=0x55).

bit_exact †

Use only bit exact algorithms (except (I)DCT). Additionally bit_exact disables several optimizations and thus should only be used for regression tests, which need binary identical files even if the encoder version changes. This also suppresses the user_data header in MPEG-4 streams. Do not use this option unless you know exactly what you are doing.

(I)DCTを除いたロスレスな変換のみを使わせます。さらにこのオプションはいくつかの最適化を無効にするため、エンコーダーのバージョンが変わったときでもバイナリとして同一なファイルを必要とする退行テストのときにのみ使用されるべきです。これはMPEG-4ストリームの中のuser_dataヘッダーも抑制します。何をするのかわからないならこのオプションは使わないでください。

threads=<1−8> †

Maximum number of threads to use (default: 1). May have a slight negative effect on motion estimation.

vcodec=<value> †

Employ the specified codec (default: mpeg4).

vqmin=<1−31> †

minimum quantizer (pass 1/2)

1

Not recommended (much larger file, little quality difference and weird side effects: msmpeg4, h263 will be very low quality, ratecontrol will be confused resulting in lower quality and some decoders will not be able to decode it).
推奨されない(とても大きなファイルになり、ほとんど画質の違いがないうえに奇妙な副作用（msmpeg4, h263でとても画質が低くなり、ratecontrolの結果に困惑されさらにあるデコーダではそれはデコード出来ない。）をもたらす。)

2

Recommended for normal mpeg4/mpeg1video encoding (default).
通常のmpeg4/mpeg1videoでのエンコーディングで推奨

3

Recommended for h263(p)/msmpeg4. The reason for preferring 3 over 2 is that 2 could lead to overflows. (This will be fixed for h263(p) by changing the quantizer per MB in the future, msmpeg4 cannot be fixed as it does not support that.)
h263(p)/msmpeg4でのエンコーディングにおいて推奨。理由は2ではオーバーフローが起きることがあるからである。(これはh263(p)においては将来、MB間での量子化手法の変更で修正される見込み。msmpeg4においてはサポートされてないため修正できない。)

lmin=<0.01−255.0> †

Minimum Lagrange multiplier for ratecontrol, you probably want it to be equal to or lower than vqmin (default: 2.0).

lmax=<0.01−255.0> †

maximum Lagrange multiplier for ratecontrol (default: 31.0)

vqscale=<0−31> †

Constant quantizer / constant quality encoding (selects fixed quantizer mode). A lower value means better quality but larger files (default: 0). In case of snow codec, value 0 means lossless encoding. Since the other codecs do not support this, vqscale=0 will be ignored and variable quantizers are used, which is the default. 1 is not recommended (see vqmin for details).

vqmax=<1−31> †

Maximum quantizer (pass 1/2), 10−31 should be a sane range (default: 31).

mbqmin=<1−31> †

obsolete, use vqmin

mbqmax=<1−31> †

obsolete, use vqmax

vqdiff=<1−31> †

maximum quantizer difference between consecutive I- or P-frames (pass 1/2) (default: 3)

vmax_b_frames=<0−4> †

maximum number of B-frames between non-B-frames:

0

no B-frames (default)

0−2

sane range for MPEG-4

vme=<0−5> †

motion estimation method. Available methods are:
動き予測

0

none (very low quality)
しない

1

full (slow, currently unmaintained and disabled)
総当り

2

log (low quality, currently unmaintained and disabled)

3

phods (low quality, currently unmaintained and disabled)

4

EPZS: size=1 diamond, size can be adjusted with the *dia options (default)
ダイヤモンド型の探索

5

X1 (experimental, currently aliased to EPZS)

8

iter (iterative overlapped block, only used in snow)

NOTE: 0−3 currently ignores the amount of bits spent, so quality may be low.

me_range=<0−9999> †

motion estimation search range (default: 0 (unlimited))

mbd=<0−2> †

Macroblock decision algorithm (high quality mode), encode each macro block in all modes and choose the best. This is slow but results in better quality and file size.

0

Use mbcmp (default).

1

Select the MB mode which needs the fewest bits (=vhq).

2

Select the MB mode which has the best rate distortion.

vhq †

Same as mbd=1, kept for compatibility reasons.

v4mv †

Allow 4 motion vectors per macroblock (slightly better quality). Works better if used with mbd>0.

obmc †

overlapped block motion compensation (H.263+)

loop †

loop filter (H.263+) note, this is broken

inter_threshold <-1000−1000> †

Does absolutely nothing at the moment.

keyint=<0−300> †

maximum interval between keyframes in frames (default: 250 or one keyframe every ten seconds in a 25fps movie. This is the recommended default for MPEG-4). Most codecs require regular keyframes in order to limit the accumulation of mismatch error. Keyframes are also needed for seeking, as seeking is only possible to a keyframe - but keyframes need more space than other frames, so larger numbers here mean slightly smaller files but less precise seeking. 0 is equivalent to 1, which makes every frame a keyframe. Values >300 are not recommended as the quality might be bad depending upon decoder, encoder and luck. It is a common for MPEG-1/2 to use values <=30.

sc_threshold=<-1000000000−1000000000> †

Threshold for scene change detection. A keyframe is inserted by libavcodec when it detects a scene change. You can specify the sensitivity of the detection with this option. -1000000000 means there is a scene change detected at every frame, 1000000000 means no scene changes are detected (default: 0).

sc_factor=<any positive integer> †

Causes frames with higher quantizers to be more likely to trigger a scene change detection and make libavcodec use an I-frame (default: 1). 1−16 is a sane range. Values between 2 and 6 may yield increasing PSNR (up to approximately 0.04 dB) and better placement of I-frames in high-motion scenes. Higher values than 6 may give very slightly better PSNR (approximately 0.01 dB more than sc_factor=6), but noticably worse visual quality.

vb_strategy=<0−2> (pass one only) †

strategy to choose between I/P/B-frames:

0

Always use the maximum number of B-frames (default).

1

Avoid B-frames in high motion scenes. See the b_sensitivity option to tune this strategy.

2

Places B-frames more or less optimally to yield maximum quality (slower). You may want to reduce the speed impact of this option by tuning the option brd_scale.

b_sensitivity=<any integer greater than 0> †

Adjusts how sensitively vb_strategy=1 detects motion and avoids using B-frames (default: 40). Lower sensitivities will result in more B-frames. Using more B-frames usually improves PSNR, but too many B-frames can hurt quality in high-motion scenes. Unless there is an extremely high amount of motion, b_sensitivity can safely be lowered below the default; 10 is a reasonable value in most cases.

brd_scale=<0−10> †

Downscales frames for dynamic B-frame decision (default: 0). Each time brd_scale is increased by one, the frame dimensions are divided by two, which improves speed by a factor of four. Both dimensions of the fully downscaled frame must be even numbers, so brd_scale=1 requires the original dimensions to be multiples of four, brd_scale=2 requires multiples of eight, etc. In other words, the dimensions of the original frame must both be divisible by 2^(brd_scale+1) with no remainder.

bidir_refine=<0−4> †

Refine the two motion vectors used in bidirectional macroblocks, rather than re-using vectors from the forward and backward searches. This option has no effect without B-frames.

0

Disabled (default).

1−4

Use a wider search (larger values are slower).

vpass=<1−3> †

Activates internal two (or more) pass mode, only specify if you wish to use two (or more) pass encoding.

1

first pass (also see turbo)

2

second pass

3

Nth pass (second and subsequent passes of N-pass encoding)

Here is how it works, and how to use it: The first pass (vpass=1) writes the statistics file. You might want to deactivate some CPU-hungry options, like "turbo" mode does. In two pass mode, the second pass (vpass=2) reads the statistics file and bases ratecontrol decisions on it. In N-pass mode, the second pass (vpass=3, that is not a typo) does both: It first reads the statistics, then overwrites them. You might want to backup divx2pass.log before doing this if there is any possibility that you will have to cancel MEncoder. You can use all encoding options, except very CPU-hungry options like "qns". You can run this same pass over and over to refine the encode. Each subsequent pass will use the statistics from the previous pass to improve. The final pass can include any CPU-hungry encoding options. If you want a 2 pass encode, use first vpass=1, and then vpass=2. If you want a 3 or more pass encode, use vpass=1 for the first pass and then vpass=3 and then vpass=3 again and again until you are satisfied with the encode.

huffyuv:

pass 1

Saves statistics.

pass 2

Encodes with an optimal Huffman table based upon statistics from the first pass.

turbo (two pass only) †

Dramatically speeds up pass one using faster algorithms and disabling CPU-intensive options. This will probably reduce global PSNR a little bit (around 0.01dB) and change individual frame type and PSNR a little bit more (up to 0.03dB).

aspect=<x/y> †

Store movie aspect internally, just like with MPEG files. Much nicer than rescaling, because quality is not decreased. Only MPlayer will play these files correctly, other players will display them with wrong aspect. The aspect parameter can be given as a ratio or a floating point number.

EXAMPLE:

aspect=16/9 or aspect=1.78

autoaspect †

Same as the aspect option, but automatically computes aspect, taking into account all the adjustments (crop/expand/scale/ etc.) made in the filter chain. Does not incur a performance penalty, so you can safely leave it always on.

vbitrate=<value> †

Specify bitrate (pass 1/2) (default: 800). WARNING: 1kbit = 1000 bits

vratetol=<value> †

approximated file size tolerance in kbit. 1000−100000 is a sane range. (warning: 1kbit = 1000 bits) (default: 8000) NOTE: vratetol should not be too large during the second pass or there might be problems if vrc_(min|max)rate is used.

vrc_maxrate=<value> †

maximum bitrate in kbit/sec (pass 1/2) (default: 0, unlimited)

vrc_minrate=<value> †

minimum bitrate in kbit/sec (pass 1/2) (default: 0, unlimited)

vrc_buf_size=<value> †

buffer size in kbit (pass 1/2). For MPEG-1/2 this also sets the vbv buffer size, use 327 for VCD, 917 for SVCD and 1835 for DVD.

vrc_buf_aggressivity †

currently useless

vrc_strategy †

Ratecontrol method. Note that some of the ratecontrol-affecting options will have no effect if vrc_strategy is not set to 0.

0

Use internal lavc ratecontrol (default).

1

Use XviD ratecontrol (experimental; requires MEncoder to be compiled with support for XviD 1.1 or higher).

vb_qfactor=<-31.0−31.0> †

quantizer factor between B- and non-B-frames (pass 1/2) (default: 1.25)

vi_qfactor=<-31.0−31.0> †

quantizer factor between I- and non-I-frames (pass 1/2) (default: 0.8)

vb_qoffset=<-31.0−31.0> †

quantizer offset between B- and non-B-frames (pass 1/2) (default: 1.25)

vi_qoffset=<-31.0−31.0> †

(pass 1/2) (default: 0.0) if v{b|i}_qfactor > 0 I/B-frame quantizer = P-frame quantizer * v{b|i}_qfactor + v{b|i}_qoffset else do normal ratecontrol (do not lock to next P-frame quantizer) and set q= -q * v{b|i}_qfactor + v{b|i}_qoffset HINT: To do constant quantizer encoding with different quantizers for I/P- and B-frames you can use: lmin= <ip_quant>:lmax= <ip_quant>:vb_qfactor= <b_quant/ip_quant>.

vqblur=<0.0−1.0> (pass one) †

Quantizer blur (default: 0.5), larger values will average the quantizer more over time (slower change).

0.0

Quantizer blur disabled.

1.0

Average the quantizer over all previous frames.

vqblur=<0.0−99.0> (pass two) †

Quantizer gaussian blur (default: 0.5), larger values will average the quantizer more over time (slower change).

vqcomp=<0.0−1.0> †

Quantizer compression, vrc_eq depends upon this (pass 1/2) (default: 0.5). For instance, assuming the default rate control equation is used, if vqcomp=1.0, the ratecontrol allocates to each frame the number of bits needed to encode them all at the same QP. If vqcomp=0.0, the ratecontrol allocates the same number of bits to each frame, i.e. strict CBR. NOTE: Those are extreme settings and should never be used. Perceptual quality will be optimal somewhere in between these two extremes.

vrc_eq=<equation> †

main ratecontrol equation (pass 1/2)

1

constant bitrate

tex

constant quality

1+(tex/avgTex-1)*qComp

approximately the equation of the old ratecontrol code

tex^qComp

with qcomp 0.5 or something like that (default)

infix operators:

	+,-,*,/,^

variables:

tex

texture complexity

iTex,pTex

intra, non-intra texture complexity

avgTex

average texture complexity

avgIITex

average intra texture complexity in I-frames

avgPITex

average intra texture complexity in P-frames

avgPPTex

average non-intra texture complexity in P-frames

avgBPTex

average non-intra texture complexity in B-frames

mv

bits used for motion vectors

fCode

maximum length of motion vector in log2 scale

iCount

number of intra macroblocks / number of macroblocks

var

spatial complexity

mcVar

temporal complexity

qComp

qcomp from the command line

isI, isP, isB

Is 1 if picture type is I/P/B else 0.

Pi,E

See your favorite math book.

functions:

max(a,b),min(a,b)

maximum / minimum

gt(a,b)

is 1 if a>b, 0 otherwise

lt(a,b)

is 1 if a<b, 0 otherwise

eq(a,b)

is 1 if a==b, 0 otherwise

sin, cos, tan, sinh, cosh, tanh, exp, log, abs

vrc_override=<options> †

User specified quality for specific parts (ending, credits, ...) (pass 1/2). The options are <start-frame>, <end-frame>, <quality>[/<start-frame>, <end-frame>, <quality>[/...]]:

quality (2−31)

quantizer

quality (-500−0)

quality correction in %

vrc_init_cplx=<0−1000> †

initial complexity (pass 1)

vrc_init_occupancy=<0.0−1.0> †

initial buffer occupancy, as a fraction of vrc_buf_size (default: 0.9)

vqsquish=<0|1> †

Specify how to keep the quantizer between qmin and qmax (pass 1/2).

0

Use clipping.

1

Use a nice differentiable function (default).

vlelim=<-1000−1000> †

Sets single coefficient elimination threshold for luminance. Negative values will also consider the DC coefficient (should be at least -4 or lower for encoding at quant=1):

0

disabled (default)

-4

JVT recommendation

vcelim=<-1000−1000> †

Sets single coefficient elimination threshold for chrominance. Negative values will also consider the DC coefficient (should be at least -4 or lower for encoding at quant=1):

0

disabled (default)

7

JVT recommendation

vstrict=<-2|-1|0|1> †

strict standard compliance

0

disabled

1

Only recommended if you want to feed the output into the MPEG-4 reference decoder.

-1

Allow libavcodec specific extensions (default).

-2

Enables experimental codecs and features which may not be playable with future MPlayer versions (snow, ffvhuff, ffv1).

vdpart †

Data partitioning. Adds 2 Bytes per video packet, improves error-resistance when transferring over unreliable channels (e.g. streaming over the internet). Each video packet will be encoded in 3 separate partitions:

1. MVs

movement

2. DC coefficients

low res picture

3. AC coefficients

details

MV & DC are most important, loosing them looks far worse than loosing the AC and the 1. & 2. partition. (MV & DC) are far smaller than the 3. partition (AC) meaning that errors will hit the AC partition much more often than the MV & DC partitions. Thus, the picture will look better with partitioning than without, as without partitioning an error will trash AC/DC/MV equally.

vpsize=<0−10000> (also see vdpart) †

Video packet size, improves error-resistance.

0

disabled (default)

100−1000

good choice

ss †

slice structured mode for H.263+

gray †

grayscale only encoding (faster)

vfdct=<0−10> †

DCT algorithm

0

Automatically select a good one (default).

1

fast integer

2

accurate integer

3

MMX

4

mlib

5

AltiVec

6

floating point AAN

idct=<0−99> †

IDCT algorithm NOTE: To the best of our knowledge all these IDCTs do pass the IEEE1180 tests.

0

Automatically select a good one (default).

1

JPEG reference integer

2

simple

3

simplemmx

4

libmpeg2mmx (inaccurate, do not use for encoding with keyint >100)

5

ps2

6

mlib

7

arm

8

AltiVec

9

sh4

lumi_mask=<0.0−1.0> †

Luminance masking is a ’psychosensory’ setting that is supposed to make use of the fact that the human eye tends to notice fewer details in very bright parts of the picture. Luminance masking compresses bright areas stronger than medium ones, so it will save bits that can be spent again on other frames, raising overall subjective quality, while possibly reducing PSNR. WARNING: Be careful, overly large values can cause disastrous things. WARNING: Large values might look good on some monitors but may look horrible on other monitors.

0.0

disabled (default)

0.0−0.3

sane range

dark_mask=<0.0−1.0> †

Darkness masking is a ’psychosensory’ setting that is supposed to make use of the fact that the human eye tends to notice fewer details in very dark parts of the picture. Darkness masking compresses dark areas stronger than medium ones, so it will save bits that can be spent again on other frames, raising overall subjective quality, while possibly reducing PSNR. WARNING: Be careful, overly large values can cause disastrous things. WARNING: Large values might look good on some monitors but may look horrible on other monitors / TV / TFT.

0.0

disabled (default)

0.0−0.3

sane range

tcplx_mask=<0.0−1.0> †

Temporal complexity masking (default: 0.0 (disabled)). Imagine a scene with a bird flying across the whole scene; tcplx_mask will raise the quantizers of the bird’s macroblocks (thus decreasing their quality), as the human eye usually does not have time to see all the bird’s details. Be warned that if the masked object stops (e.g. the bird lands) it is likely to look horrible for a short period of time, until the encoder figures out that the object is not moving and needs refined blocks. The saved bits will be spent on other parts of the video, which may increase subjective quality, provided that tcplx_mask is carefully chosen.

scplx_mask=<0.0−1.0> †

Spatial complexity masking. Larger values help against blockiness, if no deblocking filter is used for decoding, which is maybe not a good idea. Imagine a scene with grass (which usually has great spatial complexity), a blue sky and a house; scplx_mask will raise the quantizers of the grass’ macroblocks, thus decreasing its quality, in order to spend more bits on the sky and the house. HINT: Crop any black borders completely as they will reduce the quality of the macroblocks (also applies without scplx_mask).

0.0

disabled (default)

0.0−0.5

sane range

NOTE: This setting does not have the same effect as using a custom matrix that would compress high frequencies harder, as scplx_mask will reduce the quality of P blocks even if only DC is changing. The result of scplx_mask will probably not look as good.

p_mask=<0.0−1.0> (also see vi_qfactor) †

Reduces the quality of inter blocks. This is equivalent to increasing the quality of intra blocks, because the same average bitrate will be distributed by the rate controller to the whole video sequence (default: 0.0 (disabled)). p_mask=1.0 doubles the bits allocated to each intra block.

border_mask=<0.0−1.0> †

border-processing for MPEG-style encoders. Border processing increases the quantizer for macroblocks which are less than 1/5th of the frame width/height away from the frame border, since they are often visually less important.

naq †

Normalize adaptive quantization (experimental). When using adaptive quantization (*_mask), the average per-MB quantizer may no longer match the requested frame-level quantizer. Naq will attempt to adjust the per-MB quantizers to maintain the proper average.

ildct †

Use interlaced DCT.

ilme †

Use interlaced motion estimation (mutually exclusive with qpel).

alt †

Use alternative scantable.

top=<-1−1> †

-1

automatic

0

bottom field first

1

top field first

format=<value> †

YV12

default

444P

for ffv1

422P

for HuffYUV, lossless JPEG and ffv1

411P,YVU9

for lossless JPEG and ffv1

BGR32

for lossless JPEG and ffv1

pred †

(for HuffYUV)

0

left prediction

1

plane/gradient prediction

2

median prediction

pred †

(for lossless JPEG)

0

left prediction

1

top prediction

2

topleft prediction

3

plane/gradient prediction

6

mean prediction

coder †

(for ffv1)

0

vlc coding (Golomb-Rice)

1

arithmetic coding (CABAC)

context †

(for ffv1)

0

small context model

1

large context model

(for ffvhuff)

:0predetermined Huffman tables (builtin or two pass)

1

adaptive Huffman tables

qpel †

Use quarter pel motion compensation (mutually exclusive with ilme). HINT: This seems only useful for high bitrate encodings.

mbcmp=<0−2000> †

Sets the comparison function for the macroblock decision, only used if mbd=0.

0 (SAD)

sum of absolute differences, fast (default)

1 (SSE)

sum of squared errors

2 (SATD)

sum of absolute Hadamard transformed differences

3 (DCT)

sum of absolute DCT transformed differences

4 (PSNR)

sum of squared quantization errors (avoid, low quality)

5 (BIT)

number of bits needed for the block

6 (RD)

rate distortion optimal, slow

7 (ZERO)

0

8 (VSAD)

sum of absolute vertical differences

9 (VSSE)

sum of squared vertical differences

10 (NSSE)

noise preserving sum of squared differences

11 (W53)

5/3 wavelet, only used in snow

12 (W97)

9/7 wavelet, only used in snow

+256

Also use chroma, currently does not work (correctly) with B-frames.

ildctcmp=<0−2000> †

Sets the comparison function for interlaced DCT decision (see mbcmp for available comparison functions).

precmp=<0−2000> †

Sets the comparison function for motion estimation pre pass (see mbcmp for available comparison functions) (default: 0).

cmp=<0−2000> †

Sets the comparison function for full pel motion estimation (see mbcmp for available comparison functions) (default: 0).

subcmp=<0−2000> †

Sets the comparison function for sub pel motion estimation (see mbcmp for available comparison functions) (default: 0).

nssew=<0−1000000> †

This setting controls NSSE weight, where larger weights will result in more noise. 0 NSSE is identical to SSE You may find this useful if you prefer to keep some noise in your encoded video rather than filtering it away before encoding (default: 8).

predia=<-99−6> †

diamond type and size for motion estimation pre-pass

dia=<-99−6> †

Diamond type & size for motion estimation. Motion search is an iterative process. Using a small diamond does not limit the search to finding only small motion vectors. It is just somewhat more likely to stop before finding the very best motion vector, especially when noise is involved. Bigger diamonds allow a wider search for the best motion vector, thus are slower but result in better quality. Big normal diamonds are better quality than shape-adaptive diamonds. Shape-adaptive diamonds are a good tradeoff between speed and quality. NOTE: The sizes of the normal diamonds and shape adaptive ones do not have the same meaning.

-3

shape adaptive (fast) diamond with size 3

-2

shape adaptive (fast) diamond with size 2

-1

slightly special: Can be slower and/or better than dia=-2.

1

normal size=1 diamond (default) =EPZS type diamond
0
000
0

2

normal size=2 diamond
0
000
00000
000
0

trell †

Trellis searched quantization. This will find the optimal encoding for each 8x8 block. Trellis searched quantization is quite simply an optimal quantization in the PSNR versus bitrate sense (Assuming that there would be no rounding errors introduced by the IDCT, which is obviously not the case.). It simply finds a block for the minimum of error and lambda*bits.

lambda

quantization parameter (QP) dependent constant

bits

amount of bits needed to encode the block

error

sum of squared errors of the quantization

cbp †

Rate distorted optimal coded block pattern. Will select the coded block pattern which minimizes distortion + lambda*rate. This can only be used together with trellis quantization.

mv0 †

Try to encode each MB with MV=<0,0> and choose the better one. This has no effect if mbd=0.

mv0_threshold=<any non-negative integer> †

When surrounding motion vectors are <0,0> and the motion estimation score of the current block is less than mv0_threshold, <0,0> is used for the motion vector and further motion estimation is skipped (default: 256). Lowering mv0_threshold to 0 can give a slight (0.01dB) PSNR increase and possibly make the encoded video look slightly better; raising mv0_threshold past 320 results in diminished PSNR and visual quality. Higher values speed up encoding very slightly (usually less than 1%, depending on the other options used). NOTE: This option does not require mv0 to be enabled.

qprd (mbd=2 only) †

rate distorted optimal quantization parameter (QP) for the given lambda of each macroblock

last_pred=<0−99> †

amount of motion predictors from the previous frame

0

(default)

a

Will use 2a+1 x 2a+1 macroblock square of motion vector predictors from the previous frame.

preme=<0−2> †

motion estimation pre-pass

0

disabled

1

only after I-frames (default)

2

always

subq=<1−8> †

subpel refinement quality (for qpel) (default: 8 (high quality)) NOTE: This has a significant effect on speed.

refs=<1−8> †

number of reference frames to consider for motion compensation (Snow only) (default: 1)

psnr †

print the PSNR (peak signal to noise ratio) for the whole video after encoding and store the per frame PSNR in a file with a name like ’psnr_hhmmss.log’. Returned values are in dB (decibel), the higher the better.

mpeg_quant †

Use MPEG quantizers instead of H.263.

aic †

Enable AC prediction for MPEG-4 or advanced intra prediction for H.263+. This will improve quality very slightly (around 0.02 dB PSNR) and slow down encoding very slightly (about 1%). NOTE: vqmin should be 8 or larger for H.263+ AIC.

aiv †

alternative inter vlc for H.263+

umv †

unlimited MVs (H.263+ only) Allows encoding of arbitrarily long MVs.

ibias=<-256−256> †

intra quantizer bias (256 equals 1.0, MPEG style quantizer default: 96, H.263 style quantizer default: 0) NOTE: The H.263 MMX quantizer cannot handle positive biases (set vfdct=1 or 2), the MPEG MMX quantizer cannot handle negative biases (set vfdct=1 or 2).

pbias=<-256−256> †

inter quantizer bias (256 equals 1.0, MPEG style quantizer default: 0, H.263 style quantizer default: -64) NOTE: The H.263 MMX quantizer cannot handle positive biases (set vfdct=1 or 2), the MPEG MMX quantizer cannot handle negative biases (set vfdct=1 or 2). HINT: A more positive bias (-32 − -16 instead of -64) seems to improve the PSNR.

nr=<0−100000> †

Noise reduction, 0 means disabled. 0−600 is a useful range for typical content, but you may want to turn it up a bit more for very noisy content (default: 0). Given its small impact on speed, you might want to prefer to use this over filtering noise away with video filters like denoise3d or hqdn3d.

qns=<0−3> †

Quantizer noise shaping. Rather than choosing quantization to most closely match the source video in the PSNR sense, it chooses quantization such that noise (usually ringing) will be masked by similar-frequency content in the image. Larger values are slower but may not result in better quality. This can and should be used together with trellis quantization, in which case the trellis quantization (optimal for constant weight) will be used as startpoint for the iterative search.

0

disabled (default)

1

Only lower the absolute value of coefficients.

2

Only change coefficients before the last non-zero coefficient + 1.

3

Try all.

inter_matrix=<comma separated matrix> †

Use custom inter matrix. It needs a comma separated string of 64 integers.

intra_matrix=<comma separated matrix> †

Use custom intra matrix. It needs a comma separated string of 64 integers.

vqmod_amp †

experimental quantizer modulation

vqmod_freq †

experimental quantizer modulation

dc †

intra DC precision in bits (default: 8). If you specify vcodec=mpeg2video this value can be 8, 9, 10 or 11.

cgop (also see sc_threshold) †

Close all GOPs. Currently it only works if scene change detection is disabled (sc_threshold=1000000000).

vglobal=<0−3> †

Control writing global video headers.

0

Codec decides where to write global headers (default).

1

Write global headers only in extradata (needed for .mp4/MOV/NUT).

2

Write global headers only in front of keyframes.

3

Combine 1 and 2.

aglobal=<0−3> †

Same as vglobal for audio headers.

---

• Prev
  Man/CODEC固有オプション/-lameopts
• Next
  Man/CODEC固有オプション/-lavfopts
• Home
  Up