Ok, let's see the result: 
Eric Martin MBARI 2021
This assumes you've built the binaries in this directory, the goal of this notebook is to demonstrate the following:
To date the project has been collecting the data through a raw packet capture tool, tshark. We have written C tools to go through and parse these files into CSV files of interest for broader accessibility. Some sample data is present in the repository.
!ls -lh ../data/
total 1226336 -rw-r--r-- 1 emartin emartin 737B Oct 4 11:43 mbariAtSeaWecRadar_00315_202109 -rwxrwxrwx 1 emartin emartin 200M Sep 28 17:20 mbariAtSeaWecRadar_00315_20210928171518.pcapng -rwxrwxrwx@ 1 emartin emartin 200M Sep 28 17:25 mbariAtSeaWecRadar_00316_20210928172018.pcapng -rwxrwxrwx@ 1 emartin emartin 200M Sep 30 14:20 mbariAtSeaWecRadar_00684_20210930000030.pcapng
We are pretty interested in the biggest day observed yet, 9/28/2021. Let's process that one. First let's process the radar data. The program, if built and installed should print out help with no inputs:
!pcaptoradar
Let's make sure there are data in the file:
!pcaptoradar -s ../data/mbariAtSeaWecRadar_00315_20210928171518.pcapng
Opening file: ../data/mbariAtSeaWecRadar_00315_20210928171518.pcapng ---------------- REPORTS_A: 300 REPORTS_B: 300 FRAMES_A: 11469 SCANS: 367008 REVS_A: 89 AHRS: 140896 ----------------
Yes, there are reports data for both channels on the radar, and many Frames of sonar data, covering 367K radar scan lines, which add up to 89 full revolutions of the radar for the entire file.
Next, let's convert the reports, and just print the header:
!pcaptoradar -ah ../data/mbariAtSeaWecRadar_00315_20210928171518.pcapng | head -n4
Opening file: ../data/mbariAtSeaWecRadar_00315_20210928171518.pcapng time,channel,range,gain,sea_auto,sea,rain,interference_rejection,target_expansion,target_boost 1632874519.070718,3,5000,60,0,25,0,0,1,1 1632874520.070762,3,5000,60,0,25,0,0,1,1 1632874521.070754,3,5000,60,0,25,0,0,1,1
We can see that this isn't changing, which is good, at the time of this analysis, there is little change happening on the radar setting, and all of these values make sense (but range is 500.0meters, so something is a bit off). If we wanted to make a file for all the reports we simply redirecto the stdout to a file, as so:
!pcaptoradar -ah ../data/mbariAtSeaWecRadar_00315_20210928171518.pcapng > reports_a.csv
Opening file: ../data/mbariAtSeaWecRadar_00315_20210928171518.pcapng
!ls -lh *.csv
-rw-r--r-- 1 emartin emartin 631K Oct 4 12:38 ahrs.csv -rw-r--r-- 1 emartin emartin 12K Oct 5 09:48 reports_a.csv -rw-r--r-- 1 emartin emartin 745M Oct 4 13:59 scans.csv
import pandas as pd
reports_df = pd.read_csv('reports_a.csv')
reports_df.head()
| time | channel | range | gain | sea_auto | sea | rain | interference_rejection | target_expansion | target_boost | |
|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 1.632875e+09 | 3 | 5000 | 60 | 0 | 25 | 0 | 0 | 1 | 1 |
| 1 | 1.632875e+09 | 3 | 5000 | 60 | 0 | 25 | 0 | 0 | 1 | 1 |
| 2 | 1.632875e+09 | 3 | 5000 | 60 | 0 | 25 | 0 | 0 | 1 | 1 |
| 3 | 1.632875e+09 | 3 | 5000 | 60 | 0 | 25 | 0 | 0 | 1 | 1 |
| 4 | 1.632875e+09 | 3 | 5000 | 60 | 0 | 25 | 0 | 0 | 1 | 1 |
Much like the reports data, we can generate csv files using the pcaptoahrs tool.
!pcaptoahrs
ahrs lcm parser, usage pcaptoahrs <pcap filename> options: -h (enable csv header)
!pcaptoahrs -h ../data/mbariAtSeaWecRadar_00315_20210928171518.pcapng | head -n2
Opening file: ../data/mbariAtSeaWecRadar_00315_20210928171518.pcapng pcap_time,lcm_time,latitude,longitude,yaw_true,pitch,roll,x_accel,y_accel,z_accel,n_velocity,e_velocity,d_velocity,altitude,x_rate,y_rate,z_rate, x_rate_temp, 1632874518.862951,1632874518.862882,36.743080139160156,-121.881141662597656,140.372314,-1.356812,-0.747070,0.074768,0.059814,-0.993958,0.171875,-0.132812,-0.507812,-33.750000,6.517639,-0.134583,1.038208,24.157715,
!pcaptoahrs -h ../data/mbariAtSeaWecRadar_00315_20210928171518.pcapng > ahrs.csv
Opening file: ../data/mbariAtSeaWecRadar_00315_20210928171518.pcapng
!ls -lh *.csv
-rw-r--r-- 1 emartin emartin 631K Oct 5 13:04 ahrs.csv -rw-r--r-- 1 emartin emartin 12K Oct 5 09:48 reports_a.csv -rw-r--r-- 1 emartin emartin 745M Oct 4 13:59 scans.csv
import pandas as pd
ahrs_df = pd.read_csv('ahrs.csv')
ahrs_df.head()
| pcap_time | lcm_time | latitude | longitude | yaw_true | pitch | roll | x_accel | y_accel | z_accel | n_velocity | e_velocity | d_velocity | altitude | x_rate | y_rate | z_rate | x_rate_temp | Unnamed: 18 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 1.632875e+09 | 1.632875e+09 | 36.743080 | -121.881142 | 140.372314 | -1.356812 | -0.747070 | 0.074768 | 0.059814 | -0.993958 | 0.171875 | -0.132812 | -0.507812 | -33.75 | 6.517639 | -0.134583 | 1.038208 | 24.157715 | NaN |
| 1 | 1.632875e+09 | 1.632875e+09 | 36.743080 | -121.881142 | 140.460205 | -1.444702 | -0.126343 | 0.092773 | 0.031433 | -1.018982 | 0.062500 | -0.093750 | -0.515625 | -33.75 | 5.787048 | -1.538086 | 1.018982 | 24.157715 | NaN |
| 2 | 1.632875e+09 | 1.632875e+09 | 36.743084 | -121.881142 | 140.537109 | -1.713867 | 0.302124 | 0.097656 | 0.003357 | -1.034851 | 0.031250 | 0.007812 | -0.539062 | -33.75 | 3.441467 | -3.595276 | 0.826721 | 24.154663 | NaN |
| 3 | 1.632875e+09 | 1.632875e+09 | 36.743084 | -121.881142 | 140.603027 | -2.208252 | 0.532837 | 0.084229 | -0.025635 | -1.043701 | -0.054688 | 0.093750 | -0.578125 | -33.75 | 1.134338 | -5.902405 | 0.596008 | 24.151611 | NaN |
| 4 | 1.632875e+09 | 1.632875e+09 | 36.743084 | -121.881142 | 140.668945 | -2.861938 | 0.499878 | 0.058899 | -0.046997 | -1.060486 | -0.007812 | 0.148438 | -0.578125 | -33.50 | -1.345825 | -7.286682 | 0.730591 | 24.151611 | NaN |
Our last pcap file process is to take the file, restructure the packets, and turn it into something we can parse in in a higher level language, like python:
!pcaptoradar -AH ../data/mbariAtSeaWecRadar_00315_20210928171518.pcapng > scans.csv
Opening file: ../data/mbariAtSeaWecRadar_00315_20210928171518.pcapng
That took a while, but now we have the file in CSV form, let's take a look at how large it ended up.
!ls -lh *.csv
-rw-r--r-- 1 emartin emartin 631K Oct 4 12:38 ahrs.csv -rw-r--r-- 1 emartin emartin 12K Oct 4 12:33 reports_a.csv -rw-r--r-- 1 emartin emartin 745M Oct 4 13:59 scans.csv
So in the end, from a 200MB file, we are generating another file that is 745MB. Let's see if we can parse the file with pandas, and then the fun can begin.
import pandas as pd
#Set the dtypes to improve memory performance on import
dtypes_scans = {'time':'float',
'channel':'string',
'scan_line':'uint16',
'heading':'int32',
'angle':'uint16',
'angle_d':'float',
'range':'uint16',
}
for i in range(1024):
colname = f'r{i:04}'
dtypes_scans[colname]='uint8'
%%time
df_scans = pd.read_csv('scans.csv', dtype=dtypes_scans)
CPU times: user 20 s, sys: 916 ms, total: 20.9 s Wall time: 20.9 s
df_scans.tail()
| time | channel | scan_line | heading | angle | angle_d | range | r0000 | r0001 | r0002 | ... | r1015 | r1016 | r1017 | r1018 | r1019 | r1020 | r1021 | r1022 | r1023 | Unnamed: 1031 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 367003 | 1.632875e+09 | A | 444 | -32768 | 89 | 7.822266 | 808 | 15 | 15 | 15 | ... | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | NaN |
| 367004 | 1.632875e+09 | A | 445 | -32768 | 91 | 7.998047 | 808 | 15 | 15 | 15 | ... | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | NaN |
| 367005 | 1.632875e+09 | A | 446 | -32768 | 93 | 8.173828 | 808 | 15 | 15 | 15 | ... | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | NaN |
| 367006 | 1.632875e+09 | A | 447 | -32768 | 95 | 8.349609 | 808 | 15 | 15 | 15 | ... | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | NaN |
| 367007 | 1.632875e+09 | A | 448 | -32768 | 97 | 8.525391 | 808 | 15 | 15 | 15 | ... | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | NaN |
5 rows × 1032 columns
df_scans.info()
<class 'pandas.core.frame.DataFrame'> RangeIndex: 367008 entries, 0 to 367007 Columns: 1032 entries, time to Unnamed: 1031 dtypes: float64(3), int32(1), string(1), uint16(3), uint8(1024) memory usage: 373.1 MB
df_scans.time.iloc[-1]
1632874818.762966
Let's find the first full revolution, scan_line set to 0.
# Get the first full scan
idx = df_scans.index[df_scans['angle'] == 1].to_list()
scan = df_scans.loc[idx[40]:idx[41]-1]
scan.info()
<class 'pandas.core.frame.DataFrame'> RangeIndex: 2048 entries, 82287 to 84334 Columns: 1032 entries, time to Unnamed: 1031 dtypes: float64(3), int32(1), string(1), uint16(3), uint8(1024) memory usage: 2.1 MB
Good! there are 2048 scans in one full revolution, we have something to look at.
from PIL import Image as im
import numpy as np
# Get an uint8 array from the scans
scan_data = scan.iloc[:, 7:-1].to_numpy()*16
max_scan = np.amax(scan_data)
print(max_scan)
240
image_ = im.fromarray(scan_data)
image_.save('scan_nonpolar.jpg')
Ok, let's see the result:
The final step in creating a monochrome image is to reshape the array into a cartesian one. At the end of the day, we should have a square bound by the radius of the spokes, or $2*(1024)=2048$. Let's take a brute force attempt at this.
#create an array to pack
cart = np.zeros((2048,2048),dtype='uint8')
# def a remapping function
def map_polar(angle,length, max_angle=2048, max_length=1024):
'''
angle in decimal degrees
length of array to output
returns -- x and y arrays for index mapping
'''
ANG2RAD = 2*np.pi/max_angle
x = length * np.cos(angle*ANG2RAD)
y = length * np.sin(angle*ANG2RAD)
## shift into positive spaces
x = np.round(x)
y = np.round(y)
x += max_length
y += max_length
return (int(x),int(y))
map_polar(1024,1024)
(0, 1024)
# Super stupidly, let's go line by line and write over the data for the one before
it = np.nditer(scan_data, flags=['multi_index'])
for x in it:
if x == 0:
continue
r,c = it.multi_index
crow,ccol = map_polar(r,c)
cart[crow,ccol] = x
# make an image from the cartesian converted array
cart_img = im.fromarray(cart)
cart_img.save('scan_polar.jpg')
Here is the result 
ls -lh
total 1561664
-rw-r--r--@ 1 emartin emartin 617K Oct 5 12:50 Radar Image Data Exploration.html
-rw-r--r-- 1 emartin emartin 48K Oct 5 13:12 Radar Image Data Exploration.ipynb
-rw-r--r-- 1 emartin emartin 631K Oct 5 13:04 ahrs.csv
drwxr-xr-x 361 emartin emartin 11K Oct 5 00:07 img/
-rw-r--r-- 1 emartin emartin 8.9M Oct 5 09:40 output.mp4
-rw-r--r-- 1 emartin emartin 12K Oct 5 09:48 reports_a.csv
-rw-r--r-- 1 emartin emartin 61K Oct 5 13:13 scan_nonpolar.jpg
-rw-r--r-- 1 emartin emartin 98K Oct 5 13:13 scan_polar.jpg
-rw-r--r-- 1 emartin emartin 745M Oct 4 13:59 scans.csv
-rw-r--r-- 1 emartin emartin 4.0K Oct 5 13:10 wec_radar_movie.py
Just a little sandbox area for me to sort out the best way to get a single yaw offset for the whole revolution.
start_t = scan.time.iloc[0]
end_t = scan.time.iloc[-1]
print(start_t, end_t, end_t-start_t)
1632874586.014958 1632874587.689166 1.674208164215088
ahrs_df.info()
<class 'pandas.core.frame.DataFrame'> RangeIndex: 3005 entries, 0 to 3004 Data columns (total 19 columns): # Column Non-Null Count Dtype --- ------ -------------- ----- 0 pcap_time 3005 non-null float64 1 lcm_time 3005 non-null float64 2 latitude 3005 non-null float64 3 longitude 3005 non-null float64 4 yaw_true 3005 non-null float64 5 pitch 3005 non-null float64 6 roll 3005 non-null float64 7 x_accel 3005 non-null float64 8 y_accel 3005 non-null float64 9 z_accel 3005 non-null float64 10 n_velocity 3005 non-null float64 11 e_velocity 3005 non-null float64 12 d_velocity 3005 non-null float64 13 altitude 3005 non-null float64 14 x_rate 3005 non-null float64 15 y_rate 3005 non-null float64 16 z_rate 3005 non-null float64 17 x_rate_temp 3005 non-null float64 18 Unnamed: 18 0 non-null float64 dtypes: float64(19) memory usage: 446.2 KB
scan_angles = ahrs_df.loc[(ahrs_df.lcm_time >= start_t) & (ahrs_df.lcm_time <=end_t)]
scan_angles.yaw_true.describe()
type(scan_angles.yaw_true.mean())
numpy.float64
10+None
--------------------------------------------------------------------------- TypeError Traceback (most recent call last) <ipython-input-71-7364e94b6574> in <module> ----> 1 10+None TypeError: unsupported operand type(s) for +: 'int' and 'NoneType'
Let's turn this into and full script, here are the contents of wec_radar_movie.py file subject to change:
!cat wec_radar_movie.py
import numpy as np
import pandas as pd
from pandas.core.frame import DataFrame
class RadarScans():
def __init__(self, csvfile:str, ahrs_csv:str = 'None', mode='cartesian'):
print(f"opening file: {csvfile}")
self.scan_df = RadarScans.open_csv_file(csvfile)
self.ahrs_df = RadarScans.open_ahrs_file(ahrs_csv)
self.initialize_scan_indexes()
self.mode=mode
def open_csv_file(csvfile:str):
dtypes_scans = {'time':'float',
'channel':'string',
'scan_line':'uint16',
'heading':'int32',
'angle':'uint16',
'angle_d':'float',
'range':'uint16',
}
for i in range(1024):
colname = f'r{i:04}'
dtypes_scans[colname]='uint8'
scan_df = pd.read_csv(csvfile, dtype=dtypes_scans)
return scan_df
def open_ahrs_file(csvfile):
ahrs_df = pd.read_csv(csvfile)
return ahrs_df
def initialize_scan_indexes(self):
self.indexes = self.scan_df.index[self.scan_df['angle']==1].to_list()
self.num_scans = len(self.indexes)-1
print(f'found {self.num_scans} full revolutions in file.')
def select_scan(self, n:int):
if (n >= self.num_scans):
return None;
idx = self.indexes
scan = self.scan_df.loc[idx[n]:idx[n+1]-1]
return scan
def scan_angle_offset(self, scan:DataFrame):
# Guard against uninitialized access
if self.ahrs_df is None:
raise Exception('No AHRS data present')
# Find the bounds of the time window
start_t = scan.time.iloc[0]
end_t = scan.time.iloc[-1]
yaw_true_mean = self.ahrs_df.loc[(self.ahrs_df.lcm_time >= start_t) & (self.ahrs_df.lcm_time <=end_t)].yaw_true.mean();
return yaw_true_mean
def convert_to_polar(self, scan, scan_df, returns_per_scanline=1024):
#create an array to pack
out = np.zeros((2*returns_per_scanline,2*returns_per_scanline),dtype='uint8')
it = np.nditer(scan, flags=['multi_index'])
if self.ahrs_df is None:
yaw_angle=0
else:
yaw_angle = self.scan_angle_offset(scan_df)
for x in it:
r,c = it.multi_index
if x == 0:
continue
crow,ccol = RadarScans.map_polar(r,c, yaw=yaw_angle)
out[crow,ccol] = x
return out
def map_polar(angle,length, yaw=0, max_angle=2048, max_length=1024):
'''
angle in integer representaion, scaled by Max_angle (default 0-2048)
length of array to output
returns -- x and y arrays for index mapping
'''
DEG2RAD = np.pi/180
ANG2RAD = 2*np.pi/max_angle
f_angle = ANG2RAD*angle
f_angle += yaw*DEG2RAD
x = length * np.cos(f_angle)
y = length * np.sin(f_angle)
## shift into positive spaces
x = np.round(x)
y = np.round(y)
x += max_length
y += max_length
return (int(x),int(y))
def __iter__(self):
self.current_scan = 0
return self
def __next__(self):
"""Returns a ndarray of uint8 that can be used for image/movie conversion
"""
scan =self.select_scan(self.current_scan)
if scan is None:
raise StopIteration
out = scan.iloc[:, 7:-1].to_numpy()*16
if self.mode == 'polar':
out = self.convert_to_polar(out, scan)
self.current_scan += 1
return out
if __name__ == '__main__':
from PIL import Image
fname = 'scans.csv'
scanner = RadarScans(fname, ahrs_csv='ahrs.csv', mode='polar')
cnt = 0
for scan in scanner:
image = Image.fromarray(scan)
im_name = f'./img/scan_pa_{cnt:04}.jpg'
print(f'Writing image: {im_name}...')
image.save(im_name)
cnt+=1
Now we have a set of images that we can turn into a movie. There are many ways to achieve this task, for now we will use ffmpeg.
!ffmpeg -framerate 12 -i ./img/scan_p_%04d.jpg -vf format=yuv420p output.mp4
ffmpeg version 4.4 Copyright (c) 2000-2021 the FFmpeg developers built with Apple clang version 12.0.0 (clang-1200.0.32.29) configuration: --prefix=/usr/local/Cellar/ffmpeg/4.4_2 --enable-shared --enable-pthreads --enable-version3 --cc=clang --host-cflags= --host-ldflags= --enable-ffplay --enable-gnutls --enable-gpl --enable-libaom --enable-libbluray --enable-libdav1d --enable-libmp3lame --enable-libopus --enable-librav1e --enable-librubberband --enable-libsnappy --enable-libsrt --enable-libtesseract --enable-libtheora --enable-libvidstab --enable-libvorbis --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxml2 --enable-libxvid --enable-lzma --enable-libfontconfig --enable-libfreetype --enable-frei0r --enable-libass --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-libspeex --enable-libsoxr --enable-libzmq --enable-libzimg --disable-libjack --disable-indev=jack --enable-avresample --enable-videotoolbox libavutil 56. 70.100 / 56. 70.100 libavcodec 58.134.100 / 58.134.100 libavformat 58. 76.100 / 58. 76.100 libavdevice 58. 13.100 / 58. 13.100 libavfilter 7.110.100 / 7.110.100 libavresample 4. 0. 0 / 4. 0. 0 libswscale 5. 9.100 / 5. 9.100 libswresample 3. 9.100 / 3. 9.100 libpostproc 55. 9.100 / 55. 9.100 Input #0, image2, from './img/scan_p_%04d.jpg': Duration: 00:00:14.92, start: 0.000000, bitrate: N/A Stream #0:0: Video: mjpeg (Baseline), gray(bt470bg/unknown/unknown), 2048x2048 [SAR 1:1 DAR 1:1], 12 fps, 12 tbr, 12 tbn, 12 tbc Stream mapping: Stream #0:0 -> #0:0 (mjpeg (native) -> h264 (libx264)) Press [q] to stop, [?] for help [libx264 @ 0x7fcc0c008e00] using SAR=1/1 [libx264 @ 0x7fcc0c008e00] using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX FMA3 BMI2 AVX2 [libx264 @ 0x7fcc0c008e00] profile High, level 5.0, 4:2:0, 8-bit [libx264 @ 0x7fcc0c008e00] 264 - core 163 r3060 5db6aa6 - H.264/MPEG-4 AVC codec - Copyleft 2003-2021 - http://www.videolan.org/x264.html - options: cabac=1 ref=3 deblock=1:0:0 analyse=0x3:0x113 me=hex subme=7 psy=1 psy_rd=1.00:0.00 mixed_ref=1 me_range=16 chroma_me=1 trellis=1 8x8dct=1 cqm=0 deadzone=21,11 fast_pskip=1 chroma_qp_offset=-2 threads=24 lookahead_threads=4 sliced_threads=0 nr=0 decimate=1 interlaced=0 bluray_compat=0 constrained_intra=0 bframes=3 b_pyramid=2 b_adapt=1 b_bias=0 direct=1 weightb=1 open_gop=0 weightp=2 keyint=250 keyint_min=12 scenecut=40 intra_refresh=0 rc_lookahead=40 rc=crf mbtree=1 crf=23.0 qcomp=0.60 qpmin=0 qpmax=69 qpstep=4 ip_ratio=1.40 aq=1:1.00 Output #0, mp4, to 'output.mp4': Metadata: encoder : Lavf58.76.100 Stream #0:0: Video: h264 (avc1 / 0x31637661), yuv420p(tv, bt470bg/unknown/unknown, progressive), 2048x2048 [SAR 1:1 DAR 1:1], q=2-31, 12 fps, 12288 tbn Metadata: encoder : Lavc58.134.100 libx264 Side data: cpb: bitrate max/min/avg: 0/0/0 buffer size: 0 vbv_delay: N/A frame= 179 fps= 44 q=-1.0 Lsize= 9146kB time=00:00:14.66 bitrate=5108.4kbits/s speed=3.63x video:9143kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.032609% [libx264 @ 0x7fcc0c008e00] frame I:16 Avg QP:17.88 size: 88040 [libx264 @ 0x7fcc0c008e00] frame P:55 Avg QP:24.82 size: 46904 [libx264 @ 0x7fcc0c008e00] frame B:108 Avg QP:25.82 size: 49754 [libx264 @ 0x7fcc0c008e00] consecutive B-frames: 15.6% 6.7% 15.1% 62.6% [libx264 @ 0x7fcc0c008e00] mb I I16..4: 18.3% 75.9% 5.8% [libx264 @ 0x7fcc0c008e00] mb P I16..4: 0.3% 10.4% 3.7% P16..4: 1.4% 0.3% 0.1% 0.0% 0.0% skip:83.8% [libx264 @ 0x7fcc0c008e00] mb B I16..4: 0.1% 1.4% 3.0% B16..8: 14.1% 0.9% 0.3% direct: 0.3% skip:79.8% L0:50.8% L1:48.1% BI: 1.1% [libx264 @ 0x7fcc0c008e00] 8x8 transform intra:67.2% inter:33.8% [libx264 @ 0x7fcc0c008e00] coded y,uvDC,uvAC intra: 19.6% 0.0% 0.0% inter: 1.4% 0.0% 0.0% [libx264 @ 0x7fcc0c008e00] i16 v,h,dc,p: 88% 7% 5% 0% [libx264 @ 0x7fcc0c008e00] i8 v,h,dc,ddl,ddr,vr,hd,vl,hu: 51% 6% 39% 1% 1% 1% 1% 1% 1% [libx264 @ 0x7fcc0c008e00] i4 v,h,dc,ddl,ddr,vr,hd,vl,hu: 19% 13% 38% 6% 5% 6% 5% 5% 4% [libx264 @ 0x7fcc0c008e00] i8c dc,h,v,p: 100% 0% 0% 0% [libx264 @ 0x7fcc0c008e00] Weighted P-Frames: Y:0.0% UV:0.0% [libx264 @ 0x7fcc0c008e00] ref P L0: 68.6% 1.4% 21.7% 8.3% [libx264 @ 0x7fcc0c008e00] ref B L0: 82.0% 15.8% 2.2% [libx264 @ 0x7fcc0c008e00] ref B L1: 95.1% 4.9% [libx264 @ 0x7fcc0c008e00] kb/s:5020.83
from IPython.display import Video
Video('output.mp4', width=400)
!ffmpeg -framerate 12 -i ./img/scan_pa_%04d.jpg -vf format=yuv420p output_corrected.mp4
ffmpeg version 4.4 Copyright (c) 2000-2021 the FFmpeg developers built with Apple clang version 12.0.0 (clang-1200.0.32.29) configuration: --prefix=/usr/local/Cellar/ffmpeg/4.4_2 --enable-shared --enable-pthreads --enable-version3 --cc=clang --host-cflags= --host-ldflags= --enable-ffplay --enable-gnutls --enable-gpl --enable-libaom --enable-libbluray --enable-libdav1d --enable-libmp3lame --enable-libopus --enable-librav1e --enable-librubberband --enable-libsnappy --enable-libsrt --enable-libtesseract --enable-libtheora --enable-libvidstab --enable-libvorbis --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxml2 --enable-libxvid --enable-lzma --enable-libfontconfig --enable-libfreetype --enable-frei0r --enable-libass --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-libspeex --enable-libsoxr --enable-libzmq --enable-libzimg --disable-libjack --disable-indev=jack --enable-avresample --enable-videotoolbox libavutil 56. 70.100 / 56. 70.100 libavcodec 58.134.100 / 58.134.100 libavformat 58. 76.100 / 58. 76.100 libavdevice 58. 13.100 / 58. 13.100 libavfilter 7.110.100 / 7.110.100 libavresample 4. 0. 0 / 4. 0. 0 libswscale 5. 9.100 / 5. 9.100 libswresample 3. 9.100 / 3. 9.100 libpostproc 55. 9.100 / 55. 9.100 Input #0, image2, from './img/scan_pa_%04d.jpg': Duration: 00:00:14.92, start: 0.000000, bitrate: N/A Stream #0:0: Video: mjpeg (Baseline), gray(bt470bg/unknown/unknown), 2048x2048 [SAR 1:1 DAR 1:1], 12 fps, 12 tbr, 12 tbn, 12 tbc Stream mapping: Stream #0:0 -> #0:0 (mjpeg (native) -> h264 (libx264)) Press [q] to stop, [?] for help [libx264 @ 0x7fc933808e00] using SAR=1/1 [libx264 @ 0x7fc933808e00] using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX FMA3 BMI2 AVX2 [libx264 @ 0x7fc933808e00] profile High, level 5.0, 4:2:0, 8-bit [libx264 @ 0x7fc933808e00] 264 - core 163 r3060 5db6aa6 - H.264/MPEG-4 AVC codec - Copyleft 2003-2021 - http://www.videolan.org/x264.html - options: cabac=1 ref=3 deblock=1:0:0 analyse=0x3:0x113 me=hex subme=7 psy=1 psy_rd=1.00:0.00 mixed_ref=1 me_range=16 chroma_me=1 trellis=1 8x8dct=1 cqm=0 deadzone=21,11 fast_pskip=1 chroma_qp_offset=-2 threads=24 lookahead_threads=4 sliced_threads=0 nr=0 decimate=1 interlaced=0 bluray_compat=0 constrained_intra=0 bframes=3 b_pyramid=2 b_adapt=1 b_bias=0 direct=1 weightb=1 open_gop=0 weightp=2 keyint=250 keyint_min=12 scenecut=40 intra_refresh=0 rc_lookahead=40 rc=crf mbtree=1 crf=23.0 qcomp=0.60 qpmin=0 qpmax=69 qpstep=4 ip_ratio=1.40 aq=1:1.00 Output #0, mp4, to 'output_corrected.mp4': Metadata: encoder : Lavf58.76.100 Stream #0:0: Video: h264 (avc1 / 0x31637661), yuv420p(tv, bt470bg/unknown/unknown, progressive), 2048x2048 [SAR 1:1 DAR 1:1], q=2-31, 12 fps, 12288 tbn Metadata: encoder : Lavc58.134.100 libx264 Side data: cpb: bitrate max/min/avg: 0/0/0 buffer size: 0 vbv_delay: N/A frame= 179 fps= 45 q=-1.0 Lsize= 9171kB time=00:00:14.66 bitrate=5122.4kbits/s speed=3.72x video:9168kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.032349% [libx264 @ 0x7fc933808e00] frame I:15 Avg QP:17.41 size: 86581 [libx264 @ 0x7fc933808e00] frame P:57 Avg QP:24.93 size: 49589 [libx264 @ 0x7fc933808e00] frame B:107 Avg QP:25.76 size: 49179 [libx264 @ 0x7fc933808e00] consecutive B-frames: 15.1% 10.1% 16.8% 58.1% [libx264 @ 0x7fc933808e00] mb I I16..4: 19.8% 74.5% 5.7% [libx264 @ 0x7fc933808e00] mb P I16..4: 0.3% 10.8% 4.0% P16..4: 1.5% 0.3% 0.1% 0.0% 0.0% skip:83.1% [libx264 @ 0x7fc933808e00] mb B I16..4: 0.1% 1.4% 3.0% B16..8: 14.1% 0.9% 0.3% direct: 0.3% skip:80.0% L0:50.9% L1:47.9% BI: 1.2% [libx264 @ 0x7fc933808e00] 8x8 transform intra:66.4% inter:33.1% [libx264 @ 0x7fc933808e00] coded y,uvDC,uvAC intra: 20.1% 0.0% 0.0% inter: 1.4% 0.0% 0.0% [libx264 @ 0x7fc933808e00] i16 v,h,dc,p: 89% 7% 5% 0% [libx264 @ 0x7fc933808e00] i8 v,h,dc,ddl,ddr,vr,hd,vl,hu: 49% 5% 41% 1% 1% 1% 1% 1% 1% [libx264 @ 0x7fc933808e00] i4 v,h,dc,ddl,ddr,vr,hd,vl,hu: 19% 12% 40% 5% 5% 5% 5% 4% 4% [libx264 @ 0x7fc933808e00] i8c dc,h,v,p: 100% 0% 0% 0% [libx264 @ 0x7fc933808e00] Weighted P-Frames: Y:0.0% UV:0.0% [libx264 @ 0x7fc933808e00] ref P L0: 70.3% 1.4% 21.2% 7.0% [libx264 @ 0x7fc933808e00] ref B L0: 81.2% 16.7% 2.2% [libx264 @ 0x7fc933808e00] ref B L1: 95.8% 4.2% [libx264 @ 0x7fc933808e00] kb/s:5034.62
from IPython.display import Video
Video('output_corrected.mp4', width=400)