-
Notifications
You must be signed in to change notification settings - Fork 0
/
ultrafast_timelapse.R
142 lines (128 loc) · 6.07 KB
/
ultrafast_timelapse.R
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
dyn.load("lumicomp.so")
args <- commandArgs(trailingOnly=TRUE)
if (!exists("indir")) indir <- args[1]
if (!exists("filtermode")) filtermode <- args[2] # full / summer
if (!exists("nb_seconds")) nb_seconds <- as.integer(args[3]) # 10
nb_frames <- 25 * nb_seconds
noon_hour <- 13
camID <- basename(indir)
outdir <- sprintf("/tmp/tl_%s_%s", filtermode, camID)
video_output <- sprintf("tl_%s_%s_lumi.mp4", filtermode, camID)
writeLines(sprintf("Outdir: %s (%d seconds)", outdir, nb_seconds))
dir.create(outdir, showWarnings = FALSE)
images <- data.frame(datei = list.files(indir, pattern="jpg"), stringsAsFactors=FALSE)
images$pfad <- sprintf("%s/%s", indir, images$datei)
images$filesize <- file.info(images$pfad)$size
if (sum(grepl("_", images$datei)) > 0) {
images$timestamp <- as.POSIXct(sub("netcam", "", images$datei), tz="", "%Y%m%d_%H%M%S")
images$unixtime <- as.numeric(images$timestamp)
} else {
images$unixtime <- as.integer(sub(".jpg", "", images$datei))
images$timestamp <- as.POSIXct(images$unixtime, origin="1970-01-01")
}
images$date <- as.Date(images$timestamp)
images$stunde <- as.numeric(format(images$timestamp, "%H")) + as.numeric(format(images$timestamp, "%M")) / 60
images$jahr <- as.numeric(format(images$timestamp, "%Y"))
images$monat <- as.numeric(format(images$timestamp, "%m"))
images$midiff <- abs(noon_hour - images$stunde)
images <- images[images$date < Sys.Date(),] # Images from the future are not plausible
if (filtermode == "summer") subimg <- images[images$monat %in% 4:10 & images$midiff < 2,]
if (filtermode == "winter") subimg <- images[!images$monat %in% 4:10 & images$midiff < 2,]
if (filtermode == "noon") subimg <- images[images$midiff < 2,]
if (filtermode == "none") subimg <- images
writeLines(sprintf("images = %d, days = %d, filtered_images = %d, filtered_days = %d", nrow(images), length(unique(images$date)), nrow(subimg), length(unique(subimg$date))))
print(table(images$jahr))
days <- sort(unique(subimg$date))
nb_days <- length(days)
# Determine number of blocks
suitable_divisors <- function(number) {
blocksize <- numbers::divisors(nb_days)
blocksize[blocksize > 2 & blocksize < 30]
}
blocksize <- suitable_divisors(nb_days)
while (length(blocksize) == 0) {
nb_days <- nb_days - 1
blocksize <- suitable_divisors(nb_days)
}
blocksize <- max(blocksize)
nb_blocks <- nb_days / blocksize
# Remove days with least number of images
if (nb_days != length(days)) {
dayocc <- as.data.frame.table(table(subimg$date))
dayocc <- dayocc[order(dayocc$Freq),]
nb_remdays <- length(days) - nb_days
remdays <- as.Date(dayocc[1:nb_remdays, "Var1"])
nb_remimgs <- sum(dayocc[1:nb_remdays, "Freq"])
subimg <- subimg[!subimg$date %in% remdays,]
days <- sort(unique(subimg$date))
writeLines(sprintf("Removed %d days containing %d images", nb_remdays, nb_remimgs))
}
# Determine number of images per block
imgs_per_block <- round(nb_frames / nb_blocks)
if (imgs_per_block < 1) imgs_per_block <- 1
nb_frames_new <- nb_blocks * imgs_per_block
writeLines(sprintf("nb_days = %d, blocksize = %d days, nb_blocks = %d, nb_frames_target = %d, nb_frames_real = %d, imgs_per_block = %d", nb_days, blocksize, nb_blocks, nb_frames, nb_frames_new, imgs_per_block))
# Select largest image in first block as initial frame.
subimg <- subimg[order(subimg$unixtime),]
seldays <- days[(0*blocksize+1):(0*blocksize+blocksize)]
selimgs <- subimg[subimg$date %in% seldays,]
selimgs <- selimgs[order(selimgs$filesize, decreasing=TRUE),]
selimgs <- selimgs[selimgs$filesize >= mean(selimgs$filesize),] # optional
init_img <- selimgs[1, "pfad"]
writeLines(sprintf("Initial image: %s", init_img))
subimg[subimg$pfad == init_img, "Auswahl"] <- TRUE
writeLines(sprintf("First pack: %d images", nrow(selimgs)))
for (i in 2:nrow(selimgs)) {
lcr <- .C("lumi_compare", filename1=init_img, filename2=selimgs[i, "pfad"], dist=.Machine$integer.max)
selimgs[i, "distance"] <- lcr$dist
#writeLines(sprintf("%d/%d: %f", i, nrow(selimgs), selimgs[i, "distance"]))
}
# Select n with lowest distance to initial frame.
selimgs <- selimgs[order(selimgs$distance),]
top_imgs <- selimgs[1:(imgs_per_block-1), "unixtime"]
subimg[subimg$unixtime %in% top_imgs, "Auswahl"] <- TRUE
last_img <- tail(subimg[!is.na(subimg$Auswahl), "pfad"], n=1)
# Copy files
for (filename in subimg[subimg$Auswahl == TRUE, "datei"]) {
f1 <- sprintf("%s/%s", indir, filename)
f2 <- sprintf("%s/%s", outdir, filename)
if (!file.exists(f2)) file.copy(f1, f2)
}
# Loop: Select n with lowest distance to sequentially last selected of previous block.
for (packi in 1:(nb_blocks-1)) {
seldays <- days[(packi*blocksize+1):(packi*blocksize+blocksize)]
selimgs <- subimg[subimg$date %in% seldays,]
selimgs <- selimgs[selimgs$filesize >= mean(selimgs$filesize),] # optional
if (sum(selimgs$Auswahl, na.rm=TRUE) > 0) {
last_img <- tail(selimgs[!is.na(selimgs$Auswahl), "pfad"], n=1)
next
}
writeLines(sprintf("%d/%d: %d images", packi, nb_blocks-1, nrow(selimgs)))
for (i in 1:nrow(selimgs)) {
lcr <- .C("lumi_compare", filename1=last_img, filename2=selimgs[i, "pfad"], dist=.Machine$integer.max)
selimgs[i, "distance"] <- lcr$dist
#writeLines(sprintf("%d/%d -> %d/%d: %f", packi, nb_blocks-1, i, nrow(selimgs), selimgs[i, "distance"]))
}
selimgs <- selimgs[order(selimgs$distance),]
top_imgs <- selimgs[1:imgs_per_block, "unixtime"]
subimg[subimg$unixtime %in% top_imgs, "Auswahl"] <- TRUE
last_img <- tail(subimg[!is.na(subimg$Auswahl), "pfad"], n=1)
# Copy files
for (filename in subimg[subimg$Auswahl == TRUE, "datei"]) {
f1 <- sprintf("%s/%s", indir, filename)
f2 <- sprintf("%s/%s", outdir, filename)
if (!file.exists(f2)) file.copy(f1, f2)
}
}
cmd <- sprintf("median_images.py %s 3", outdir)
writeLines(cmd)
system(cmd)
cmd <- sprintf("date_to_timelapse_monyear.py %s_median3", outdir)
writeLines(cmd)
system(cmd)
cmd <- sprintf("ffmpeg -hide_banner -loglevel panic -y -framerate 25 -pattern_type glob -i '%s_median3/*.jpg' -codec:v libx264 -crf 18 -profile:v main %s", outdir, video_output)
writeLines(cmd)
system(cmd)
# 1280x960 -> 960x540: 320 420 1280 960
# crop_images.py /mnt/big/nick/cams/lauterbrunnen 320 420 1280 960
# crop_images.py /mnt/big/nick/cams/scheidegg 160 150 1120 690