Cameras are set up by the captwo program to produce five images in uncompressed (CR2 format) at each time step, with exposure times of 1ms, 8ms, 66.6…ms, 500ms, and 4000ms. Making a time-lapse video usually takes a lot time.Although the ultimate goal is stereo time lapse, an interesting first step is to create video files from a single camera view. Combine image sequences with different frame rates into one video file with our best time-lapse program yet. Apply vignette, tilt focus and other effects instantly. Time-Lapse Tool’s amazing variety of effects make it uniquely easy to use for creating stunning videos.I did begin to implement my own but it was taking too long, so took the (slightly ugly) shortcut of using the publicly available ImageMagick executable to convert to PNG first.SofortBild is a free tethering software for Mac users working with Nikon cameras. I searched for a CR2 loader written in Python and couldn’t find one. (Matlab may also have been good but I don’t have that installed at the moment). So, open Time Lapse Assembler and well pick the folder that contains all the photos from our hyperlapse sequence and then well set our Codec to mp4, Framerate of 24 frames per second.Some image processing clearly is required to combine the multiple exposures into a single image and for this I wanted to use Python and NumPy for convenience of implementation.
![]() Program For Doing Time Lapse Photography On A Full Listing OfAlthough captwo puts all images into a single directory, I had manually moved each month’s images into its own directory on the TS-109, since this makes it easier to browse over the network. The code assumes that CR2 images are organised into sub-directories by month, so that for example all of February 2013 would be in sub-directory 201302. Python for that looks like: brightness = (image.sum() / (720*480)) / durationImage_mean = (image * np.tile(w0/w_norm, (480, 720, 3))) + (image * np.tile(w1/w_norm, (480, 720, 3)))Image_mean = np.clip(image_mean, 0.0, 1.0)A full listing of Python code for producing a PNG sequence is here.0.00-1.00 sec 25.0 MBytes 209 Mbits/sec 1.00-2.00 sec 16.9 MBytes 142 Mbits/sec 2.00-3.00 sec 17.0 MBytes 143 Mbits/sec 3.00-4.16 sec 10.2 MBytes 74.3 Mbits/sec 4.16-5.26 sec 9.25 MBytes 70.3 Mbits/sec 5.26-6.17 sec 4.75 MBytes 43.9 Mbits/sec 6.17-7.28 sec 8.75 MBytes 66.0 Mbits/sec 7.28-8.29 sec 6.25 MBytes 52.1 Mbits/sec 8.29-9.00 sec 7.38 MBytes 86.5 Mbits/sec 9.00-10.30 sec 11.6 MBytes 75.0 Mbits/sec 0.00-10.30 sec 117 MBytes 95.4 Mbits/sec sender 0.00-10.30 sec 117 MBytes 95.3 Mbits/sec receiverHowever, this is with data being sent from the Mac to the TS-109. The results from this look ok. Tar xvf iperf3.tarThen run a server on TS-109 using: src/iperf3 -s4(The -s means configure as a server and -4 means use only IP4 not IP6 as the TS-109 isn’t set up for IP6)And attempt a client connection from a Mac using: src/iperf3 -4 -c my_ts109_name_or_ip(Again -4 means to use only IP4). Suitable command looks like:Ffmpeg -i seq.%%04d.png -r 24 -b:5000k video video_year_daily_0800.mp4Still having problems with downloads of images over the local network from TS-109 to a Mac or PC (and having swapped network cables a few times in case it was that) I decided to try some diagnostics.Downloaded iperf3 from: and it seems to build quite nicely on both OSX and ARM (on the TS-109). But if I connected a single cable from Mac to the NAS and set fixed IP’s on both then all works nicely, as follows: 0.00-1.00 sec 22.0 MBytes 184 Mbits/sec 1.00-2.00 sec 21.6 MBytes 181 Mbits/sec 2.00-3.00 sec 21.6 MBytes 182 Mbits/sec 3.00-4.00 sec 21.1 MBytes 176 Mbits/sec 4.00-5.00 sec 21.8 MBytes 183 Mbits/sec 5.00-6.00 sec 21.8 MBytes 182 Mbits/sec 6.00-7.01 sec 21.6 MBytes 181 Mbits/sec 7.01-8.00 sec 21.6 MBytes 182 Mbits/sec 8.00-9.00 sec 21.8 MBytes 182 Mbits/sec 9.00-10.00 sec 21.6 MBytes 182 Mbits/sec 0.00-10.00 sec 218 MBytes 182 Mbits/sec receiver 0.00-10.00 sec 216 MBytes 182 Mbits/sec receiverNow to copy the images and make some interesting time lapse sequences…It wasn’t too difficult to look at images from the NAS box when there were only a few days worth of data, as I was using the public folder which is automatically shared on the network using samba. Now just need to find out what that might be……and in the end it turned out to be something to do with cabling or my network switch – still haven’t figured out which. 0.00-3.36 sec 384 KBytes 937 Kbits/sec 3.36-3.36 sec 0.00 Bytes 0.00 bits/sec 3.36-7.69 sec 256 KBytes 484 Kbits/sec 7.69-7.69 sec 0.00 Bytes 0.00 bits/sec 7.69-9.37 sec 256 KBytes 1.25 Mbits/sec 9.37-9.37 sec 0.00 Bytes 0.00 bits/sec 9.37-11.05 sec 256 KBytes 1.25 Mbits/sec 0.00-11.05 sec 1.25 MBytes 949 Kbits/sec receiver 0.00-11.05 sec 1.12 MBytes 854 Kbits/sec receiverSo there’s something seriously wrong with the data rates obtained when the TS-109 is asked to send data to the Mac. The roles can be reversed using the -R parameter: src/iperf3 -4R -c my_ts109_nameWhich gives disappointing and revealing results. Download torrent windows 7 professional 64 bit ita iso crackSo I followed instructions at. Ones which hadn’t already been downloaded) using -ignore-existing only seemed to make things even slower, perhaps because querying what’s already downloaded is not done efficiently in the case that there are tens of thousands of files in one directory.I was wondering if NFS could be used to improve things. Also, any attempts to use rsync to update with only the newly taken images (i.e. The z option means to compress on the box before sending, though this didn’t make much difference in speed, and I’m guessing this is because what was gained in network bandwidth was lost in compression time.But it does seem like this does seem overly slow for the amount of data – after all there must be considerable overhead in using the ssh transport layer. I was able to do an initial download after a few months of images by using rsync with the following command typed into an X11 terminal on a Mac: rsync -avz -e ssh /Users/OneYearTimeLapse/Data/I left the command running overnight to do this. Even getting a Mac Finder or Windows Explorer browser window to list this number of files seems to take many minutes and was occasionally failing, let alone trying to download. Keeping the delay between multiple exposures as low as possible is desirable so that not too much movement occurs in the scene (people, traffic, trees etc) and so multiple exposures can be combined without ghosting effects.In the end the best I could do using the Canon 350D’s was 2. This was mainly because I was trying (in vain) to find a way to capture only to the camera’s SDRAM rather than memory card, so that there would be minimal delay between exposures at the different shutter speeds. The network light on the box is busily flashing so hopefully it will indeed build a nice file list and download everything swiftly…let’s see how long it all takes…It took about18 hours of coding and testing to choose an appropriate software implementation to support hardware sync. And that was ten minutes ago.
0 Comments
Leave a Reply. |
AuthorAmber ArchivesCategories |