Wikipedia:Reference desk/Archives/Computing/2012 April 2

From Wikipedia, the free encyclopedia
Jump to: navigation, search
Computing desk
< April 1 << Mar | April | May >> April 3 >
Welcome to the Wikipedia Computing Reference Desk Archives
The page you are currently viewing is an archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.

April 2[edit]

MP3 player to PC USB cable[edit]

What is that bulb on the cable? Thanks. (talk) 08:34, 2 April 2012 (UTC)

Ferrite bead. -- Finlay McWalterTalk 08:57, 2 April 2012 (UTC)

Extracting hi-res image from Adobe Flash[edit]

Is there a way to extract a full-resolution image file from Adobe Flash viewer like this one (at the maximum magnification offered by zoom)? (talk) 11:35, 2 April 2012 (UTC)

In this specific case, click the icon labelled 'Save Image' at the top-right of the picture, then right-click on the image that appears and save the image. (On my particular combination of browser and operating system it's called 'Save Image As...') This produces a file named NSARM200715624.jpg which is at the highest resolution possible for that picture. - Cucumber Mike (talk) 11:47, 2 April 2012 (UTC)
That method seems to produce a pic at much lower resolution than it can be displayed. One option is picking the "Print" option above "Save Image", and check "Print to file" from the panel. Unfortunately, this seems to lock up my computer, so I can't tell what res it produces. This option may not exist on your computer. StuRat (talk) 17:54, 2 April 2012 (UTC)
Another option, for any image which fits on the screen all at once, is to do a screen grab. You may need to set the monitor's resolution to max to fit a large pic. Assuming you are on a Windows operating system, next hit the Print Screen button on your keyboard, above the Insert and Delete buttons (might be different on a laptop). Go to an application like Microsoft Paint, then do Edit + Paste. Some editing of the pic is often needed after, to remove ads and such from the edges. You can then save the image in the desired format using File + Save. StuRat (talk) 16:43, 2 April 2012 (UTC)
I just took a look at your example, and, since zooming only lets you view it through a tiny window, this method wouldn't work very well in your case. However, zooming in on the page increases the size of the frame (View + Zoom + Zoom In, repeatedly, in Firefox). Then, using the + at the bottom of the pic, repeatedly, will increase the image size within this frame. This will allow you to capture the entire pic at moderate resolution, or the pic at full res as a series of panes, after panning to show each portion at max res. StuRat (talk) 16:45, 2 April 2012 (UTC)
And you could do it programmatically fairly easily via (where zoomifyInitialZoom=100 is 100% zoomed in, and the X and Y values are exactly what you'd think they are). ¦ Reisio (talk) 19:20, 2 April 2012 (UTC)
Thanks, but I think we need a bit more. I copied that URL, and it gave me the image at 100% zoom, without the tiny frame around it. However, how do we save it from there ? A screen grab will take many panes, and require stitching together, as discussed below. The right click on the pic, which normally has an option to save the pic, doesn't, in this case, at least in Firefox. It does have an option to print, and, at least on my computer, this includes a "Print to file" option, but that locks up my computer, and I'm not sure if it would provided full res, in any case. The "File + Save Page As" option wants to save it as a SWF file, and it's size of 49KB indicates it's not full res (which I estimate around 5000×3000 pixels), although my computer doesn't know how to open an SWF file, in any case. StuRat (talk) 00:09, 3 April 2012 (UTC)
My apologies. I think I thought this was simpler than it was. In order to try and make up for it, I found a Python script called Dezoomify that will do what you want. Details about it on Commons are here, and the code itself is here. Unfortunately, having spent a little while faffing around with it, I can't actually make it work, but I think that's just my incompetence rather than a fault with the script. Maybe one of our Python wizards can help out if you get stuck. - Cucumber Mike (talk) 20:12, 2 April 2012 (UTC)
It looks like the Sourceforge version of Dezoomify will losslessly stitch the tiles together (using jpegtran), so it's not only easier than the alternatives but also avoids generational loss. I was able to get it working on Windows with Python 3.2.2 after a few changes:
  • It expects a file named jpegtran in the same directory as even on Windows, where it's named jpegtran.exe instead. Work around this by specifying -j jpegtran.exe on the command line.
  • Replace urlConcat(url, imagePath) with urllib.parse.urljoin(url, imagePath). (The author was apparently unaware of urllib.parse.urljoin and wrote a buggy replacement that doesn't work for this site.)
  • The stitching takes much longer than it should, so be prepared to wait.
But I now have a stitched version of that Titanic image on my hard drive, and it looks fine. -- BenRG (talk) 05:27, 3 April 2012 (UTC)
Could you upload it to Wikipedia or Wikicommons and link to it here ? (Presumably a 100 year old newspaper article is out of copyright protection by now.) StuRat (talk) 18:23, 3 April 2012 (UTC)

Stitch panes together to form large pic ?[edit]

Related to the above Q, is there a program that will accept multiple panes, find where they overlap, and stitch them all together to form a single, large pic ? StuRat (talk) 18:16, 2 April 2012 (UTC)

Hugin and a number of others. ¦ Reisio (talk) 18:25, 2 April 2012 (UTC)
Does it have an automatic option where it will find the overlaps and stitch them together without the user having to tell it which pane goes where ? (This should work with screen grabs, but not so well with separate photographs, as slight differences in scale, exposure, etc., would make the panes not match exactly.) StuRat (talk) 18:31, 2 April 2012 (UTC)
It does, though it also works surprisingly well with real photographs: it'll even extract properties of your lens from the way that the images fit together. Paul (Stansifer) 18:37, 2 April 2012 (UTC)
Excellent. Is it free ? StuRat (talk) 18:44, 2 April 2012 (UTC)
Hugin is free software. It is both free of cost to use, and its source code is available and free to modify. You can download it at - including its source and binary installation programs for common platforms. Nimur (talk) 19:01, 2 April 2012 (UTC)
Regarding your above comments: real photographs are stitched together by sophisticated software. Among the corrections necessary, exposure must be compensated (using an autoexposure algorithm); color must be matched (using white balance algorithm); and geometric distortions of the lens must be corrected (e.g., as explained in our barrel distortion correction mathematical overview). Exposure and color must be adjusted in pure software, as (obviously) the photo can not be re-shot with different shutter-speed, sensor gain, and so forth. If I recall, Hugin is able to stitch extended dynamic range images ("raw" files), but this requires quite a lot of technical knowledge and effort, because the user-interface is set up for "most common" usage workflows. Hugin's algorithmic implementation of each of these systems is quite complicated; there is a graphical utility to assist you with feature point extraction, and the photos are precisely aligned using cross-correlation. Because each photo may be exposed differently, colored differently, and geometrically distorted, this is actually quite mathematically intense and nontrivial; that is why Hugin's source-code is so much larger than, say, Image Magick, another tool that can stitch images together. Nimur (talk) 19:22, 2 April 2012 (UTC)
For the interested: Photometric alignment and vignetting correction. Nimur (talk) 19:44, 2 April 2012 (UTC)
What does ImageMagick do, as far as stitching ? StuRat (talk) 00:13, 3 April 2012 (UTC)
It rasterizes, appends, and re-encodes. See Appending Images. It can also composite, Composite Operator. Nimur (talk) 01:35, 3 April 2012 (UTC)
In this specific case, I recommend Dezoomify (see above). -- BenRG (talk) 05:27, 3 April 2012 (UTC)

How does UnoTelly work?[edit]

You can access US only content from outside the US, but it's not a VPN service. All you do is set their servers as your DNS server. I don't understand how that works. Anyone know? Thanks. - Akamad (talk) 22:47, 2 April 2012 (UTC)

It appears to be an open proxy. Providing a DNS server simply makes it a transparent proxy. Nimur (talk) 01:46, 3 April 2012 (UTC)
They must support only certain high-profile services like Netflix and Hulu. For those, they identify the servers that actually check the originating IP, allocate a separate proxy IP address for each one, and return that address through DNS. For all other servers, they return the real IP address. In at least some cases, the servers that actually send the video don't check the originating IP address, so don't need to be proxied. This would make UnoTelly's service a lot cheaper to provide than a universal proxy (and in practice probably faster). -- BenRG (talk) 03:28, 3 April 2012 (UTC)
Ahh yes, that all makes sense. They do claim faster speeds than VPNs, so they probably do only route certain traffic through their systems. Thanks Nimur and BenRG. - Akamad (talk) 04:14, 3 April 2012 (UTC)