Wikipedia:Reference desk/Archives/Computing/2009 December 13

Computing desk
< December 12 << Nov | December | Jan >> December 14 >
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.


December 13

edit

Images being stretched when photo editing with basic programs

edit

I downloaded the high-resolution TIFF of this image visible here, as I have done with dozens or hundreds of images, and for the first time used the new Windows Live Photo Gallery to crop it and Paint to convert it into .jpg; the resulting image I uploaded as File:Spring Lake Bridge.jpg. As you can see here, the resulting image is much more of a square than the original. Any idea why this happened? Before Windows made me download Windows Live, I had been using the Microsoft Office Picture Manager for cropping, so I redownloaded, cropped with that program, converted with Paint, and uploaded the current version of the image, which you can see fits the original far better. What went wrong? Does Windows Live simply stretch the image, or did I do something I shouldn't have done? Nyttend (talk) 00:59, 13 December 2009 (UTC)[reply]

This is not at all what you asked, but for what it's worth, I use and recommend Irfanview for this sort of thing; it is free and gives you a lot more control over this stuff than what you're using now. Tempshill (talk) 02:30, 13 December 2009 (UTC)[reply]
Well, since you got us off onto a tangent, I'll follow it :-) I find Picture Manager quite simple — I find the Crop tool, cut off the edges, and save, and then open it in Paint and immediately save as a .jpg. How would Irfanview help? Nyttend (talk) 02:48, 13 December 2009 (UTC)[reply]
Sounds faster, because you don't have to use 2 apps to do a single simple crop. And it doesn't change the aspect ratio of your pictures randomly! Tempshill (talk) 06:14, 13 December 2009 (UTC)[reply]
While I personally would prefer to use something other then Microsoft Office Picture Manager (I have bad experience with MS stuff of this sort as they tend to be oversimplified so don't provide the versatility I enjoy and do stuff without telling you) as I understand it the aspect ratio thing only occurs with Windows Live Photo Gallery Nil Einne (talk) 06:22, 13 December 2009 (UTC)[reply]
The problem is that the initial image has a Pixel Aspect Ratio setting embedded in it which is unusual (0.802), whereas to display "properly", it needs to be set to something else (1.0). Now why it ever got set this way in the first place, I have no idea, probably has to do with the device that created it (though now that I think of it, it is not uncommon for TIFFs in general to have this problem). The result is that unless you have a program that lets you edit such a thing, it'll come out squarish. Photoshop can handle this setting and even announces to you when you open the file that something is wonky. But I imagine that lots of other "simpler" programs either just ignore it or mistranslate it or whatever (it looks like they take the property seriously, but don't give you any option to change it, whereas it would actually display perfectly fine if they just ignored the property and treated each pixel as equal in size). --Mr.98 (talk) 15:18, 13 December 2009 (UTC)[reply]
Thanks for the input! I'm the type of person who would rather use a familiar program instead of a new one that has only slight benefits, so I'd rather continue using Picture Manager and Paint. Two questions from this — (1) Does converting tiff to jpg in Paint result in a much smaller file (because of compression-related issues) than it would if I saved it in Irfanview? If so, Irfanview might appear to me to have more than slight benefits. (2) Is there any way to uninstall Windows Live other than attempting to delete all Live-related programs from my hard drive? I installed it because I thought it was just another Windows update, not one that would make it harder for me to use the programs I want to use. Nyttend (talk) 02:50, 14 December 2009 (UTC)[reply]

Co-primality Algorithm for Many Large Numbers

edit

I'm normally inquiring at the mathematics desk, but I have a question on algorithms that perhaps someone here may be better equipped to tackle. If you are choosing large numbers sequentially so that each new one must be relatively prime to all of its predecessors, what is a good algorithm for determining each succeeding term's suitability for this? It's apropos of determining a largest possible number created by adjoining a digit to the left each time, and the problem is just barely tractable if one limits the field from the start to composite numbers. My sense is that one neeeds to maintain a list of small prime factors that have been used already (by adding to and removing from as one goes up and down the full tree of numbers), check these with each term added to the list after some suitable period of using standard algorithms (each time a digit is added in the specific case at hand), and--as is also best done with much primality testing--checking the coprimality completely using the standard Euclidean or Stein's algorithm only at the end with as many candidate sets as necessary.

I'd appreciate any good attempt at an answer, but of course something already in the literature would be best. Thanks.Julzes (talk) 09:54, 13 December 2009 (UTC)[reply]

Have you reviewed the primality test article? There is no (currently known) more efficient way than the techniques described in those articles; proving primality (or even relative primality) with an efficient algorithm remains one of the most challenging unsolved problems in mathematics and computer science. For this reason, prime factorization is the fundamental basis for many cryptographic algorithms, detailed in the public-key cryptography article. For some literature references investigating fundamental complexity limitations,, take a look at the complexity section of the primality test algorithm. Nimur (talk) 21:53, 13 December 2009 (UTC)[reply]
He doesn't want primality, he wants co-primality, which is vastly different, and much, much easier to calculate.
May I ask, exactly how big are these numbers of yours? And how many are they? I just ran the GNU Multi-Precision Library gcd function on two 4096-bit numbers, and it took, like, a few milliseconds. I seriously doubt that you're going to find a method that is much faster than simply using gcd to check for co-primality with the rest of the numbers. Also remember that the expected value for the gcd between two large numbers is 1; you'll very rarely need to look again for a new number.
Also, unless you are doing this as a programming exercise or in general to learn, please don't implement the algorithms yourself. The people developing arbitrary-precision libraries are way smarter than you and me, and they've worked way harder than we will. For instance, while us mere mortals might think we are clever when using Binary GCD and thinking that this possibly couldn't be done much faster, these people use awesome stuff like Lehmer's algorithm and Subquadratic GCD. Seriously, if this is to be used for something real, use a proper library. Belisarius (talk) 23:04, 13 December 2009 (UTC)[reply]
Ah, that's a very important distinction - whoops. Sorry if my response was off on the wrong track. Nimur (talk) 16:33, 14 December 2009 (UTC)[reply]

Let me describe the sizes of the numbers involved. I just got the last plausible tractable result for the problem in which primes are allowed, which is the base-7 case. Odd numbered bases are a little easier, and base six is totally intractable. In base ten the answer to the base-seven question is 99 digits, just in as a result in the last half hour. There are a total 29735375 terminal nodes (maximal numbers of the type involved, in which there is no way to add a further digit.) The first interesting case in which primes are not allowed is base nine, and preliminary results are around, I'd say, 400 digits. I think the size of the overall tree may be manageable, pending some research. The person who said that I should only use standard algorithms is probably both partially right and partially wrong. I'm interested in a specialized question that might better be handled with deep research. It involves co-primality of increasing numbers of increasingly large numbers, and I don't think that the gcd algorithms are optimized for the specifics, though it may not matter. That's why I was interested in the question I specifically asked. If I write an optimal program, it might mean a difference between a few months and a decade. Never mind that it's a worthless question in real terms.Julzes (talk) 04:37, 14 December 2009 (UTC)[reply]

Oh, also, the problem in which base seven is complete is totally intractable for base six (well over 1500 base-ten digits and an obviously wide tree to get there). I call these the primality-allowed and -disallowed left-truncatable coprimality questions.Julzes (talk) 04:48, 14 December 2009 (UTC)[reply]

I guess the question I'm asking comes down to how much faster divisibility testing for small primes is compared with coprimality testing for large numbers (with a large number). I envision as an alternative to simple coprimality testing, sequentially flagging and checking every small prime (excluding those which cannot come into play--as in base 9 one simply is never going to get a number divisible by 3 with right digit chosen properly; and, further, 2 does not come up past the final digit, with the only allowable final digits being 4 or 8, an odd penultimate digit, and then only even digits; and further still, once 5 is on the list of factors the algorithm should be particularized so that it isn't ever really checked until travelling back down(up?) through the tree delists it). Thus, generating candidate answers, one only has to check these answers for coprimality rather than every node in the excluded part of the tree. Of course, if divisibility testing for many small numbers is as slow or slower than coprimality testing against many but a fraction fewer large numbers, then the simple route using the gcd function provided all the time is best.Julzes (talk) 05:02, 14 December 2009 (UTC)[reply]

Permissive Copyleft License

edit

I'm looking for a software license that is similar to the GPL but slightly more permissive in the way its share-alike clause is handled. (And I don't want to write my own if I can help it, though I'm willing to modify existing ones.) It should be free to distribute and modify, with the following clauses: The original work must be attributed to the author. The work may not be used for commercial purposes, with exceptions granted by permission of the author. Release of derived work requires notification of the author. All derived work must include this license, with the exception that it need not be free to modify. The author may not be used in promotion of derived work without permission. Attribution to the author must be removed following the request of the author.

Basically, the GPL (with a couple of CC and BSD restrictions) but with the provision that derived work does not need to be free to modify (though it still must be free to redistribute, and then only non-commercially barring permission from the author). Put another way, it's the cc-by-nc-sa license that allows derivatives licensed with cc-by-nc-nd. Does anyone know if there exists a license matching this criteria?

Also, as far as I'm concerned derived work and linked libraries are synonymous, so that rules out any LGPL-style licenses. Deshi no Shi (talk) 15:30, 13 December 2009 (UTC)[reply]

Well, the problem in writing our own clause is that your license is viral, meaning that license would need to be ported along forever with the code. Which doesn't work well unless everyone is using the same licenses (if everyone is on GPL, that makes the license permissive; if you are trying to mix GPL with CC and BSD, then licensing gets really complicated, and effectively the copyleft license becomes difficult to comply with). This is just a general warning about picking too obscure licenses, if you really want the licensed material to be used. My intuitive feel is "you're trying to micromanage this a bit too much"—either sign on wholly to a GPL-like framework, or sign on wholly to it being with proprietary or PD. But having excessively convoluted licensing terms is just a guarantee that they will either be ignored or the code will be ignored. --Mr.98 (talk) 15:51, 13 December 2009 (UTC)[reply]
Well, that's why ideally the license in its entirety /wouldn't/ be viral; there would be the terms the project is released under, and the terms that the modifications /can/ be released under. The goal is to allow derivative works with more or less restrictive licenses. Under the clauses that I have defined, any derived works /could/ use the GPL, or even more restrictive licenses should their authors choose. Are there no licenses which define two different criteria: that which a given work is released under, and that which modified works may be released under? Deshi no Shi (talk) 16:11, 13 December 2009 (UTC)[reply]
You can multi-license. But if your code is released with an extremely liberal license, somebody else cannot come and re-release it with GPL - that just doesn't make sense. They can re-release a project, including your code, and license any new portions that they have synthesized or included under GPL - but they can't extend that license to your work which is not GPL. If you really need help with licensing, you probably need to talk to an attorney who is familiar with software licensing in your area. Nimur (talk) 16:36, 13 December 2009 (UTC)[reply]
Well, if there is such a license, I haven't seen one. A NC license is already fairly problematic for most software development, but letting people release it as NC-ND... I'm just not sure there's ever been much of a market for that option. Also, there's kind of a snag in doing that—you allow them to release it NC-ND "barring permission" from the author to abrogate NC, but after a few derivative works, which author matters? All of them? You can see how this gets kind of problematic. Frankly, I think you're better off just saying "it's GPL, and if you need it otherwise, get in touch." That is actually far more flexible than trying to get weird custom licenses that do all sorts of things for all sorts of people. (Though of course GPL is not NC, really.) --Mr.98 (talk) 17:27, 13 December 2009 (UTC)[reply]
That's a pretty seriously limiting license - unless your software is pretty amazingly special - most honest people are going to run a mile from using it - and the dishonest people will use it no matter what you write. If you found someone infringing - you'd be on your own in defending it because none of the software rights people will be interested in defending a closed license like that. Refusing free redistribution and commercial access means that it's never going to be on a Linux distribution. If you're that concerned - just keep it private. You can't describe this as OpenSourced. SteveBaker (talk) 03:55, 14 December 2009 (UTC)[reply]
I could be wrong, but is this really any more restrictive than a FOSS license? You can't use GPL'd code in proprietary software, for instance (or am I mistaken?). And as far as I know, much of what's used on FOSS Linux distros, like Debian, is code that refuses "commercial access." And it's more permissive than any of the CCs (with the exception of cc-by and cc-by-nc). Deshi no Shi (talk) 06:00, 14 December 2009 (UTC)[reply]
You can make derivatives of most FLOS software without having to tell anyone (which may be problematic, what happens if you can't find the person? and unless the person gives permission you're going to have to tell them everytime you make a derivative work, heck there may even be some confusion over whether the same derivative work by a different person or hosted on a different site requires notificiation). And you can reuse their code in anything with a compatible license. For GPL since quite a lot of software is GPL this is quite a useful thing although many BSD developers hate GPL probably more then Microsoft does. And you can use their code commercially, you just have to release it under the appropriate license. Debian does refuse to have any non FLOS software in things they release but they do release their code under a FLOSS license (usually GPL) so anyone can re-use their code and release their software with whatever they wish provided they comply with the license including of course commercially. Nil Einne (talk) 07:46, 14 December 2009 (UTC)[reply]

lf2 network game help with hamachi

edit

well i hve learned how to connect n play lf2 using hamachi...but with only one opponent..how can i connect n play with several friends at a time...plz help —Preceding unsigned comment added by 59.184.156.28 (talk) 16:10, 13 December 2009 (UTC)[reply]

Ask all your friends to connect to the person (IP) waiting for players (with the same version of Little Fighter 2 of course)? Nil Einne (talk) 04:02, 14 December 2009 (UTC)[reply]
It was my understanding that Little Fighter 2 only supported head-to-head connections, with four people on each computer. I don't think it supports more than two computers at a time, unless there was an update recently that I am not aware of. --Coreycubed (talk) 21:47, 14 December 2009 (UTC)[reply]
You're probably right, I presumed the OP had already checked whether Little Fighter 2 supports more then 2 computers but don't know myself Nil Einne (talk) 12:56, 16 December 2009 (UTC)[reply]

nothing works...plz help

Windows 7

edit

What is the best way to put Windows 7 onto a PC without loosing the programs that are currently on the system. The PC is running XP home with 4 Gigs of Ram. BigDunc 17:41, 13 December 2009 (UTC)[reply]

Microsoft has an update analyzer which will see if your hardware is compatible, and whether you can install Windows 7 over the existing XP in a way that will preserve programs and data, or whether you will have to do a clean install, which will wipe out all your programs and data. In the latter case you will have to restore your data from the backup you make before you start, and reinstall all your software from the original disks (or re-download it, if that's how you got it). If you ordered commercial software as a download with no CD, and the time period when you're allowed to download it has expired, you're SOL.
Some people always do a clean install, because they don't trust the process of integrating the old programs with the new operating system, or they view it as an opportunity to get rid of a lot of crud that might have accumulated. --Jc3s5h (talk) 17:54, 13 December 2009 (UTC)[reply]
Here's advice from Microsoft on migrating from XP to Windows 7. This page has more general info on installing 7. —Preceding unsigned comment added by 82.41.11.134 (talk) 19:44, 13 December 2009 (UTC)[reply]

Network

edit

Hi, I'm trying to set up a home network (to connect this computer and a computer with a printer), but I'm having some problems. The computer I want to network to is XP, but this one is Vista. Help? 76.230.211.192 (talk) 17:52, 13 December 2009 (UTC)[reply]

Generally speaking, XP and Vista computers can coexist on the same network. --Jc3s5h (talk) 17:55, 13 December 2009 (UTC)[reply]
OK, so how do I set it up? ```` —Preceding unsigned comment added by 76.230.211.192 (talk) 19:17, 13 December 2009 (UTC)[reply]
Are you wanting wired or wireless networking? Do you have a router? You can find lots of information at these sites [1][2][3][4] —Preceding unsigned comment added by 82.41.11.134 (talk) 19:52, 13 December 2009 (UTC)[reply]
The ultimate decisive factor is, do you already have a router? If yes, tell us what type, so we can help you with it, or direct you to the user manual for it. If you do not already have a router, you have a few options. You can purchase one suitable for home use; the price ranges anywhere from $10 to $100, depending on your needs and location. Alternatively, you can set up internet connection sharing, using one computer as the server and sharing its network connection. This is more complicated and has certain limitations; it also requires certain hardware features (e.g. an extra ethernet port on the server computer). So, if you can answer the first question (do you have a router), we can help you proceed. Nimur (talk) 21:17, 13 December 2009 (UTC)[reply]
Something simple to check - make sure both networks have the same name - XP and Vista have different default names for their network! Exxolon (talk) 23:53, 13 December 2009 (UTC)[reply]

transfering files

edit

How do I go about moving stuff from a computer with XP to one with Vista? So far the only memory storage device my old computer has even recognised is the card from my camera, so I found everything I wanted to move and saved it all on that card. When I went to copy it onto my new computer, it had somehow scrambled every name into a random series of letters and symbols that no computer has been able to recognise even so far as to let me delete them, I have had to throw that card away and get a new one. I once again copied everything over, saved it on the card and moved it to my new computer, only to find that the entire card was empty, that nothing had saved on it. Returning to the old computer for one last try, I managed to carry a file across from one to the other, and my new computer refuses to recognise it, instead replacing every letter I had typed with randomness, looking as though someone has sat on the keyboard for a while. This time at least the file still reads perfectly on my old computer.

Is there anything else I can try, or will I simply have to copy everything over by hand? what if I were to connect a wire straight from one computer to the other?

80.47.153.175 (talk) 19:28, 13 December 2009 (UTC)[reply]

Assuming both are desktops, if you're reasonably confident the best thing to do is take your old hard drive out of your old PC and install it in your new PC. Then copy any files you want if you don't want to leave it in your new PC. Exxolon (talk) 23:52, 13 December 2009 (UTC)[reply]
The problems you are having, the randomly scrambled file names and the inability to read files on other computers, should not happen. It sounds to me like something in your old PC is messing with the file system on the memory cards. If it is malware, it is something you should get fixed before copying anything.
That said, how are you copying the files? You are presumably taking the card and putting it in a card reader connected to your old PC by a USB cable. When you do that, the card appears like another hard drive in XP and you should then be able to drag stuff from the PC's hard drive to the hard drive representing the card. And I presume you are using the "safely remove hardware" icon in the system tray (near the clock), before you remove the card from the reader. If you are using another method, perhaps some file copying software or maybe the camera software, then that could do all kinds of weird things. Astronaut (talk) 02:06, 14 December 2009 (UTC)[reply]
It kinda sounds like this could be an old-style short versus long filename thing? I have no clue why - but that's the kind of thing that gives you scrambled-looking filenames. SteveBaker (talk) 02:23, 15 December 2009 (UTC)[reply]

Network without ethernet

edit

I get internet through an ethernet cable. However I just got a laptop that doesn't have ethernet port, but I need to get the internet on it temporally to activate something. Is there a way I can connect the laptop to the desktop computer via USB and share internet? It has no other ports and I don't have a wireless card or router, or any access to them. The laptop is win xp, and desktop is win 7 —Preceding unsigned comment added by 82.43.88.54 (talk) 20:52, 13 December 2009 (UTC)[reply]

USB to ethernet adapters are available. --Jc3s5h (talk) 21:15, 13 December 2009 (UTC)[reply]
Presumably the laptop has WiFi - so you need to find someplace nearby that has free WiFi access - a coffee shop maybe? Some fast-food place? Maybe a hotel lobby. Go there - do your thing! WiFi routers are pretty cheap these days (Amazon has them for $30 for example) - you could get a USB ethernet interface for maybe half that - but having cable-free ethernet at home is worth having. SteveBaker (talk) 03:40, 14 December 2009 (UTC)[reply]
No it's an old laptop it does not have wifi. —Preceding unsigned comment added by 82.43.88.54 (talk) 14:27, 14 December 2009 (UTC)[reply]

So there's no way to just plug a usb cable into both computers and have them link via it? —Preceding unsigned comment added by 82.43.88.54 (talk) 16:41, 14 December 2009 (UTC)[reply]

Well, Jc3s5h had a suggestion. You could buy a USB-to-ethernet adapter, and plug one end into your connection and the other into your laptop. Thanks, 174.114.4.18 (talk) 21:14, 14 December 2009 (UTC)[reply]
Yes, but all I have is a usb cable and no money. I was hoping for a solution with my current equipment —Preceding unsigned comment added by 82.43.88.54 (talk) 22:44, 14 December 2009 (UTC)[reply]
Well, you're out of luck - USB definitely won't work. It's not a software matter - or a cable matter - it's hardware - so you can't fix it. The problem is that USB is not a 'symmetrical' protocol. Every USB device is either a USB host (ie a computer) or a USB slave (eg a mouse or a memory stick). Hosts can only talk to slaves - and slaves can only talk to hosts. Since both your PC and your laptop are USB hosts - they can't talk to each other via USB without some kind of an intermediary...which will probably cost you more than the USB internet contraption.
I guess if it's an old laptop, it might have a telephone jack and an onboard modem...if so, connect it up to your phone line and get a month of free-trial service from some dialup ISP or other. Maybe that'll get you onto the Internet for long enough to do what you need to get done.
Failing that - do you maybe have an old fashioned RS-232 serial port on both computers? With an RS232 "null modem" cable, you could make them talk and (possibly) you could get a "SLIP" connection working. If you are handy with a soldering iron, and both machines have a parallel printer port, you could make a PLIP connection cable (a "null printer" cable)...but we're getting into seriously antique technology here and it may be tough to find software for more modern OS's that'll do that.
SteveBaker (talk) 02:19, 15 December 2009 (UTC)[reply]
Thank you for the informative answer SteveBaker :) —Preceding unsigned comment added by 82.43.88.54 (talk) 14:43, 15 December 2009 (UTC)[reply]

Photo formats

edit

Which photo format (out of GIF, JPEG, PNG, TIFF, and "Device Independent Bitmap" (which I don't even know what is)) would have the highest quality for saving Microsoft PowerPoint slides as images for importation into Windows Movie Maker, considering that file size is not an issue? Thanks in advance, Ks0stm (TCG) 21:14, 13 December 2009 (UTC)[reply]
Please leave me a {{talkback|Wikipedia:Reference desk/Computing}} on my talk page when responding. Thanks, Ks0stm (TCG) 21:16, 13 December 2009 (UTC)[reply]

This will depend on the contents of your powerpoint slide. If you have mostly text, a format like GIF or PNG will give you the best tradeoff of size to fidelity. JPEG will often distort text, and depending on the quality of the export (related to the final file size), the text may look garbled or unreadable. TIFF has the capacity to store images without compression (or with lossless compression). PNG also has the ability to use lossless compression, depending on your software export tool. Consider also that if you import the images into Windows Movie Maker, and then and re-encode with a video codec, you may introduce additional loss - so consider using a lossless format for the intermediate stages and choose your video coding settings carefully. Nimur (talk) 21:23, 13 December 2009 (UTC)[reply]
As an aside, "device independent bitmap" is another name for BMP file format. This is an uncompressed format, so it should have suitable characteristics for your needs. Keep in mind, though, that just because the file format is lossless, your export toolchain may still introduce image distortion (this happens commonly, for example, by using a "thumbnailed" version of embedded images from a Powerpoint slide, instead of the full resolution image). Nimur (talk) 21:25, 13 December 2009 (UTC)[reply]
You probably want to use PNG, exported losslessly. It is a much easier format to deal with than TIFF, on the whole. --98.217.71.237 (talk) 21:37, 13 December 2009 (UTC)[reply]
I would export it either as an EMF or WMF file -- which are vector formats. Windows Movie Maker can import EMF files. GIF is limited to 256 colors. BMP and PNG are lossless. BMP does not compress the image format very much at all (aside from RLE) whereas PNG does compress it to a greater degree. JPEG is only used to decrease the file size of photographs. But, in my experience, PowerPoint exports slides at too low of a resolution. What I do instead is select the slide in the pane on the left and copy it, then open Adobe Illustrator and paste it. It is then a vector object that you can export from Illustrator at any resolution. I then export the slide as a high-resolution PNG. If you don't want to use Illustrator, then EMF would be my recommendation.--Drknkn (talk) 21:43, 13 December 2009 (UTC)[reply]
Where can I find Adobe Illustrator? I don't recall having that program on my computer... Ks0stm (TCG) 23:26, 13 December 2009 (UTC)[reply]
You can buy it. If you have not already bought it, it is probably not installed on your computer! --Andreas Rejbrand (talk) 23:33, 13 December 2009 (UTC)[reply]
Or download the trial version. But you can also export the slides as EMF vector files without Illustrator, if you prefer. EMF is one of the "Save as" formats, at least in my version of PowerPoint.--Drknkn (talk) 00:34, 14 December 2009 (UTC)[reply]
In my version of Powerpoint (2004, on Mac), you can set the resolution of exported slides to be arbitrarily high. File > Save As > PNG > Options > Dots Per Inch. Setting it to 96 dpi makes each slide 960 x 720 which is plenty good for most video -- upping it to 180 dpi makes it better than any hi-def format. I would skip the vector option (it isn't going to get you any higher than the minimum that the video format supports, which isn't terribly high). --Mr.98 (talk) 00:11, 14 December 2009 (UTC)[reply]
I use PowerPoint 2003. EMF images are smaller (in terms of file size) than PNGs and complete fidelity is assured until the movie is rendered.--Drknkn (talk) 00:34, 14 December 2009 (UTC)[reply]
Well, as he said, file size is not an issue. And I really don't see any advantage, especially if he doesn't have Illustrator. The rendered frame will be of the same quality either way. You can also export all the PNGs in one go, automatically, rather than copying and pasting. --Mr.98 (talk) 01:33, 14 December 2009 (UTC)[reply]
As I mentioned above, EMF files can be exported from PowerPoint without using Illustrator. Maybe the Macintosh version does not have the EMF option? It is in my version of PowerPoint. I mentioned that I copy and paste between PowerPoint and Illustrator and then export the image from Illustrator to Adobe Premiere as an EPS or PDF file because I can vouch for the effectiveness of that workflow with more certainty.--Drknkn (talk) 01:36, 14 December 2009 (UTC)[reply]