raw dpi? 5d mkiii, quick Q

<p>Dear all, can some please tell me what the dpi is when shooting raw, straight out of the camera (5d mkiii). <br>

On my old 5d it's 240.</p>

<p>Thanks Nick</p>

<p>RAW images don't have a DPI. You select a DPI when you convert to JPEG.</p>

<p>ok sorry here's what I'm getting at. When I open a raw image from my 5d in camera raw, and then directly open it in ps, and check the image size the dpi is 240. Not sure why that it's, but it's always been that way. Is that true with the 5D mk iii ? Thanks</p>

<p>It's meaningless. The box has to have a number so they put one in it. There is no meaning to dpi until you are going to print.</p>

<p>The DPI is a printing variable, and can be changed. 240 DPI means if you start a print dialogue, say in Photoshop, it will start off with assumption you want to print 240 pixels horizontal by 240 pixels vertical, per square inch.</p>

<p>You could change the dpi value, higher or lower. If you do so and then say yes to "do you want to save changes" when exitting the program, you'll have saved the new DPI value. It doesn't change the pixel dimensions of the file, just that default print setting.</p>

<p>(I sincerely hope I haven't hung myself out to dry. I believe that's what it means, LOL.)</p>

<p>Don't know, but if you open it in DPP and save it as anything from an 8 bit JPEG to a 16 bit TIFF file, there's a box where you type in whatever DPI you want. The default setting appears to be 350dpi.</p>

<p>Maybe Photoshop imports with a default 240dpi setting? If so then it's Photoshop selecting the DPI, not the Camera or the RAW file.</p>

<p>This kind of thread has popped up quite a bit lately. Basically the camera itself has nothing to do with dpi because the camera only thinks in total pixels. DPI settings are entirely related to photoshop (or whatever program is used) and only need to be considered when printing. I'm sure there is probably a photoshop setting so that the default dpi when opening a photo is 240 or whatever but I've never really bothered because I don't bother with dpi for my print workflow unless I am working with a low res image and worried about how large I can make it without it looking horrible.</p>

<p>The camera does, in fact, embed DPI settings in JPEG files. They don't mean much, they can be changed at any time, but they are there as a result of the in-camera RAW to JPEG conversion. Rather than leave the field blank, they just put something in there. Could be 72dpi, could be 240dpi, could be 350dpi. It really doesn't matter</p>

<p>On the other hand, I don't think there is any embedded DPI info in RAW files because since you can't print a RAW file it would be meaningless. DPI is essentially a printing instruction (which many printer ignore anyway).</p>

<p>I ask for reasons related to printing from file to film recorder, and artifacting due to interpolation . Original DPI, as interpreted and laid in there by ps is meaningful to me, ok. And I do believe it's beyond commonly understood that dpi relates number and size of pixels to size image viewed, because it does, not for no reason. And once you start res-ing up or down, depending if you have checked or not checked the re sample box in >image size, some interpolation is happening to account for more or less dip, or your image is getting smaller as dpi goes up. And so, my question very simply asks what is coming directly from your camera raw into photoshop with your canon 5D mk iii. Who's got one? Is it 240? Anyone?!</p>

<blockquote>

<p>Original DPI, as interpreted and laid in there by ps is meaningful to me</p>

</blockquote>

<p> <br>

There is no such thing as "original DPI." DPI is the relationship between pixels and inches, period. It has nothing to do with the size viewed if it's on a screen. Screens display pixels. You have completely misunderstood this. There are no inches involved until you output to a device that measures in inches.</p>

<p>If they had called it "printing pixel ratio", the concept would have been accurate and not confused anyone. And it would be the same thing.</p>

<p>Raw files have pixels. How many pixels you print per inch is entirely up to your choice of print size.</p>

<p>Let's say that the file is 2000x3000 pixels, and assume that you don't resize it through interpolation. If you print it at 20x30 inches, you have printed it at 100 dpi.</p>

<p>If you print the file at 10x15 inches, you're printing at 200 dpi.</p>

<p>If you print at 40x60 inches, you're printing at 50 dpi.</p>

<p>The value that the camera put there by default is meaningless.</p>

<p>I have to wonder how the DPI value is perceived in the metric world.</p>

<p>ok, I got it. thanks .</p>

<blockquote>

<p> have to wonder how the DPI value is perceived in the metric world.</p>

</blockquote>

<p> <br>

They use DPKM. Dots per kilometer, keeps the number countable.</p>

<p>I think it is PS settings, not camera. Change your default DPI settings in PS to any number you want, just do not check resample box.</p>

"I have to wonder how the DPI value is perceived in the metric world."

 

Can't tell how it's like in a 'real metric' world, but down here in Mexico it works fine: we print 5x7s and 8x10s, so we keep

DPI (maybe because of the influence from our neighbor north of the border). On the other hand I can't tell what's an A4

size, which seems to be a 'real metric' printing size (and even my calculator doesn't includes it with the area conversions).

 

By the way, I always wondered if Canon played a metric-imperial units conversion joke when defining their standard APS-

C 'crop factor' to be the same as the 'mile-kilometer crop factor' (one mile is roughly the same as 1.6 kilometers, but I

would ask how we 'crop' miles to get kilometers... maybe zooming in Google Earth).

I need to add, regarding using imperial units for printing, that most people in Mexico usually convert the sizes to

centimeters to understand, we photographers simply assume they will get it because print sizes are supposed to be in

inches: Just remembered a discussion with my brother regarding a 20x24 being a 'good size' for a specific use according

to my sizing standards (inches), but 'way to small' to my brother's (centimeter) sizing standards. For a reference, a

20x24cm print would be a little bit smaller than an 8x10in (regardless of the DPI).

<p>Well, based on a 5760-pixel-wide sensor that's 36mm across, that's 4064 pixels per inch. That's the only figure which is defined by the hardware, and as such the only "raw" ppi value; as others have said, anything else in a JPEG file describes a print size, but only the arbitrary print size that's encoded in the JPEG (a field that I suspect a lot of us are now wishing hadn't ever been included in the image format, or most others), not any guarantee of how large you actually decide to print it.<br />

<br />

If you want to avoid resampling, you don't want the resolution of the camera - you want the resolution of the output device. For most printers that's usually meaningless (the resolution is very high, but the colour depth at each pixel is typically very low, so they rely on having more print resolution than the input image to support dithering; still, sharp edges and thin lines ought to get represented slightly better if you hit a multiple of the printer dpi). For a screen, a 1:1 pixel mapping means matching the PPI of the image to the PPI of the display - typically around 100, though somewhat higher if you're using something like a tablet or phone with a high-res display. Pretty much nothing would ever look at the PPI of the source image in order to do this, though.</p>

<p>so a dpi that's a multiple of that of the output device is key to printing. ok.</p>

<blockquote>

<p>have to wonder how the DPI value is perceived in the metric world.<br>

<br /> They use DPKM. Dots per kilometer, keeps the number countable.</p>

</blockquote>

<p>Jeff, I think you meant DPCM...dots per centimeter.</p>

<p>Millimeter, that's what I should have said.</p>

<p>this talk of mm leads me to where I'm concerned again, so here goes, though not sure this is the right forum.. I'm setting up an LVT film recorder cos I want to print b&w in my darkroom from digital files, and I got it for a song. The negs I'm so far are producing terrific 8x10 test prints, but I'm just getting it going, and there's much I don't understand. Obviously. The LVT resolution is now calibrated to give an exposure to neg resolution 60pixels per mm (res 60) or 1524 dpi. If I have a image file 6016 x 4016 pixels, I size it for a 4x5 neg (tmax 100) in ps, actually 3x4.5 is what it allows for w/that film size. Resampled is OFF. DIP then becomes 1336.88 dpi. Close to res60. I can up res to make it res60/1524 dpi or adjust image size a bit to make it exactly that- probably better. What I don't yet know is how large I can print. My darkroom accommodates paper only as large as 16x20. I'm also wondering at what point is an 8x10 neg preferable. What size original file, or print requirement would initiate that need? It would only be for contact printing in my case, but... there are some fantastic silver printers around still.</p>

<p>Nick: That may be over-stating it. :-) It's sort-of true, though it's typically not all that relevant.<br />

<br />

Let's have an example. Say you have an image that you've scaled before printing such that it's 300ppi. Should you print this on an ink jet at 600dpi, each image pixel corresponds to a square of 2x2 printer pixels (assuming everything lines up). Should you print this on an ink jet at 720dpi, each image pixel corresponds to a slightly larger area that doesn't perfectly correspond to pixel sizes. Therefore there's some interpolation happening, and the result will appear somewhat uneven.<br />

<br />

Unless you deliberately run your printer at low resolution when printing photographs (even then, the behaviour is questionable) or you're trying to print extremely high resolution images and you care about the result (for example, because there are repeated high frequency lines and you care about moiré that will be visible at range), this isn't likely to be such a problem in the days of high resolution printers. For example Epson's current range typically hits 5760x1440 dpi, and HP claim 9600x2400 dpi (from 1200dpi input) for some of their photo printers.<br />

<br />

With older printers, or non-photo printers it's more visible. I've seen at least one test that printed a black square with a 600ppi single-pixel white stripe through it. It printed beautifully on a 600dpi HP printer, because it aligned perfectly to the print resolution. It looked awful on a 720dpi Epson printer, because it was interpolated. The reviewer concluded that the Epson was less good at resolving detail.<br />

<br />

It's not usually such a major issue for photographic images because the pixels are typically reproducing a real-world signal that wasn't perfectly aligned to the camera sensor sites anyway. Therefore any smudging introduced by the printer will only be added to a degree of smudging that's happening as part of the capture process (the pixels can only reliably sample frequencies up to a point) - this is why viewing HD content on a 1366x768 HDTV only looks <i>quite</i> bad, not awful (sorry, pet peeve of mine).<br />

<br />

Matching the resolutions is a far more significant issue if you have computer-generated content that's aligned to the pixel grid, such that high frequency data there is genuine, not aliased. For example, if you add a text caption or line art, it can get visibly mangled by printers at even quite high resolutions - if I'm doing this, I always try to target the print resolution. The eye has very good resolving power with high contrast edges - certainly the steps in a curve printed at 600dpi on a monochrome laser printer can be quite visible on a close look. A moderately-low-resolution monochrome bitmap that's part of my employers' logo is visibly wonky on a print-out I have in front of me, due to "beating" between the print and image frequencies.<br />

<br />

If you're using a 5000+ ppi colour printer instead of a 600dpi laser, you're less likely to have problems with most images unless the interaction between the print resolution and the image detail that's visible at the macro scale. If you're aiming at an electronic device, you might want to be aware; depending on how you sample it, it's often a better idea to go straight from the full resolution image to the output, rather than using an intermediate shrunk version. I recently had to produce a 96-pixel-wide version of my employers' logo; I could have used a 100-pixel version, but the rescaling would have looked terrible. Fortunately I found a thousand-pixel-wide high res version to use as a starting point.<br />

<br />

[Note: There's historical advice for downsampling in stages. This is because the downsampling filters in some software are a bit primitive, and didn't fully sample all the source pixels "under" a destination pixel; shrinking incrementally ensured that all the source pixels were accounted for. If your software downsamples properly, this is probably a bad idea; if you only have bilinear sampling available, it's not such bad advice.]<br />

<br />

Anyway, I hope all that helps, and I've not contributed too much to the internet's wealth of misinformation.<br />

<br />

Jeff: I thought you were just being facetious! I enjoy converting between dpi and lppm for yuks and hijinks. We should measure focal length in microfurlongs.</p>

 

<p>Nick: Interesting. (Apologies for the cross-over; as you may imagine, I was typing.)<br />

<br />

I'm assuming that your LVT has analogue writing; some variants (I'm trusting Wikipedia here) seem to use LCD or LEDs to do faster writing but with a fixed native resolution. I presume there's still a limit to the precision of the light movement; whether the 60pix/mm corresponds to some multiple of the stepping technology's limits I couldn't say. It does sound like trying to aim exactly for whatever setting you select would be a good idea, especially if you're making enlargements from it.<br />

<br />

Good luck - it sounds like a fun toy!</p>