thanks; does anyone disagree?

does anyone disagree?

Nope.
You can't make a silk purse from a sow's ear so don't bother trying.
The advice you've received is correct.

Mike

first dpi without a size in pixels or in cm/inch says quite nothing, because it says only something about resolution per inch.
if you have to rescan the picture depends on how big the image should be for printing
for example: if you scanned a A4 (29,7x21cm)size picture with 100dpi
and the print will be only 10x7cm or smaller, then you do not have to rescan the image, only change the size of the image to 300dpi, without re-sampling it.
but if you need the picture printed at the same size like the original, then you have to rescan it.
and if you need the picture printed bigger than the original, then you have to scan with even more dpi

A computer monitor is 72 dpi, so for viewing on a computer monitor at 100% resolution the 100 spi scans are great.

Upsampling for editing purposes is standard procedure.

Imagine a grid of square pixels on the 100 spi image, there would be 10,000 squares in a square inch to edit with.

Now imagine a 200 spi image. There would be 40,000 (4 times bigger) pixels to edit with.

In video a 640X480 image is 4 times bigger than a 320X240.

Scanning over 300 spi is a waste, and should be considered the maximum working size.

Budzr,

Upsampling for editing purposes is standard procedure.
Quadrupling the number of pixels does not increase detail. You could upsample to gigapixels and the detail would still be the same as the original scan.

Scanning over 300 spi is a waste, and should be considered the maximum working size.
300 spi scans are considered the minimum acceptable for printing these days, not the maximum as you state. Scans for high gloss photo printers are often 1200spi or better.

BTW, Windows default display is 96 ppi, not 72.

The OP asked for a "definate" (sp) answer. That one came in the first three responses. It would be helpful if you would stop posting misinformation and nonessential trivia.

Excuse my stupidity but I assume that means that a pixel is a pixel. That is all pixels are the same size (maybe the same area would be more accurate?) I know they can be different shapes, square 1:1 or rectangular1:1.33 etc but are they the same area. That wouold make sense with the explanations that I have read.
Wikipedia defines pixel as :<<In digital imaging, a pixel (or picture element[1]) is a single point in a raster image. The pixel is the smallest addressable screen element, it is the smallest unit of picture which can be controlled>>.
That didn't help me much.

Terry,
http://desktoppub.about.com/cs/intermediate/a/meas_resolution.htm

Source pixel size is dependent on the scanning resolution. At 100 dpi an pixel is 0.01" x 0.01". At 300 dpi a pixel is 0.0033" x 0.0033", or 1/3 the height and width and 1/9 the area.

In general though the dpi isn't anywhere near as critical as the overall pixel size of the image. Imagine scanning a postage stamp at 300dpi and getting an image that is 240x270 pixels. Compare that to scanning a movie poster at 50dpi and ending up with an image that is 1200x1800 pixels. Which has more detail, the 300dpi postage stamp scan or the 50dpi poster scan? If you answered "300dpi" then you don't get what resolution is all about.

And, in video, dpi is completely meaningless. It only has relevance to scanned & printed material.

Hey Musicvid, your comments have nothing to do with MY comments. I didn't SAY the detail would improve, I didn't SAY it was the maximum ideal for photo printing.

And what I DID say is absolutely 100% correct based on the advice asked for.

The purpose of upsampling is for EDITING! For adding more pixels and decreasing the number of colors in each pixel so selections and color corrections are more accurate.

Uhh, all of which has exacty what to do with the original question?
Oh, absolutely nothing.

Resolution has nothing to do with spi.

8X10, 5X7, 4X6, 16:9, 4:3 are all "resolutions".

100, 200, 300 are all samples per inch.

Computer screens nowadays are more likely have around a 96-100 dpi resolution, with many exceeding that.

When the next generation of OSes come around from Apple and Microsoft, we'll even get resolution-independent text where the font sizes are defined in points rather than pixels.

It is very possible to increase the apparent resolution of an image with a wide variety of upscaling methods, including even the most basic methods built-into Photoshop. Lots of plug-ins available to go beyond that, and they have worked very well for me over many years of use.

When working with video, we can even use interpolation across multiple frames to increase resolution, and this has produced amazing results that are surprisingly close to what's faked on TV shows.

Please get your terminology right.
100, 200, 300, etc. are DOTS per inch, not "samples".
8x10, 5x7, 4x6, etc. are PRINT SIZE, not "resolutions".
16:9, 4:3, etc. are ASPECT RATIOS, not "resolutions".

Mike

Resolution has nothing to do with spi.

You should make a resolution to take a deep breath (sorry).

8x10 is a size, and it could be for a photo printed from a small 72 dpi file, with of course gigantic pixels thanks to the blowup.

DPI is the common term, not spi, even if it is of course sampled.

CHECKPOINT PLEASE:

The original question quoted verbatim:

"If I scan a picture into photo shop at 300DPI vs having the same picture already scanned at 100DPI and then increasing it to 300 within photoshop, is there a difference in quality?"

The only definite answer is, "Yes. The one originally scanned at 100dpi will have lower quality, and the one originally scanned at 300dpi will have higher quality."

Since there was only one question asked, and clearly so, trying to be "helpful" by answering questions that were not asked, or introducing a bunch of irrelevant hypothetical scenarios, or by introducing a lot of extra information that is not applicable to the original question, is not only "not helpful," but it is confusing at best, and just plain wrong at its worst.

This is another example of a thread that should have ended by volition after four or five posts. All of the discussion that has come after should have gone into another thread, or unless the OP follows up or asks a collateral question here. After all, there exists an implied intention of the thread-starter to keep it reasonably on-topic and concise.

That being said, this is my last post in this discussion.

Any good teacher understands that knowing when to stop talking is more important than knowing when to talk.

BudWzr spewed: "A computer monitor is 72 dpi"

No, it isn't. It depends on the size of the monitor and the resolution at which the video card is set.

I realize that you think you know what you're talking about and that may be why you post things to the forum. (Well, that and just wanting to see your name.) But you rarely, if ever, know what you're talking about and it would be best if you just stopped posting.

What you're looking for is a photo uprez. Photoshop does this poorly but can access various plugins and export into external programs that do it quite well. An uprezzed picture isn't as good as a high resolution picture, but it's a heck of a lot better than a picture enlarged without uprezzing. Because uprezzing is such an important thing that is useful in many ways, I have spent quite a bit of time comparing the various options. By far the best (IMHO of course) is a program called Photozoom Pro 3. It doesn't work as a plugin to Photoshop, but you can export from Photoshop into Photozoom Pro which in practice is almost as good.

Photozoom Pro 3 is just a wonderful uprezzer that I can't recommend highly enough. I use it for all sorts of things like adding low rez photos to HD videos and making DVD covers and posters from HD screen grabs. It's not as good as a higher res photo, but it is a heck of a lot closer than many of you might think. It certainly is way better than anything else I've tried:

http://www.benvista.com/main/content/content.php?page=ourproducts&section=photozoompro_1

Reading over your first post again, I would rescan the photos at 300dpi rather than try to uprez them even with a good uprezzer like Photozoom. The uprezzer can save your butt when you don't have access to higher resolution scans, but nothing beats real resolution.

There was an old "Peanuts" cartoon where Lucy would tell Linus all sorts of things that were misleading or simply not true, and when Charlie Brown would overhear, his stomach would hurt. "Ow, ow, ow!" was usually the caption on the final panel.

Well after reading some of the BudWzr posts in this thread, my stomach is going "Ow, ow, ow," and I feel that need to respond.

As some of you know, I ran a desktop publishing company back in the 80's, so I had to learn all about dpi and scanning. Here is what you need to know, both for printing and for video editing.

First, the concept of dpi is only useful when printing. Forget about it for anything else. When you print, you need to output at least 75 dpi to get newspaper quality printing; about 150 dpi for magazines; and 300 dpi for high quality work. When I say "output," that is what has to go to the printer. It says nothing about what dpi is used for scanning, unless the scanned object is the same size as what you are printing.

When the scanning is being done for producing photographs using continuous tone technology (like old-fashioned chemical-based developing), you can see quality improvements at dpi levels beyond that.

Second, the reason my stomach hurts is that when scanning objects, the dpi you use also depends on the size of the object. For instance, if you scan a 35mm negative at 300 dpi, you will end up with a totally useless file that doesn't have enough detail for anything. Thus, to say that 300 dpi is the most you'll ever need is just plain wrong. This holds true when scanning prints because if you scan a wallet sized print (which isn't much bigger than a 35mm slide or negative) you'll likewise end up with something that isn't useful for much of anything if you only use 300dpi.

Like so many other engineering and technical problem, success often depends on choosing the correct perspective from which to view the problem. In this case, dpi is the wrong perspective from which to view scanning. The right way to think about it is pixels. For video, you start with how many pixels you need to match your project resolution (720x480, 1920x1440, etc.). You then decide how far you might want to zoom into the still photo you are scanning. You multiply that zoom factor by the project resolution, and that tells you how many pixels you need. Now, if your scanning software doesn't read out directly in pixels, but instead gives you only dpi, you simply divide the physical size of the object you are scanning into the pixel number you just derived, and that gives you the dpi. So, for example, if you have a 720x480 project, and need to zoom into the photo by a factor of two in each direction, then you will need to scan at a high enough resolution to create a 1440x960 pixel file. If your original is 5x3 inches, then you will need 288 dpi in the horizontal direction and 320 dpi in the vertical direction. Since I always want a little "padding," I would choose the next largest dpi setting, which on my scanner is 400 dpi.

I have written about this many times before in these forums. Here are links to other posts:

Scanner questions (for photographs)

What resolution do you scan images for HD?

Oh, and in direct answer to the original post: up-conversion creates nothing and is only used because it helps reduce staircasing artifacts. However, if you don't have enough detail in the original scan to read the license plate on the car, you still won't be able to read it after you do up-conversion (although sometimes it can make it a little easier to guess what the number on the plate might be). There are technologies which can intelligently guess on what detail might have been there if the image had been scanned at higher resolution, but in the end the only way to get that detail is to scan at the higher resolution in the first place.

And, in final answer to the original post, if you are starting with the usual 4x6 photos, and you want to print duplicates of those (4x6) then the short answer is that 300 dpi is the resolution you want to use, and the 100 dpi scans you already made should be thrown out.

How does that differ from my comment?

Using your logic I guess just about anything anyone says can be wrong. If they're scanning a piece of microfilm or a tse tse fly 300 spi wouldn't work either.

And they're not "dots", they're squares, and the terminology has been updated accordingly. It's not 1979 anymore.

It differs from your comment in that it's a studied and thoughtful response from a well-respected member of this community, who has contributed credibly and prodigiously for many years.

Had you written his exact words, there would still be a difference...

How does that differ from my comment? Let me be very specific on what differs, and what I consider to be incomplete, misleading, or simply incorrect.

A computer monitor is 72 dpi, so for viewing on a computer monitor at 100% resolution the 100 spi scans are great.

The fact that many monitors, including mine, are not 72 dpi has already been pointed out. The bigger issue is that saying 100 dpi (spi) scans are great is incomplete and misleading because, as I pointed out, unless you specify the size of the object being scanned, the dpi level has no meaning. A 100 dpi scan of a small object, like 35mm slide, would be useless for almost anything.

Upsampling for editing purposes is standard procedure.

As others have said, this is not correct. Upsampling is only done as a last resort, when a higher resolution version of the material is not available. No professional would intentionally scan at a lower resolution than needed, and then upscale, although it is true that a long time ago (the 1980s) images were routinely handled via proxy because computers didn't have enough RAM memory to handle a full-resolution image in memory all at once. Thus, we usually scanned a small resolution image -- or one was created for us by the software -- and that was used for editing. Then, just before printing or pre-press, the full resolution image was substituted, and all the editing commands were then applied to the full-res version. The last company I ran was Live Picture, and we had a technology called "FlashPix" that stored multiple resolution versions of the same photo in the same file, and then let the computer deal simultaneously with both the low and high-res versions.

Scanning over 300 spi is a waste, and should be considered the maximum working size.

This is actually the statement that caused me to take the time to post. I have already explained, in my post above, why this is incorrect: the dpi used depends on the object size, and 300 dpi is not even close to enough when scanning a small object.

The purpose of upsampling is for EDITING! For adding more pixels and decreasing the number of colors in each pixel so selections and color corrections are more accurate.

I am not sure how to classify this statement other than I am not sure what you were trying to say, but what you actually said doesn't make sense. Upsampling normally does not change "the number of colors" used to describe each pixel.

Resolution has nothing to do with spi.

Again, not really true, although perhaps you were trying to make some other point. The equation below shows that resolution and spi are directly related through a simple formula:

Resolution = Size * spi

Thus, if you scan a 3" x 5" image at 400 dpi, you will end up with

3 * 400 x 5 * 400 = 1200 x 2000 pixel image.


In summary, I am truly not trying to be too argumentative (although it is impossible to avoid this entirely), but I really do think you were probably misleading the person who originally posted, and that person might end up doing something wrong or suboptimal if that person based their actions solely on your posts.