Thought I understood it, PPI/DPI

[Ryan_]

When in photoshop, I make a document 70" wide and 36" tall and set a PPI of 100 and add my image to enlarge it. Then I do the same thing except I make the PPI 300. When viewing each at 100% the 300ppi one looks terrible and the 100ppi one looks decent. The 300ppi file is much larger in MB as well.

I thought PPI was irrelevant for someone like me (just shooting photos and printing and only worried about amount of pixels in final image and quality of said pixels) and wasn't supposed to actually effect the file except for adding a tag to it storing the PPI number. Is the PPI in photoshop really DPI?

I use two printers (companies). One says don't enlarge just worry about the pixels and they'll set the appropriate DPI when it goes to print based on the print size requested (they say they'll go as low as ~100dpi give or take 10dpi to accommodate large prints if the pixels look good enough). However according to them my file (5D3) doesn't have enough pixels to get to 70inches, the DPI would be too low.

The other company says that "canvases are rendered at 300 dpi at the print size". And they said the image looks a bit soft when they set the file up for print.
But I've experimented in Photoshop and set up a document at 70x36inches at 100PPI. Enlarged the photo to fit the Photoshop canvas, did some sharpening and it looks decent enough especially if its gonna be printed on canvas and viewed from a few feet away. But I don't know if thats correct?

So I guess my big question is, is PPI in photoshop equivalent to DPI? I'm used to Lightroom where I just export the file at whatever pixels I want and thats that. Though I've never enlarged a photo in Lightroom or any program for that matter.

 

[Skirball]

How are you putting the image into the file? Are you changing the size of the image you put in there? Doing it properly, there should be no difference. Other than the 300ppi document will have a lot more boarder around it. It'll look like the photo in that document is smaller, but it's really that the document (when viewed on a monitor) is bigger, since it has more PPI.

PPI is irrelevant to you, unless you use a cheap print house that either doesn't know what to do, or refuses to change the size themselves. The safe path when dealing with people like this is to just send everything at 300 ppi. But just go into your image size and change it to the ppi (resolution) you want - just make sure to have "resample image" off! Then all you're changing is the width and height values, the image doesn't change.

 

[Ryan_]

How are you putting the image into the file? Are you changing the size of the image you put in there? Doing it properly, there should be no difference. Other than the 300ppi document will have a lot more boarder around it. It'll look like the photo in that document is smaller, but it's really that the document (when viewed on a monitor) is bigger, since it has more PPI.

PPI is irrelevant to you, unless you use a cheap print house that either doesn't know what to do, or refuses to change the size themselves. The safe path when dealing with people like this is to just send everything at 300 ppi. But just go into your image size and change it to the ppi (resolution) you want - just make sure to have "resample image" off! Then all you're changing is the width and height values, the image doesn't change.

Gotchya! Thats where I was getting confused then. I was just dragging the image into a blank document that was at 70x36inches. In one doc I set PPI to 100 and then enlarged the photo to fit the space. In the other I set the PPI to 300 (still 70x36inches) and enlarged the photo to fill the space. So that explains why it was looking crappy because it had so much more space to fill.

Thanks for clearing that up!

 

[Ryan_]

Ok its somewhat cleared up. But still lost.

Articles like this: http://laurashoe.com/2014/09/08/lightroom-myth-busting-when-resolution-matters-and-when-it-doesnt-and-how-to-set-it/ are making me more confused (no offense to author). Because it says PPI doesn't matter but then says it matters.

So when I export 2 of the same photos from lightroom, same pixel dimensions, but different resolutions (PPI), why am I getting different results? I never touched the PPI setting until now, and I'm only doing so because I'm trying to see what a 70" print might look like. Any advice?

 

[Marsu42]

So when I export 2 of the same photos from lightroom, same pixel dimensions, but different resolutions (PPI), why am I getting different results? I never touched the PPI setting until now, and I'm only doing so because I'm trying to see what a 70" print might look like. Any advice?

The x*y pixel dimensions and ppi are the connection between pure digital data and the real world (of printing).

If you just view the file on a monitor, ppi is usually ignored. But if you print, the ppi value tells you a) how large you can print w/o losing quality and/or b) how much data you need to export to get sufficient printing quality w/o having a unnecessarily large file.

Example for ppi use: Export a file with 300ppi for a desired physical size (in inc/cm) from Lightroom, drag it into a Indesign or similar which has an actual physical dimension as the page layout, and it will match both. Or know what print size you'll order, and enter this physical size plus the ppi value the print shop wants (usually 200 or 300 ppi).

Example for pixel dimensions use: You've got a monitor of 1600*1200 pixels, so enter these pixels into the Lightroom export for optimum viewing pleasure as one input pixel (of the shot) is one output pixel (on the screen) and the viewing software doesn't need to up/downscale.

 

[Ryan_]

So when I export 2 of the same photos from lightroom, same pixel dimensions, but different resolutions (PPI), why am I getting different results? I never touched the PPI setting until now, and I'm only doing so because I'm trying to see what a 70" print might look like. Any advice?

The x*y pixel dimensions and ppi are the connection between pure digital data and the real world (of printing).

If you just view the file on a monitor, ppi is usually ignored. But if you print, the ppi value tells you a) how large you can print w/o losing quality and/or b) how much data you need to export to get sufficient printing quality w/o having a unnecessarily large file.

Example for ppi use: Export a file with 300ppi for a desired physical size (in inc/cm) from Lightroom, drag it into a Indesign or similar which has an actual physical dimension as the page layout, and it will match both. Or know what print size you'll order, and enter this physical size plus the ppi value the print shop wants (usually 200 or 300 ppi).

Example for pixel dimensions use: You've got a monitor of 1600*1200 pixels, so enter these pixels into the Lightroom export for optimum viewing pleasure as one input pixel (of the shot) is one output pixel (on the screen) and the viewing software doesn't need to up/downscale.

Ok so first just to clear this up, I assume the PPI everyone is talking about is really DPI then?
And secondly, so are you saying that if I export my file at 70" length, and 300 "PPI", then the 100% view is, in theory, how it should look? Because it looks really bad when I do that. Which is kind of understandable.

However if I export 70" length, and 100 "PPI", it looks decent enough at 100%, especially if viewed at proper distance. One of my printers says they will go as low as 90-100dpi IF the pixels are high quality (low noise, sharp, in focus, etc), so that could work (although they say never enlarge the file) but I can't order custom sizes from them which I need for this order.
And the other printer says they print at 300dpi. And unfortunately the results Im getting wouldn't work that route. But I feel like that was just a blanket statement. I guess ill email them again and try to work this out, or maybe as the client if they will go down to 60" length which I'd feel more comfortable with.

Thank you for your input.

 

[Skirball]

My suggestion is to first not mess with the dimensions at all, let them be what they are. Stop dragging files in until you figure this out a bit better. Either 1) open up your image from Photoshop (File > Open) or 2) copy the image to clipboard and then go to "new document". If you do it this way the default document size will be the size of the image on the clipboard. Don't mess with it on import (it'll most likely be at 72 ppi, you need to change that in Preferences if it bothers you)!

Once you have your image open in Photoshop go to Image>Image Size. Make sure "resample image" is not checked and change the resolution to 300 ppi. The width and height will automatically recalculate. This is the maximum size you can print at that resolution without a loss of quality (100%). If you want to print smaller at 300 ppi, then turn on resample image (bicubic is fine), and set the size you want. Photoshop will downsample the image to fit the dimensions at 300 ppi. Save the document as a different name, and you have your file to send to the printer.

Not to start another LR vs PS war, but in my opinion Lightroom is far better at this process if you're doing it for more than an image or two. After I edit in Photoshop I save and it goes right back into LR. If I need files at a certain size and PPI for a printer I export copies to a separate folder and never mess with my originals. Just my $0.02 on the matter, in case you have access to LR.

 

[Ryan_]

My suggestion is to first not mess with the dimensions at all, let them be what they are. Stop dragging files in until you figure this out a bit better. Either 1) open up your image from Photoshop (File > Open) or 2) copy the image to clipboard and then go to "new document". If you do it this way the default document size will be the size of the image on the clipboard. Don't mess with it on import (it'll most likely be at 72 ppi, you need to change that in Preferences if it bothers you)!

Once you have your image open in Photoshop go to Image>Image Size. Make sure "resample image" is not checked and change the resolution to 300 ppi. The width and height will automatically recalculate. This is the maximum size you can print at that resolution without a loss of quality (100%). If you want to print smaller at 300 ppi, then turn on resample image (bicubic is fine), and set the size you want. Photoshop will downsample the image to fit the dimensions at 300 ppi. Save the document as a different name, and you have your file to send to the printer.

Not to start another LR vs PS war, but in my opinion Lightroom is far better at this process if you're doing it for more than an image or two. After I edit in Photoshop I save and it goes right back into LR. If I need files at a certain size and PPI for a printer I export copies to a separate folder and never mess with my originals. Just my $0.02 on the matter, in case you have access to LR.

Cheers yea I definitely have a bit of learning to do in this area, been reading everyday trying to grasp it better.
And 99% of the time I'm strictly using Lightroom. I love it. I only recently bit the bullet on Photoshop CC because I need it for work and occasionally it helps me out for my photos as well. Like in this case, photo is going to be printed as polyptych, so PS helps.

 

[Lurker]

Essentially P = DPI = PPI in that they are all referring to discrete data points. This is only part of the story, you also have to consider dimensions because the DPI and PPI have the extra per inch factor. Note that your original image dimensions are in pixels and not pixels per inch. DPI and PPI are the same thing just monitors tend to us pixels and printers use dots. Pixel is probably the more generic a term.

A 70x36 image at 300 DPI requires 21000x10800 discrete data points. (70*300 x 36*300)
A 70x36 image at 100 DPI requires 7000x3600 discrete data points.

Your camera provides 5760x3840 discrete data points (Pixels). You fall a bit short but are not too bad for a 100 DPI print. Way short of what is needed for 300 DPI prints.

There are 2 ways to make up the missing points.

You can simply stretch/compress the points you have to make them large/small enough to fill the desired area. So on the long side you're about 18% [ (7000-5760)/7000 ] short of points so each pixel has to fill 1.22 dots on the print. On the short side you're about 7% [ (3600-3840)/3600 ] long on points so each pixel can only fill .94 dots on the print. This isn't so bad but the image will be distorted some.

Do the same math for the 300 DPI prints and the distortion will be the same but now you introduce pixelation and color banding. On the long side each pixel the camera gave you has to cover over 3.5 dots on the printed image. On the short side each pixel 2.8 dots on the printed image. To be fair, these 3.5x2.8 dots cover the same area as the 1.22x.94 dots in the example above. When viewed in the final print at 5 feet you may not be able to tell the difference. But when sitting at a computer monitor with the image at 100% you certainly will.
(This is because 100% shows 1 pixel in the image on 1 pixel on the monitor so the 300 dpi print will cover 3x the area on the monitor relative to the 100 dpi print)

The other way to fill in the missing dots is to resample the image. The question is who does the resample. Since resampling in your case would make a huge file, it may be better for the print house to do the resample. This saves on disk storage and transmission times. Also, if you let them resample they can take the base image and resample for any image size needed. If you did it you'd have to send a file for 70x36, 8x10, 3x5, 4x6 images and on and on. If you want to print in house then you have to do the resample. Resampling keeps the pixel to dot ratio at 1:1 and uses more sophisticated math to generate the missing pixels rather than just stretching (or compressing) the pixels you already have.

To me it sounds like the first company is stretching/compressing the pixels you provide to fit the space. The second company is resampling your image to make up the dots so they can do 300dpi. The resampling maybe isn't that great so they get a soft image.

I have not done much printing because I've been confused about this stuff too and things never seemed to work out right. I think I've finally figured it out. Resampling maintains image quality if you're enlarging or shrinking the final print. (Think about printing a 3x5 image at 300 DPI. You only need 900 x 1500 points.)

Hope this makes sense. It's all about translating the pixels you have from the camera to the dots/pixels you need on the output device and the area that those dots cover.

 

[Ryan_]

Essentially P = DPI = PPI in that they are all referring to discrete data points. This is only part of the story, you also have to consider dimensions because the DPI and PPI have the extra per inch factor. Note that your original image dimensions are in pixels and not pixels per inch. DPI and PPI are the same thing just monitors tend to us pixels and printers use dots. Pixel is probably the more generic a term.

A 70x36 image at 300 DPI requires 21000x10800 discrete data points. (70*300 x 36*300)
A 70x36 image at 100 DPI requires 7000x3600 discrete data points.

Your camera provides 5760x3840 discrete data points (Pixels). You fall a bit short but are not too bad for a 100 DPI print. Way short of what is needed for 300 DPI prints.

There are 2 ways to make up the missing points.

You can simply stretch/compress the points you have to make them large/small enough to fill the desired area. So on the long side you're about 18% [ (7000-5760)/7000 ] short of points so each pixel has to fill 1.22 dots on the print. On the short side you're about 7% [ (3600-3840)/3600 ] long on points so each pixel can only fill .94 dots on the print. This isn't so bad but the image will be distorted some.

Do the same math for the 300 DPI prints and the distortion will be the same but now you introduce pixelation and color banding. On the long side each pixel the camera gave you has to cover over 3.5 dots on the printed image. On the short side each pixel 2.8 dots on the printed image. To be fair, these 3.5x2.8 dots cover the same area as the 1.22x.94 dots in the example above. When viewed in the final print at 5 feet you may not be able to tell the difference. But when sitting at a computer monitor with the image at 100% you certainly will.
(This is because 100% shows 1 pixel in the image on 1 pixel on the monitor so the 300 dpi print will cover 3x the area on the monitor relative to the 100 dpi print)

The other way to fill in the missing dots is to resample the image. The question is who does the resample. Since resampling in your case would make a huge file, it may be better for the print house to do the resample. This saves on disk storage and transmission times. Also, if you let them resample they can take the base image and resample for any image size needed. If you did it you'd have to send a file for 70x36, 8x10, 3x5, 4x6 images and on and on. If you want to print in house then you have to do the resample. Resampling keeps the pixel to dot ratio at 1:1 and uses more sophisticated math to generate the missing pixels rather than just stretching (or compressing) the pixels you already have.

To me it sounds like the first company is stretching/compressing the pixels you provide to fit the space. The second company is resampling your image to make up the dots so they can do 300dpi. The resampling maybe isn't that great so they get a soft image.

I have not done much printing because I've been confused about this stuff too and things never seemed to work out right. I think I've finally figured it out. Resampling maintains image quality if you're enlarging or shrinking the final print. (Think about printing a 3x5 image at 300 DPI. You only need 900 x 1500 points.)

Hope this makes sense. It's all about translating the pixels you have from the camera to the dots/pixels you need on the output device and the area that those dots cover.

Thank you so much for laying that out.
I read it, but I'm gonna have to read it 5 more times I think before I get my head around some of it. However this bit "To me it sounds like the first company is stretching/compressing the pixels you provide to fit the space. The second company is resampling your image to make up the dots so they can do 300dpi. The resampling maybe isn't that great so they get a soft image." has me wondering about the first company. Because they claim that they simply take the file we give them, and set the DPI when printing to accommodate the size ordered. They say they go as low as 100dpi if necessary. So their math example is like this: if you have a 4000x3000 image, then you can print up to 40x30inches.
So with that said, does that mean that there is a DPI setting on the printer/printer software and they just plug the appropriate DPI to match the size ordered? Because I've ordered a 60inch print from them before, and it came out very nice. It was on canvas, so I'm sure that helped. With your nose to it you could say it was a slight touch soft but honestly when viewed even at arms length it was beautiful, and the client was happy. So when the second company is telling me that a 72inch print will look like the photo attached (Capture.JPG), I'm confused. I realize that 72" is one foot wider than 60" so yes it technically won't be as crisp as 60". But the 60" print literally looked 100x better than the attachment. None of that pixelation/artifacts. I even emailed this printer back asking if they were sure thats how it'd look and they said yes, that they print at 300dpi for canvas no exceptions. And for reference, I've also attached a very close up crop of the photo (_U8A7932.jpg). I will say its certainly not the clearest, however I shot with L glass (on 5d3), heavy tripod, live view focusing, shutter delay, etc. FWIW, the unprocessed RAW is much cleaner and sharper, but still a bit of the jaggies on the tree branches. I feel I pushed it a bit too much in post but I still don't think its terrible. And don't mind the blurred/painter-ish reeds/water, thats from the 30 second exposure. Main subject is tree in center.

Hopefully I haven't totally confused anyone reading with all this. I'm just trying to get to the bottom of this and figure out if I can actually print this photo at 70". I may end up just telling client that 60" is the max because at least I know I can do that, and feel confident about it.

Anyways, Lurker, Marsu42, and Skirball, I appreciate you taking the time and offering your input/help/advice very much. Definitely helps. Cheers

 

[IMG_0001]

Hi,

First, 72'' over 60'' is a fifth larger and that means you need 44% (from 1.22) more pixels to maintain the resolution if you keep the aspect ratio constant. Put another way, it means that each of the pixels you have has to cover almost 50% more area at 72'' than it did before at 60''. Also, any noise, sharpening artifact, blur or other image deficiency is magnified by a similar amount, resulting in an image that looks like your sample.

Also, on stretching the pixels, the apparent sharpness may be better than on resampling, but the pixelisation may seem worst. That is because on resampling, pixels of intermediate colors fill the gaps were information is missing and this results in a blurry looking image. On the other hand, stretching the pixels won't look blurry, but might look blocky as the missing information is replaced by the color of existing pixels. That is the tradeoff.

I guess there is a reason why larger format cameras with higher resolution exist after all...

Regards,

 

[Ryan_]

Hi,

First, 72'' over 60'' is a fifth larger and that means you need 44% (from 1.22) more pixels to maintain the resolution if you keep the aspect ratio constant. Put another way, it means that each of the pixels you have has to cover almost 50% more area at 72'' than it did before at 60''. Also, any noise, sharpening artifact, blur or other image deficiency is magnified by a similar amount, resulting in an image that looks like your sample.

Also, on stretching the pixels, the apparent sharpness may be better than on resampling, but the pixelisation may seem worst. That is because on resampling, pixels of intermediate colors fill the gaps were information is missing and this results in a blurry looking image. On the other hand, stretching the pixels won't look blurry, but might look blocky as the missing information is replaced by the color of existing pixels. That is the tradeoff.

I guess there is a reason why larger format cameras with higher resolution exist after all...

Regards,

Wow thanks for pointing that out. Never really thought about it like that, I just looked at another foot longer as no big deal but according to the numbers, it is kind of a pretty big deal. So maybe their sample is accurate then? Still doesn't explain the print my friend had done of one of my images, printed to about 8ft long, and looked amazing. Now that I think about it though, I'm thinking maybe it was because of the content. It was shot with a 50mm 1.8, probably at 2.8ish, with main subject close to minimum focusing distance, so much of the photo was blurred out anyways, and this contrast perhaps gives the impression of a sharp image even at very large sizes? Well that may be part of it anyways, but the fact is even the in focus part looked great. Sure a touch soft up close but not even close to the sample that the lab sent me of a 72" print.
I will email the tent company who made the 8ft print again to try and get some more info on what they did and how they printed that big to keep it looking so good. They told me initially they print straight from the file if there is enough pixels, but I'm thinking there was more to it then that.

Thanks IMG_0001

 

[TexPhoto]

How about: What matters is original pixel count, don't change it. Select print size and print.

An image file in your computer is not 70 inches wide (or any inches wide) and is not 300 dpi or any dpi. 70 inches and 300 dpi is meta data, just like date, or GPS location, or camera model. An image file that is 600x 400 pixels is not changed when you se the size to 6 inches, or the dpi to 100. Now set both, and resample, and you are forcing a change that is the same as changing the pixel count.

When you resample an image to a different number of pixels, you may loose detail, but will never gain it. Less pixels (down sampling) is always less detail. More pixels (up sampling) is debatable depending on the technology, but in general, your going to loose a little detail or stay the same.

In general, if your image comes out of the camera at 5760x3840, and you want a 6 inch wide print, or 60, or 600, set the image size you want , "DON"T RESAMPLE!, and then print it. The computer and printer will handle the DPI all by themselves. Your RGB photo is going to be changed to a CYMK (or similar print technology) and the dpi will change during the printing. Changing the DPI yourself just changes it 2X, and there is no gain by doing this, only same results or loss.

 

[IMG_0001]

A few more things here,

About printing 8ft wide, it would mean that each of the pixels from the 5D3 sensor would grossly cover a little over 1/64th of an inch. Possibly, from a certain distance, which is necessary to look at a print that big, and if it was printed on canvas, it might do ok. Moreover, perceived sharpness is relative, so given that as you stated a lot of the image would be out of focus, the in focus parts might look sufficiently detailed.

The thing about your original image is that it has relatively deep DOF,. Yet the long exposure blurring the weeds, the leaves and the water have low details. Added to the low level of details in the horizon and the blue sky, it results in the fact that wath catches the eye are noise or sharpening artifacts. Particularly, I'm sure if you pixel peep the original RAW, you'll find that the blue share the same uneveness as the sample you provided.

Also, the water, the leaves and the weeds (and some branches probably) are not blurry enough to look 'deliberately blurry'. They just look like they are poorly rendered.

My opinion is that some images just scale better than others. This particular one might simply not be a good candidate for printing that large.

Best regards,

 

[Ryan_]

How about: What matters is original pixel count, don't change it. Select print size and print.

An image file in your computer is not 70 inches wide (or any inches wide) and is not 300 dpi or any dpi. 70 inches and 300 dpi is meta data, just like date, or GPS location, or camera model. An image file that is 600x 400 pixels is not changed when you se the size to 6 inches, or the dpi to 100. Now set both, and resample, and you are forcing a change that is the same as changing the pixel count.

When you resample an image to a different number of pixels, you may loose detail, but will never gain it. Less pixels (down sampling) is always less detail. More pixels (up sampling) is debatable depending on the technology, but in general, your going to loose a little detail or stay the same.

In general, if your image comes out of the camera at 5760x3840, and you want a 6 inch wide print, or 60, or 600, set the image size you want , "DON"T RESAMPLE!, and then print it. The computer and printer will handle the DPI all by themselves. Your RGB photo is going to be changed to a CYMK (or similar print technology) and the dpi will change during the printing. Changing the DPI yourself just changes it 2X, and there is no gain by doing this, only same results or loss.

Well that definitely makes it easier to get, by simplifying it and thats how I thought it was supposed to be. Because this is how the first printer does it. They say leave the file alone, all that matters is the amount of pixels, printer takes care of the rest. And they will print up to 60inches with this file but since I need to split it into 5 pieces, I need custom sizes and they might not be able to do that. Where the other printer who only does 300dpi, can do custom sizes but since they won't print at a lower dpi like say 100, the file looks pretty bad. Such an annoying dilema. Surely there are other printers out there but I don't want to experiment with a clients orders. These 2 printers have done me good until now.

Thanks for your input TexPhoto.

A few more things here,

About printing 8ft wide, it would mean that each of the pixels from the 5D3 sensor would grossly cover a little over 1/64th of an inch. Possibly, from a certain distance, which is necessary to look at a print that big, and if it was printed on canvas, it might do ok. Moreover, perceived sharpness is relative, so given that as you stated a lot of the image would be out of focus, the in focus parts might look sufficiently detailed.

The thing about your original image is that it has relatively deep DOF,. Yet the long exposure blurring the weeds, the leaves and the water have low details. Added to the low level of details in the horizon and the blue sky, it results in the fact that wath catches the eye are noise or sharpening artifacts. Particularly, I'm sure if you pixel peep the original RAW, you'll find that the blue share the same uneveness as the sample you provided.

Also, the water, the leaves and the weeds (and some branches probably) are not blurry enough to look 'deliberately blurry'. They just look like they are poorly rendered.

My opinion is that some images just scale better than others. This particular one might simply not be a good candidate for printing that large.

Best regards,

I agree with you about the reeds and water. And that might be a problem in a print this big.
I'm thinking maybe as you said, this file just wasn't meant for the big screen. Maybe 60inches on canvas but may even be pushing it then. I'll see what happens.

Thanks again for your input IMG_0001


Hopefully I can get this sorted ASAP and will try and update thread once I do. Appreciate all the help.

 

[kbmelb]

The confusion of DPI and PPI comes down from from printing & graphic design industry (I've worked in both).
It always confused the heck out of customers. DPI (dots per inch) referred to the printer line screen (mostly, theres more to it and it will confuse even more) and was relevant when doing color separations and going to plate for color press. Black and white it could determine gray scale/shadow depth and contrast when printing newspaper.

PPI is mistaking just because people were so use to saying DPI, but when talking sensor resolution PPI is what is actually meant.

Now when you have a print done, it comes of an ink jet or a dye sublimation. On the former, dots are random and so densely populated to the human eye they may as well be continuous tone. Dye sub, is pretty much continuous tone.

With theses type prints, especially back in the day, PPI is important because under 200 PPI the human eye can start to detect the image is made up of millions of squares (pixels). 240ppi use to be the ink jet default since it was 20% over the threshold of 200dpi and it saved buffer time and RAM space. So with most printers and there supplied printer software, they'd essentially, through software, throw out the extra pixels if you actually sent a 240+ppi or higher image. Modern printers I'd imagine are faster and isn't an issue but as people we rarely accept change and will think everything is the same it was 10-15 years ago.

Now where 300ppi comes from is the print standard set in Europe some years ago of 12pixels/mm when you do the American conversion comes out to 304.8ppi. which for simplicity has been bumped down to 300ppi. I'd guess this was due to certain art mags and Nat Geo types they wanted everything to be very sharp (they even used higher line screens) so a 33% increase over the threshold made sure, to the most discernible eye, they'd look like photographs and not printed repros.

All that said, for prints 30" (widest) and under, if you can achieve 200ppi you will be good. You start to go over you can get away with a little due to the distance things need to be viewed.

I've seen breath taking prints from a 12MP 5DI blown up to 60". The particular print I' thinking of was simply sent to the printer and software handled the up res/scale.

Longer story short, if you are shooting with a 5D3 and haven't cropped away half of it you should be solid even at 70". If you printer knows what they are doing.

 

[Ryan_]

The confusion of DPI and PPI comes down from from printing & graphic design industry (I've worked in both).
It always confused the heck out of customers. DPI (dots per inch) referred to the printer line screen (mostly, theres more to it and it will confuse even more) and was relevant when doing color separations and going to plate for color press. Black and white it could determine gray scale/shadow depth and contrast when printing newspaper.

PPI is mistaking just because people were so use to saying DPI, but when talking sensor resolution PPI is what is actually meant.

Now when you have a print done, it comes of an ink jet or a dye sublimation. On the former, dots are random and so densely populated to the human eye they may as well be continuous tone. Dye sub, is pretty much continuous tone.

With theses type prints, especially back in the day, PPI is important because under 200 PPI the human eye can start to detect the image is made up of millions of squares (pixels). 240ppi use to be the ink jet default since it was 20% over the threshold of 200dpi and it saved buffer time and RAM space. So with most printers and there supplied printer software, they'd essentially, through software, throw out the extra pixels if you actually sent a 240+ppi or higher image. Modern printers I'd imagine are faster and isn't an issue but as people we rarely accept change and will think everything is the same it was 10-15 years ago.

Now where 300ppi comes from is the print standard set in Europe some years ago of 12pixels/mm when you do the American conversion comes out to 304.8ppi. which for simplicity has been bumped down to 300ppi. I'd guess this was due to certain art mags and Nat Geo types they wanted everything to be very sharp (they even used higher line screens) so a 33% increase over the threshold made sure, to the most discernible eye, they'd look like photographs and not printed repros.

All that said, for prints 30" (widest) and under, if you can achieve 200ppi you will be good. You start to go over you can get away with a little due to the distance things need to be viewed.

I've seen breath taking prints from a 12MP 5DI blown up to 60". The particular print I' thinking of was simply sent to the printer and software handled the up res/scale.

Longer story short, if you are shooting with a 5D3 and haven't cropped away half of it you should be solid even at 70". If you printer knows what they are doing.

Thanks for the input kbmelb! I'm glad so many have a say on this, its definitely helping me understand more.
So you mentioned I could be solid even at 70". Then what do you think of the 2 photos I posted (abound half way down/up the previous page of this thread)? The first photo being the digital sample of what a lab told me the photo would look like at 70", and the second being a very small center crop at 100% of the actual image. Do you think the sample is accurate? I've dealt with the lab before but only printed up to 30". However the quality was great and their customer service was just as great and I do feel they are reputable but best to get other opinions I suppose. As of now I still plan on telling client that 60" is max, but I'd love to run with 70" if possible. Can't wait to see what happens, so at least I will know for future requests.

 

[LDS]

When in photoshop, I make a document 70" wide and 36" tall and set a PPI of 100 and add my image to enlarge it. Then I do the same thing except I make the PPI 300. When viewing each at 100% the 300ppi one looks terrible and the 100ppi one looks decent. The 300ppi file is much larger in MB as well.

That's expected, because if you create a "blank" image and then put a photo into it, you have to manage yourself any resampling needed to show the image at the correct size and resolution. Simply enlarging it will just add pixels with the simplest interpolation algorithm (add new pixel identical to the neighboring ones), which will lead to ugly result, especially for large changes.

IMHO, creating a blank image is mostly useful for people using Photoshop to draw and paint from scratch, and being able to define the desired output size and resolution helps them to create an image with the desired specification for the selected output device, or to be used as a part of a more complex work.

With a photo, unless you wish to create something more complex, like an ad or magazine page, there's little need to create a blank image and paste an image into it, and even then a DTP application would manage this kind of project better than Photoshop.

If you're printing yourself, you can simply select the output size (and sometimes resolution, depending on the printer), and let the print engine apply the transformations needed to output the image at the desired size. Of course, if there are not enough pixels in the input image, and the process needs to resample and "create" a lot of new pixels, some quality loss is inevitable (and if an image is reduced too much, as well). When it becomes unacceptable depends on several factors.

If you have to send an image to an external service, and they won't perform the above step for you, or you want more control, you can just resize the image for the final output size and resolution, have Photoshop resample it with the proper algorithm, and then perform the final tuning for the output device (using the soft proof features), including sharpening. There are plug-ins like PixelGenius PhotoKit Sharpener which have pre-sets for different output devices, which simplify this step a lot. Then send the image for printing, or, if you need it, copy it into another image (at the same resolution) if needed.

Basically, as an image metadata both PPI and DPI settings tells the relationship between the source physical size and the capture/create medium resolution at the creation/capture stage, and can thereby be used later to understand when the image needs to be resampled to be shown at a different size and resolution on a different device. Otherwise devices with different resolutions would have no way to know how to properly transform it.

Because an image is made of pixels, not dots, PPI is the correct term. Photoshop uses correctly PPI when creating a blank image, because it is made of pixels, not dots.

Pixel and dots may not be interchangeable. Why? Because usually a pixel can represent any color value for a given color depth and color space. A dot may not, depending on the output technology (usually dots are used by output devices only). In many printing technologies, like inkjet printers using four or more colors, each dot can't represent the whole color range of a single pixel. Thereby, more dots in a dithering pattern are needed to "fool" the eye and look like the pixel they have to display.

The printer process will analyze pixel data, and turn them into the required dot pattern. The more dots per inch a device can output, the more complex and refined the dithering pattern can be. Once there was a rule of thumb that the input resolution was OK at 1/3 of the output resolution, so an image set at 240ppi (for a given size) would print well on a 720dpi printer (for the same size). Today photographic printers are far more capable.

Other output technologies may deliver more colors for a single dot, and thereby require less dots for pixels. Thereby comparing the quality of an output device looking only at the DPI value regardless of the technology is useless.

Input sensors capturing image in pixels have also a PPI value (often incorrectly labeled DPI) defining the sensor resolution, but it is usually useful only for devices like scanners, because they can know the source size, and thereby from the image pixel size and the stored PPI metadata for an image you can compute the original image size (it's pixels/PPI), and when the resolution can be set, it is useful to capture only the required number of pixels for a given output device, to reduce the captured image size (for storage, transmission, etc.), and avoid/reduce later resampling which can always alter somewhat an image.

For a camera, the sensor PPI value has little meaning. For example a 24MP APS-C sensor has an higher PPI than a 24MP full frame one, but at the same equivalent focal they will deliver more or less the same image at the same pixel size, the lens actually changes the captured image size, and usually you're more interested in this. There is also no strong relationship between the source image size and its final output size, the latter size is usually chosen to fulfill display needs only. You could still capture less pixels if only using low-res outputs, but because most cameras work better at the native resolution, it's what you mostly use, and let software resample the image later to achieve the desired size for a given output.

Screens resolution should also be defined in PPI, not DPI, but it looks a lot of developers has little knowledge of this. Also, when the only one interested in these data where graphic professional, they were more used to the older DPI term than the newer PPI which was specific to the newer electronic devices, once screen (and not only screens) resolutions were defined in "lines", not pixels.

Higher res screen again allow to display larger images without need of resampling while keeping the screen size not too large (just smaller pixels may be harder to see).
Higher res can be useful, for example, Lightroom asks you to perform sharpening at 1:1 because otherwise the image is resampled, at the changes applied makes more difficult to understand the "correct" sharpening settings. Just, on a screen, you can resample an image on the fly, and simulate different resolutions.

That's why also you should apply "input sharpening", to correct for the capture stage loss of details only, and "output sharpening" specific for a given output device (screen, different types of printers, etc.), to ensure it looks OK when the image is resampled for the final output.

 

[MrFotoFool]

Lots of good answers, some very in depth. I have worked at a photo lab for 20 years and will try to keep things simple. (I am not a super technical person anyway).

It sounds like you are creating a new document in Photoshop and dropping the photo in as a roundabout way of cropping to your desired dimensions of 36x70 (which are slightly more panoramic than the file the camera shot). The more straightforward way of doing this is use the cropping tool and set the crop dimensions to 36 inch tall by 70 inch wide (and leave the ppi box blank).

I think 60 inch wide is probably the maximum for a really sharp photo from 5D3. 70 inch is pushing it (unless you do a panoramic composite of two or three files). However, as you guessed, it might be ok if printed on canvas and viewed at a certain distance.

Here is what I do to get a completely accurate test, although it is easier for me because I can do it at work. Crop the photo to the desired size (36x70) as I stated. Then, reset the crop box to 8x10. Crop out an 8x10 section of the photo, using the rulers along the edge of the image as a guide. What I mean is, say you set the crop horizontal (10 wide and 8 tall), then just place the cursor to start the crop box at the ten inch mark of your 70 inch print and drag it down until the other end is at 20 inch. Once you let go, you can use the arrows on your keyboard to move the crop box to a position where it will show the best detail (since you will only have a small section of the photo). Now save that and print an 8x10 at the lab, which is cheap, and that print will show you exactly how sharp the final 36x70 will be. You do NOT have to print the 8x10 test on canvas, just on their standard photo paper.