Comparing formats at output size

[Paul J]

Advertisement (gone after registration)

I thought I would ask this here as there are some well informed people that can help verify something for me which I believe to be true. It seems obvious but at a technical level I may be misjudging something.

 

Please let me state that to begin with this is for round-about comparisons, not scientific.

 

If on my 26" NEC monitor I resize a pic to 72dpi at 1150 pixels height, this gives me an image in physical visual form that is comparable in size to what the majority of my output it is. So for a horizontal image this is very close to the image that I will see printed double page in a magazine. For a portrait image it's very close to what I will see represented in a single full page magazine.

 

Now depending on what magazine it is and the quality of print I have to say in general it is VERY comparable to what I see in the magazine when I look at both side by side to compare. (ie. hold the mag up to or along side the monitor) The level of detail is closely represented within the printing process and it closely compares in tonality and colour too.

 

Does this sound right? The case being that my shots from a P65 at 1150px height and my shots from the M9 at 1150px height look really very comparable. I have work that requires much bigger output and this is when you really see the difference but at this size it seems there's really not much difference at all. I'm just wondering if this is something I can rely on to a degree, it really feels that way.

 

I really must set up a test and do some prints at that size to be sure but I'm guessing it's still a close comparison on the monitor.

[skinnfell]

...and the question is?

 

As far as I can understand from your post, you want us to agree that a P65 image and a leica m9 image looks roughly the same at 1150 pixels height. And yes, I DO agree that they might look similar, although I don't own a P65.

 

According to the interwebs these have both a Kodak sensor, and arguably excellent optics on both. That said, with careful sharpening and resizing, almost everything will look good at 1150 pix.

 

Many glossy magazines are printed at raster of 133 lines per inch, and most designers want a 2X quality factor, meaning that the delivered image should measure at least the equivalent of a full page spread at 266 dpi.

Lets say the magazine is 12 inches wide, then your image should measure at least 266x12= 3192 pixels the longest side. Which is way less than even an M8 performs.

 

(In reality, the 2x quality factor comes from the olden days of scanning, so especially with good lenses and no AA filter you can easily go down to a quality factor of 1.5x. The reason we still use it is because its easier to remember and allows for some cropping.)

 

DPI is a measure that only applies to printing (printed dots per inch)

 

That said, most monitors nowadays have a resolution of far more than 72ppi. Lets say you have a res of 1920x1200 on your 27" screen, that amounts to a horizontal length of roughly 23.5 inches. Provided the pixels are square, there are now (1920/23.5=) 81,7 pixels per inch. If you have higher resolution, then the PPI (what most people refer to as DPI) will be higher.

The rumored iPad3, will have a on-screen resolution of more than 260 PPI if the rumors are correct.

[Paul J]

Not wanting for anyone to agree with me for the sake of it. Hoping someone can give a better informed answer.

 

My question is: Is it reasonable comparison to make if you resize an image to appear the same size as the printed image when held side by side?

 

If I view an image 1150px height @72dpi and viewed in photoshop at 100% it looks the same size as the printed image when I hold the magazine up to the monitor. When I look at the two side by side what I see in the magazine and what I see on my monitor is really quite comparable. One can only tell so much from only shifting eyes back and forth between the two. There is slightly more detail in print but at the same time it's masked by the offset printing dots, appearing as grain.

 

Also can anyone give information on downsampling images? I hear it can be more destructive than up sampling but I don't see this specifically reflected in images from a P65 Downsized to 11x14 for example (resized at RAW output not in PS).

 

What I'm trying to arrive at is that the Leica M9 is actually, for me, the preferred format to use for these sizes for many reasons and the quality it seems is really very similar at this size. I've only had the M9 a month now and have yet to "jump of the cliff with the hang glider" so to speak and try it. I'm finally ready to try it out on assignments but just want to be armed with as much knowledge on the situation as possible.

 

In terms of what designers ask for, I've never had that issue and even though I always ask new clients if they have specific requirements they seem to be quite relaxed about it! Even for the more recognised magazines that I shoot for.

[skinnfell]

A monitor can show much more detail than printed images, because the printed picture is actually a mosaic of colored dots (yellow, cyan, magenta and black). These are easily visible through a loupe or a fast leica lens.

 

So your method of making the image roughly the same size on screen as on print is not a bad way of estimating what it would be on print. Of course color calibration is a separate issue.

 

Really high-quality printing is an art. So is proper upsampling (making the image larger than it is originally) and downsampling (making the image smaller).

However, Photoshop has come really far in this regard and downsampling is not really an issue anymore, in my opinion. But the more you downsample, the more you negate the benefit of the larger format.

 

A good way of downsampling is to find out how large the image is going to be on print at the desired resolution. For example if, like in my previous post the optimal picture size is,lets say 3900 pixels. In other words, you want to downsamle the image just enough so that it fits exactly in the page without further scaling. Being that adding/removing sharpness is an art, YOU want to take control over this process if you can.

 

So lets say you downsample the image to the 3900 pixels in my example, you will probably want to add a little sharpness. Keep in mind that some sharpness gets lost in printing, so you will need to add a bit more than what looks good on screen. But not so much that it turns into sandpaper. The grainier the image the less sharpness you want to add.

 

There is really no holy values to sharpening (unsharp mask in photoshop), but I find that a radius of well under 1 and a amount of 60-70% works well for M9 files.

 

I generally recommend no more or less than 2x upsampling in PS. In other words, Image size=200% This way, each pixel will get upsampled to exactly 4 pixels. PS now has a separate algorithm for upsampling. Its possible to go over that too, but then I would recommend a separate software for that.

 

A little trick I have employed is to add a touch of noise to the upsampled picture. This helps printing, since very uniform areas are hard to print accurately. It also smooths over areas of moire and antialiasing which might occur with upsampling.

 

Pictures with just a bit of grain print better than pictures with no grain whatsoever, especially with inkjet printers, but also on rasterized magazine print.

 

The M9 delivers files that are 5212 pixels wide. With 2x upsampling you can create an image 10424 pixels wide. Printed at 133 lpi raster and 1.5x quality factor, you can get away with 200 ppi, and create an image 52 inches wide in your magazine with pretty good quality. That is 1,3 meters wide... (or about 1 meter wide with 2x quality factor)

 

Now if you do the same conversion with your P65 files you will get a pretty darned large print!

 

There is also the issue with viewing distance. How far away will your viewers be?

 

Bottom line, a properly photographed and postprocessed file from the M9 can stand up to any magazine job, and all but the largest billboard or wall-sized prints.

 

Let me know if you want some sample images.

[skinnfell]

Ok here are the samples:

 

In order from top:

1. 100% crop, straight from raw file (no other processing or cropping)

2. Full image, downsampled with "Bicubic sharper"

3. Full image, downsampled with "Bicubic" , (standard setting)

4. Crop from 200% upsampled image.

5. same but with 5% noise added.

 

As you can see the standard downsampling works pretty well for screen uses.

The 200% crop doesnt contain much aliasing, but if it did, a little noise would smooth over it. No other sharpness added. The 200% crop corresponds to an area roughly 9 centimeters wide out of a print 1.3 meters wide.

(this was a handheld image, and focus somewhere in the middle. And since we are on the leica forum - shot with M9 with 24/2.8 asph)

[Paul J]

Thanks Skinnfell for the detailed post and yes pretty much goes along with what I have found also in terms of up sampling and downsampling. Bicubic smoother or sharpener does a really good job these days and adding 2-5% noise goes a long way to any image too. There's always sharpening to do once images have been resampled. It can really make or break an image. I have found you should do as much sizing in the RAW conversion process rather than in photoshop for maximum quality. It's actually best to out put from RAW the final print size and I have one large format mag that actually demands this.

 

After testing the M9 I think a 60" print is really quite achievable. Billboard too, depending on the style of shot, usually as the viewing distance on a billboard is quite far away. I have a job next week which a few of the shots are going to be billboard and I'm going to be shooting with the Leica for a certain look so it could be that at least one of the shots are run as billboard. I used to own a Leaf Aptus 22 which was up to the task so i think it should be fine.

 

Out of interest I've many times had images run at A5 and smaller which have been shot on the P65 and not thought they had looked damaged in anyway beyond what detail is lost in offset printing so I'm dubious of the down sampling damage argument but still interested to hear from an prepress expert on it.

[Paul J]

Advertisement (gone after registration)

Just saw the samples. Thanks for uploading those, nice shot too.