Understanding File Sizes

[KellyL]

For the last two months I have become a convert to RAW shooting, and am soooooo grateful that I finally took the step. I am somewhat confused by the file sizes, I am using lightroom, and when I export my files, I leave the settings at highest quality jpeg, 300 dpi. The resulting image is 3504 x 2336, the measurments are 11.68 x 7.87 inches.

How does that work when I upload to my lab and order a 11 x 14 or a 16 x 20 print and use their online cropping tools? Is the resolution of the image being degraded because I am uploading the 3504 x 2336 image and not creating an image set to the largest print I am ordering?

I don't know if this question makes sense or not, but I don't quite get this aspect of digital photography.

[blumesan]

Kelly,
I don't have Lightroom so there may be an easier way to accomplish what I will suggest. I assume you have some version of PhotoShop.

For the best image quality it is best to perform any resolution changes (pixel dimensions) and cropping on a PSD or TIFF file in 16 bit mode rather than using a jpeg file. So save your lightroom adjusted file as a TIFF. Open in PS and use the crop tool there to crop the image and set the resolution to whatever your lab recommends. This is a one step proceedure using that tool. Once done, convert the image to 8 bit mode and save as a jpeg at highest quality. Then send it off to the lab.

Cheers/Mike

[Mr. Pickles]

I think what you want to know is two parted, so here goes.

If you want a 11x14 or 16x20 print, from an image that doesn't indicate that at 300dpi it would make a print that big, upload it anyway. If your image is sized a certain way to only allow but 150dpi, upload it. You will get pretty good results.

If the image so will save is not in the same "format ratio" as the print you want, then I would crop accordingly, and upload that. A 4x6 and a 8x10 are two different format ratios. One is 0.6666 and the other is 0.8.

Typical DSLR's cameras shoot in the 0.6666 format (Olympus 4/3 sensor is different at 0.75. As you can figure, an 11x14 (0.786) and a 16x20 (0.80) are different ratios too.

Inside of Lightroom (and in PS) you can use the crop tool to make a certain ratio. In Lightroom, it is a drop down option (Custom), and in PS it is done along the top when crop tool is selected via a drop down (Aspect Ratio). You draw a "bounding box" as big as you can (typically the length or width of the image, with the other dimension 'cropped' off. this way YOU are controlling where the crop is, and not the place you will get your print from.

Some print houses will let you crop or set the portion of the image you want, but it is best to check first, or just upload your own "cropped" version.

I'd save the cropped one as a JPG and upload that. If you have to "export" and use PS, then what Blumesan stated is pretty typical of "people who want the best possible"...

[JohnC]

What kind of printer is your lab using? Inkjet? Lightjet? Does the printer do any up/downsampling in its hardware?

I usually prepare my files at the size/aspect ratio/resolution I want to print at and upload those to the lab. If I want an 8x10 and an 11x14 from the same file I upload two separate files because of the different aspect ratio.

I do not strictly adhere to 300 dpi. For larger prints (11x14, 16x20, 20x30) that will be viewed from further away I may use 240 dpi. Sometimes even 180 dpi. I would suggest sending some test files to the lab so you can see the differences directly. Prepare the same image at a few sizes and order the same size print from each. Make sure to add text to each image so you know which is which when looking at the prints.

It sounds like you understand the different aspect ratios and cropping, but for anyone following along who isn't sure they understand, I've written up a brief bit about aspect ratios: Aspect Ratio

[KellyL]

I understand the different crop ratios and how they effect what a picture will look like. My concern is I upload a file that is roughly 12 x 8, and when I crop for a 16 x 20 it seems like I am relying on the hardware of my lab company to interpolate the smaller file for a larger print. It seems like that might be a messy thing to do. I have used Roes, LabPrints and Full Color, all three allow you to upload a file and then use online cropping tools to create your order. Since I know that my file is a certain size coming out of the RAW state, it seems like when I order a print larger than the processed RAW file yielded then I am literally "stretching" my picture to fit a larger canvas. It seems like this would cause image quality to suffer. I have wondered if I shouldn't be exporting larger files from Light Room. I usually use Full Color, love their service and online ordering tools, but I don't know what kind of printer they use.

[heartseye]

There are some quite sophisticated "sizing" programs......most seem to agree on Genuine Fractals on the LizardTech website.....bob

[Mr. Pickles]

Originally Posted by KellyL

I understand the different crop ratios and how they effect what a picture will look like.....
Since I know that my file is a certain size coming out of the RAW state, it seems like when I order a print larger than the processed RAW file yielded then I am literally "stretching" my picture to fit a larger canvas.

Whichever way you slice it, you're "stretching" the picture to fit a larger canvas. Does it matter if you use Lightroom to do it, or Genuine Fractals, or the lab? Maybe, but maybe not if you have a decent size image to start with, and aren't going to a 32x40 print or larger.

If you want a 16x20, and one thinks they need 300dpi to get a good print, then they need image dimensions of 4800x6000, or 28.8mp. You got a camera that can do that?

I'd think not. So, you need to "stretch". If you got 8 or 10mp, and you aren't cropping much if any of the long side, then you'll get good results if you upload to a good print lab. If you upload to "Freds Prints-N-More", then you might not. Might want to uprez them yourself, or find a new print lab. Or, run a test like mentioned before.

[KellyL]

Thanks gentlemen. I appreciate your comments and you have helped to calm my fears.

[heartseye]

Another small thing Kelly......you likely know this.....if you crop to the final size yourself do not use the crop tool. Do it using the rectangular seection tool then "crop to selection" This way the image doesn't get resampled. Cheers...Bob

[KellyL]

No...I did not know that. Thank you.

[JohnC]

Either way should work about the same, as long as you set the tool's options.

Using the Marquee Tool you can set a fixed aspect ratio (i.e., 8x10). But then you need to go to Image > Image Size and resample to set the size of the file.

You can use the crop tool and set the size and leave the resolution field blank and it won't resample. Or you can set the size and resolution and do it all in one step.

The advantage to using the Marquee Tool is that you can select which type of resampling to do when you resize (bicubic, bicubic smoother, etc.). If you use the Crop tool you use the interpolation method you have set in Photoshop's Preferences.

With either method you are going to end up resampling somewhere in the process.

[KellyL]

I thought that was one of the beauties of the crop tool, the ability to preset all the numbers, including resolution.

I am not sure what the difference is between bicubic and bicubic smoother, etc. I always just leave it on the default settings. I have read in Jack Davis' books that he leaves them on default...I believe.

[jfrancho]

Kelly, Take a look at the differences in resampling methods in this tut: http://www.photocamel.com/forum/imag...ts-images.html. By using the snapshot feature of the history pallet, you can try all three methods in the bicubic resampling family and quickly compare the difference. I definitely do not recommend using the default. Each suits each image differently, though in general its bicubic smoother for upres, and sharper for downres, but not always.

[KellyL]

Thanks John. I missed this reply somewhere along the way. I have some studying to do as today I was just thinking to myself, I have no idea what these different resampling options do.

[Max@Home]

Originally Posted by KellyL

I have some studying to do as today I was just thinking to myself, I have no idea what these different resampling options do.

since you will be trying and testing anyway: another 'golden tip' I'd like to share: if you have to upres (to keep a high DPI), do it in small steps, say a maximum of 10% per step,and first reset the DPI so you start off with the intended end-goal.

or: from 3000px x2000px @ 300DPI (printsize 10x6.66) to 30x20 printsize at 220DPI (6600px x 4400px): first reset DPI, so you get 3000px x 2000px @ 220 DPI = printsize 13.64 x 9.1, from there upres in steps of 10% till you are 'almost there' and let PhotoShop fill the final % to 30x20. & I agree on JohnF's Bicubic Smoother for upressing, *most* of the times

I use this method I once picked up from the internet for my companies 'hallposters' (printsize A0, 80cmx120cm at 240DPI) printing on OCÉ 4-colour ColourPlotter (TCS500), and get very pleasing results

...smaller steps (5%) might even yield a better result... worth a test?

...€0.02...

Kindest regards,

Max@Home

[KellyL]

Max, do you know what version those instructions were written for? I had read about the 10% increase when using PS7, but I had heard in an interview or read in a book that with CS or was it CS2 that the 10% issue had been resolved.

Anybody else hear/read this?

[Mr. Pickles]

The 10% increase in steps does come form the PS7 (or earlier days). The introduction of the extra Bicubics has been said in many places, and by many pros, to make the multi-step procedure a waste of time because the better result isn't there.

BUT, some still prefer the multi-step approach, and feel like it offers them the results they want. Might be some "if it ain't broke..." mentality too, but it might be best to try both ways and find the one that works.

And like others said, there may not be one method that works best on all images...