16x20 in prints from 20D

[magmac21]

I am getting a bit confused by image size and file size again. I have an image shot in raw from a canon 20d . Converted to tiff in photoshop has a file size of 46.8MB at 10x6 inches at 350 pixels/ inch. I need to make 16x20 inch prints from it and I followed a previous thread on this forum suggesting using the crop tool and setting dimensions required etc and cropping. This works but I am ending up with 164.8 MB files ! I will be sending these to a lab over the web and I assume these are far too big. Is this just what happens when you make 16x20 prints or can i reduce the file sizes at all and also am I going about this the right way ?

[BrandonSi]

I save my final prints in level 10 .jpg and upload to mpix.. they're usually about 2mb each. I'm not sure how you're trying to get your final prints, so I'll just tell you how I do mine..

Say I have a 10x6. Obviously 10x6 is not the same ratio as 16x20. You need to crop to 7x5.6 to get the same ratio, so using your crop tool put that in as your measurement and crop away. Then when you have your cropped photo on it's own, go to "Image Size" and change the size of your image to 20x16, and I'd specify the PPI at 250. Make sure the sample method is bicubic smoother. Hit ok. You know have a properly sized 16x20 upsampled to 250 ppi. At this point, optionally play with the USM and see if sharpening helps the photo any, sometimes it doesn't. Convert it sRGB profile and then save as .TIF for future reference. Then "Save As" jpg with level 10 compression, ensuring the sRGB profile box is checked. Send that to your printer/printing company.

How the image actually looks is going to be determined on the original quality of the image you cropped from.

[ScottE]

First I would crop the image to the 4x5 ratio needed for a 16x20 and do all the processing for colour, levels, curves, etc.

Second I would find out what the lab that does your printing wants. Many have a RIP program that does a much better job of resizing than you can do with bicubic resampling in Photoshop, or most other programs. If not, they should be able to tell you what pixels per inch is optimal for their printer. (If they don't have a RIP program and don't know the optimal ppi you should consider looking for a different lab.)

If the lab has a RIP program, I would just apply a modest amount of unsharp mask and send them the file. (If sending over the internet you may have to convert to JPG to get a small enough file size.)

If the lab does not have a RIP program, I would resample to give the pixels per inch they consider optimal. Bicubic resampling is good, but Pyramid resampling in programs like Qimage is better. After resampling I would apply unsharp mask as the last step before sending to the lab. Again, you may have to convert to JPG to send over the internet.

Scott

[jfrancho]

I would find out what the lab prefers to get. Anything we could recommend would be a best guess. Once you know what the lab's prerequisites are, many here can help you with the process.

[Poco]

I am getting a bit confused by image size and file size again. I have an image shot in raw from a canon 20d . Converted to tiff in photoshop has a file size of 46.8MB at 10x6 inches at 350 pixels/ inch. I need to make 16x20 inch prints from it and I followed a previous thread on this forum suggesting using the crop tool and setting dimensions required etc and cropping. This works but I am ending up with 164.8 MB files ! I will be sending these to a lab over the web and I assume these are far too big. Is this just what happens when you make 16x20 prints or can i reduce the file sizes at all and also am I going about this the right way ?

Regardless of why you are resizing the image or whether you need to (unlikely) I will say that is about the correct size for a 7000x5600 pixel (20x16 @ 350 ppi) uncompressed tiff. So if you really want the image to be 350ppi then you are stuck with that size.

Now, as for whether you really need the image to be that big, as others have said, you will have to say what the lab wants.