Big, bigger, Biggest!

You have more than one issue here.

The tree removal would depend on the picture, but usually the clone tool is your best best.

As for size, what resolution has the printer requested?

Bob

Reasmple to anywhere from 240 to 300dpi and use unsharp mask right at end, save your very large file in any format the printer asks for. Do all your editing at the lower end before resampling.
Good luck

wrote in message

Hi

I have a really good 1.3Mb jpg photo taken with a 5mpx digital camera. I

need to be able to enlarge the photo for use on a banner measuring 2 metres x 1 metre. Also, I need to "remove" a tree from the photo and rotate the photo slightly because a sign is crooked and add some wording(no problem at doing these changes).
You must also cut something out because digital cameras do not employ a 2:1 aspect ratio.

What is the best way to put the picture into Photoshop, make the changes I

need to make and store the photo at its highest resolution for use on the banner eg do I open the file in Photoshop and save it as a psd, then make the changes, then enlarge the photo (I’ve heard you can only enlarge in 10% increments?) or, instead, do I place the photo in Illustrator and work on it there, or do I export it to PDF, or …?
Install a memory card reader in your computer. Remove the card from your camera and install in the reader. Copy the file from the reader to your hard drive. Eject the card and replace it in your camera. If you don’t eject (that is a software command that you will find in the device properties tab), you will most likely damage the files on the card.

Once on your hard drive, you just open in Photoshop and continue onward.

For some reason, when the people doing the banner opened my Photoshop

file, they said it was only at 72dpi even though my file was 20Mb in size. This seems strange because when I printed the file on an A4 page, the picture was extremely good quality.
You heard wrong. The 72 dpi is the default setting. A picture that looks quite good at A4 may look terrible when enlarge to 1 meter x 2 meters. Your image needs more pixels, and hence it will be bigger.

20 MB isn’t all that big. I usually get files of 33 MB from a 35mm slide (and nobody would think very long about displaying the images at 1 meter x 2 meters, even if you could).

I’m using Photoshop 6 (will be upgrading to Photoshop CS soon when budget

allows).
Photoshop CS is no more useful for your project than is Photoshop 6. Jim.

eg do I open the file in Photoshop and save it as a psd, then make the changes,

Yes. Don’t do your editing on the jpeg as each time you save you will lose info. You will need to work on a psd in any case if you want to keep layers.

Sandie,

At 2m x 1m (approximately 78 x 39 inches), you would need a file that is 11,700 x 5850 pixels if the viewing distance of the banner was such that 150ppi gave a "perceived" high quality image. Now, let’s assume your camera image is 2500 x 200 pixels, more or less. From that, you can see that it is the long dimension where the highest degree of enlargement occurs…about 468%. I don’t know how extreme an enlargement can be made that maintains reasonable quality, but I’d think this is really pushing the limits, particularly from a compressed image to begin with (too bad it wasn’t a TIFF). For a higher resolution image, the demands will be even greater and I wonder if that would even buy you anything quality-wise? If the printer requires a 240-300ppi image, you may not have any choice but to go that high. That this is a banner, presumably to be viewed from a signficant distance, may be your saving grace.

You should definitely save your working image in a PSD or TIFF format. Peter’s suggestion to edit first, then enlarge, is a good one explained by that your edits will be making use of original rather than interpolated data, which should yield better quality edits.

As for enlarging the image, you can do it however you like. The 10% increments you’ve read about are not a limit but rather a suggestion on how the best quality enlargement might be obtained. That is multiple resampling at 10% per instance seems to provide higher quality results than a single resampling by some extreme amount. Fred Miranda’s Stair Interpolation action for Photoshop uses this approach and can be downloaded from <http://wwww.fredmiranda.com/software>. At one point in time, the action was a free download but I see now that there is a fee charged for it, but it is now called Stair Interpolation Pro. I’m not familiar with the new action but the older one did seem to do a good job for me. It may still be available freely elsewhere if you do a web search. I don’t see that the new action can be downloaded unless purchased first, which would make it difficult to assess if it is worth having or not, but I suspect it would be a good tool to have in your editing arsenal. Note that even within Fred’s examples, he only goes as high as a 300% enlargement. That’s not to suggest higher isnt’ possible, but it may suggest what one might consider "reasonable limits".

Finally, that others see your image as a large image at 72ppi is common. It is quite likely tagged by your camera as having a 72ppi resolution even though resolution doesn’t mean a thing unless you’re scanning or printing an image. If your image file has been tagged with a higher resolution, then the applications others are using to view it are ignoring the resolution information and are displaying it at the common 72ppi default. If you view the image in Photoshop and open the Image Resize box, then uncheck resample, you can change the resolution and see the affect that has on-screen by applying the change and then selecting to view it at print size. The on-screen size would then approximate what the printed size would be. The truly important parameters are simply the image size in pixels.

Hope that helps,

Daryl

Sandie,

even though my file was 20Mb in size. <<

A 2m x 1m RGB file at 150 ppi will have an uncompressed file size of close to 200 megs, not 20!

The pixel size will be 11811px by 5906px giving a total pixel count of 69,755,766.

Now you are starting of with 5,000,000 (approx) pixels. That is an enlargement of roughly 14 x. Your quality will suffer BADLY! In fact, it is probable that an image re-sampled by that amount will be totally unusable.

Oops…I just noticed a goof in my earlier message. On the approximate size of a 5Mpx image, that should’ve been 2000, not 200 pixels on the short side.

Len,

Reading your response I see we are basically saying the same thing, but I’m curious about your "14x" enlargment factor. It seems you are quoting that based upon the total area (total pixel count) than upon the linear dimensions. While I think it is correct to quote things either way, I’d have thought that "x" factors as we often see them on loupes, binoculars, etc., were all referring to linear magnification. Perceptually, I would think that would be easier to sell, since one can more readily recognize the doubled height or width of an object vs. simply a doubling of the area occupied. Nonetheless, having never actually paid attention to such details when using a loupe or other optics but rather just noticing they were "bigger", do you know if the "x" factor is actually referring to the amount an area is magnified by?

Just curious,

Daryl

If you have PS CS, there is no need for step interpolation for upsizing. If the file went through ImageReady as some time, that is probably how it got changed to 72 ppi. Any method of upsizing more than about 4x is dubious for good results, unless you are a great artist that can paint back in details that are lost. (unless you are in a hollywood movie, and then all things are possible)

Philo,

Why is there no need for step interpolation when upsizing in PS CS? Is that the approach PS now takes in an upsizing?

Sandie,

Another way to look at the numbers Len and I have thrown your way is, going from Len’s numbers, by looking at the percentage of final data that your original data contributes to. That is, 5Mpx/69.8 Mpx is about 7% original data…93% interpolated!

Regards,

Daryl

Daryl: the bicubic smoother gives better results than SI. I’ve tested up and downsampling (bicubic smoother and bicubic sharper) and they are noticably better than PS7’s routines.

Daryl,

Just curious,<<

Just re-enforcing that Sandie is expecting too much <g> and giving the pixel-count figure comparison makes the point very clearly…

Phil,

Thanks…I’ve never really done any such comparison, so that’s good to know.

Daryl

I actually think that you have enough pixels for your needs without upsampling. The key is that you are going to use the picture on a 2 meter wide banner, presumably viewed from a distance where even very low resolutions are fine.

To test this I took a 5×7 photo and downsampled it in Photoshop to 20 pixels/inch. Down to 100×140 pixels total! I printed this out at 5"x7", and of course at normal viewing distance it looked awful – unusable.

However when I put it up on a wall and viewed it from 10 feet away it looked just fine. Not quite as good as the original, but absolutely acceptable.

As to where to get a print, all the online places will banner print, but since it will be expensive, I suggest you print out 1/10th of your banner on single sheet of paper and see how it looks from the expected viewing distance

John.