Image Content issues vs Output Needs

I'll certainly give it a go - thanks for that Rob, and Will too. I guess as everything is being 'upscaled' together its reasonable to assume that both the 'good' quality aspects as well as the 'bad' that I am seeing during the sharpening stages will equally be exaggerated. But its worth running a few tests.

I don't think it needs to delay any conclusions based on the methodology I've been pursuing - but sometime in the future I'll call up my PS Elements 'Resize Image' dialogue and punch in some numbers .... just have to check out those 'Beginners Guides' first

Yes, sort of. Traditionally the upscaling is done at the end of the workflow to optimise printing and when used like that all it does is allow better quality resampling than the native print driver so that there are better quality jaggies.

But it is of course possible to do actual editing at the larger upscaled image size as well. If you do that it is entirely possible that the sharpening, or any other microcontrast/denoise/edge-aware processing will work differently. You may find in practice it isn't very practical speed/memorywise to work on the largest files but it may be possible to work on an intermerdate upscaled image - sorry to introduce yet another variable here.

I couldn't resist

IMAGE RESIZING TEST

Ok, I fired up the Resize Image box. Taking the KIIs .psd file that had the 200% sharpening applied, I see the pixel dimensions as 4928x3264, and document size is 52x34 cm.

I choose from the drop-down box 'Bicubic Smoother'.

I've been talking about my 100x70 cm poster prints, so lets try that - leaving 'Constrain Proportions' ticked (my maths is rubbish ..) , I activate 'Resample Image' and key in 100 cm width - straight away this gives me 66cm height, and my pixel dimensions are 9449x 6258. Resolution remains at 240. Wow, I hope this algorithm is a good one! Maybe its not such a big increase?

I save out to a new version, and open up the smaller one alongside.
I can see it straight away. The new resampled version has lost its edge completely. The fine crisp, slightly jagged, lines have lost their 'jag', but probably lost a good 70 - 100% of sharpening as well!

Hmmmm..., maybe this algorithm is not such a good one? There's no way I'm going to start resampling like this and sending the file to my printers. They clearly have something much better up their sleeve, based on the supurb results I've always had on these 100x70cm posters.
I don't see this affecting the outcomes of the 'content' assesments on the comparison thread.

I find it best to do your sharpening after any resizing.

You might find that what looks horrid on screen prints OK though!

My lazy-man way is to print from lightroom, I have no problem printing A3. The biggest change I've made recently in my printing is, I've gone from an epson to a canon printer. The epson wanted 300ppi but the canon wants 600ppi to get the best out of things.

Don't expect good results if you take an image you have optimised the sharpening for at smaller size and then just resize it without further work. Try comparing after a) do basic input sharpening on two duplicate images b) upscale one of them to larger size c) do final sharpening to now optimise each of the images separately d) now compare - obviously at equal image size not 1:1 pixel ratio.

If you don't see a benefit it's not worth it. Note that other experiments have shown differences between the two methods for actual printing (where printer interpolation was thus avoided) but it was subtle. In fact for moderate enlargements it was only apparent with a magnifying glass on areas such as the curve of letters text. Whether that is worth the extra work is a personal question, and whether it will actually show on screen is another.

Thanks guys, I really appreciate your input on this. Finding the right point in such a complex workflow for all the various enhancement and re- sizing operations is going to take a bit of experimenting.

For now, I am happy to leave the re-sampling out of the K5II comparisons. Its an extra dimension that won't really affect a camera purchase choice. Individuals will have to consider their personal printing needs and factor their own knowledge in to it.

I'm certainly glad that I've found a good online service where it doesn't need consideration at all

If you allow a printer to re-size your prints, you don't have the option of sharpening them yourself. Enlarged prints should be sharpened after re-sizing; you will not get the same results sending an already sharpened file and allowing them to enlarge it. On smaller enlargements this may not be particularly noiticeable, but on very large prints it will be. I'm printing sizes up to 60 inches, and send the file already re-sized and sharpened. I am certain that sending a sharpened file and allowing the printer to enlarge it will not get the same results.

Another factor with large prints is viewing distance. While sharpening for screen can be done at 100%, for prints it's best to do sharpening at 50% or even 33% for a 2x enlargement, 25% for an even larger print. A print viewed at 4 feet also does not need to be printed at 300ppi. I usually use 240ppi for large prints that will go on a wall, which is more then enough imo. (though I noticed, Greg, that you said "Resolution remains at 240 ppi". This seems to be a default setting in Lightroom. An un-enlarged file should always be printed at 300ppi imo). As the ppi decreases, the optimum radius for sharpening increases, as it also does with viewing distance. Cambridge Colour have a sharpening radius calculator here ,which I find works especially well using the High Pass Filter in Photoshop to sharpen an enlarged image. Using Unsharp Mask it doesn't seem to work as well, with the eyeball being a better calculator in this case.

I am always being amazed here on this forum, that it is possible to draw out such a range of knowledge and experience, and what you're saying just above aliengrove just adds even more to that.

Its all food for thought and further experimenting, just like the contributions from the members further above also. It seems I am very lucky indeed with my chosen online printing service (FOTO.COM) - even though the sharpening is done before upload and there's no specific resizing, they seem able to create prints of any size that exactly match each other, and look the same as on the screen. I have peered close up at 12' x 8' prints and the same shot at 100cm x 70cm and can't tell any difference, so somehow they are just managing to reproduce exactly what they're sent at any size.

With this thread I tried to draw a dividing line between questions of print resizing and the specific image comparisons on the other thread. I would have struggled to produce meaningful conclusions about printed output as the variables would have been too many. It seems to me that it is best to take the conclusions from the comparisons as to type of image content and processing you use, then interpolate (not extrapolate ) to your own normal print process and preparations.

I think the point Aliengrove makes about viewing distance is key..there's no point viewing a 36x24 inch enlargement at 100% on your PC.

Print media also makes a difference, especially if printing on canvas as the "grain" in canvas is much more forgiving, infact I have 36x24inch canvas prints done at the native image ratio, which normally equate to around 120-130PPI, they look perfect

Simon

Just to clarify, we do mean viewing at 36x24 with the image resampled to that size don't we? If we are just talking about the physical size property of the image then obviously the image itself isn't changed and wahatever the print size is set to doesn't matter for image viewing.

Assuming we are talking about resampled images: There seems every point in viewing the image to me as, otherwise, how would you ever judge if you have a suitable amount of sharpening for the intended print?. Agreed that you don't need to view that at 100% though - unless you are interested in such things as how good your resizing/resampling process is working.

What do you mean here by "native image ratio" ?

Q1.. yes, I'm talking about resampled images, I view mine at 50% on the PC once sharpened at the larger size..more representative of "real world" viewing.

Q2...I mean the standard print size the camera will produce at 300PPI without any resizing, on the K5 its 15x10 inch so a 36x24 comes out around 120-130 PPI

Simon

There's nothing to stop you doing a 'test strip' with the image size set to whatever you want... you can do this without, or with, resampling, to get a preview of what the print will look like. I would suggest it's not necessary to resample unless you are going 'huge', but this way you can find out for sure without wasting an expensive print.

Of course disable the 'fit to print' feature in your printer, and print off an 8x4 print or something.

I find that the screen doesn't perfectly reflect what comes out of the printer, even if viewed at a 'representative size'.