ACR workflow option

[rogprov]

With an image in CS3 ACR there there are number of options in the Workflow dialogue box - Space, Depth, Size and Resolution. The first two are self explanatory but I'm having difficulty in understanding the correct use of the last two - Size and Resolution. Why would one choose to change the default settings to decrease or increase the output size and what good or bad effect would a resolution change have? Would any change from the default bring advantages in some situations? In short - why are they available?

Regards,
Roger

[David Kilpatrick]

Resolution just sets the dpi - 72, 300, 240 or whatever you choose - which is convenient for further uses. Example, if you use DTP layout programs you don't want 72 dpi as a placed image ends up being six times the size of your entire page at native size. So you set 300dpi.

The size option is useful. You may only want to export the smallest size, for web use, as sRGB - or the largest size for photo library use aRGB. Once you have set a combination of these things, ACR remembers it for the next photo (you must either click DONE or actually open an image, to have it remember the settings). All my images are exported to 5120 pixels wide (17.5 megapixel size) whether shot on D7D, A100, A350, A700 - whatever. They all get sized identically to go into my library master file archive where they are keyworded, captioned, etc.

David

[rogprov]

Thanks David.

I thought the resolution change must be for use as you describe but I'd not realised the up-sizing option was so convenient for submissions.

[01af]

... I'm having difficulty in understanding the correct use of the last two---Size and Resolution.

Resolution, in PPI, is just a number in the image's metadata which has no meaning until the image gets materialised---i. e. sent to a printer, a printing press, or a layout program (in the latter, the 'materialisation' is only virtual but still it's some sort of output where the absolute sizes of several components coming from different sources, like text, graphics, and photographs, do matter). You may enter any number that makes you happy, except zero. The EXIF 2.2 standard suggests '72' as the number to enter when the output resolution is unknown or yet to be defined. You can alter that number at any time later, with no consequences to the actual image whatoever.

Size, in pixels, is just that---the size of the image. Usually you'll choose the image's native size which is just the width and height as created by the camera. You can also interpolate the image to a larger size or reduce it to a smaller size, according to your requirements.

However upsizing in ACR, albeit convenient, is not recommended. ACR's resampling algorithm will add ringing artifacts to high-contrast edges. They are tiny and hard to detect but still undesirable---see the attached file, part (A).

To avoid the ringing artifacts, simply convert the image in ACR to its native size, import it into Photoshop, resize the image there using the resampling method Bicubic Smoother, then apply capture sharpening. The result will be basically the same as upsizing in ACR but without the ringing artifacts---see the attached file, part (B). Furthermore you can select any arbitrary size in Photoshop rather than being restricted to just a few fixed choices in ACR.

The attached file shows a sun-lit part of a sunflower leaf against a dark background at 400 % view---so it's a tiny detail at huge magnification. When viewed on an average monitor, the full image would be 3,30 × 2,20 m (11' × 7' 4"). It's from a Dynax 7D raw file so the whole image's native size was 3008 × 2000 pixels. In Adobe Camera Raw 4.5, version (A) received some capture sharpening (Amount = 25; Radius = 1.0; Detail = 50; Mask = 35) and got upsized to 5,120 × 3,404 pixels (17.4 MP). Version (B) was converted from the same file through ACR 4.5, too, but at native size and without any capture sharpening (Amount = 0). It was then upsampled in Photoshop CS3 to 5,120 × 3,404 pixels using Bicubic Smoother. Finally it received some sharpening for source, to approximate version (A) visually. Click on the attachment to see it at full size (560 × 540 pixels, 51 KB).

-- Olaf

[rogprov]

Thanks Olaf for all your time and trouble, it's all most helpful and gives me a lot to experiment with