Workflow question-when to uprez

[Bill Jaynes]

I'm beginning to do some moderate up-rezzing. Very interested to hear at what stage in their workflow people are doing this and, briefly, why.
Thanks,
Bill

[jbrembat]

I'm beginning to do some moderate up-rezzing. Very interested to hear at what stage in their workflow people are doing this and, briefly, why.
Thanks,
Bill
[a href=\"index.php?act=findpost&pid=124804\"][{POST_SNAPBACK}][/a]

Upsampling or downsampling, in any case resampling, is performed, generally, for printing and/or for web publishing.

For printing, resampling is done to get the right PPI value to avoid driver intrerpolation.

For web publishing, resampling is done to reduce image size and image weight.

In any case resampling is output referred, so it mut be the last operation (I suggest to save the image in original size before resampling).

Depending on resampling algorithm and destination (monitor or printer), you may need some final sharpening.

Jacopo

[eliedinur]

I'm beginning to do some moderate up-rezzing. Very interested to hear at what stage in their workflow people are doing this and, briefly, why.
Thanks,
Bill
[a href=\"index.php?act=findpost&pid=124804\"][{POST_SNAPBACK}][/a]

Uprezzing can quickly increase file size. For instance, a 46 Mb 16 bit file upped by 1.7x will balloon to nearly 130 Mb. Depending on your setup, of course, this can put quite a strain on resources during editing and is therefore best left to the end.
Elie

[nemophoto]

I'm beginning to do some moderate up-rezzing. Very interested to hear at what stage in their workflow people are doing this and, briefly, why.
Thanks,
Bill
[a href=\"index.php?act=findpost&pid=124804\"][{POST_SNAPBACK}][/a]

I frequently have to upres images for a client who then prints 8' and larger posters. The images are usually 30" on the long side, at 400dpi. The when and why are important. The when: as the last element of production because working on a 300-500MB image is a pain and slow. I also prefer to do all retouching and levels, curves, color correction in 16-bit mode. Depending upon software used to res-up, I also do that in 16-bit, before converting the final file to 8-bit. The why is because, for their printing process, they need larger, "denser" files than the 1Ds2 produces. The 1Ds2 is natively about 9x12 at 400dpi. Obviously, the file must go up over 200 percent -- sometimes much more, because the AD invariably crops even tighter. (I just had to massage a 1D2n file to over 800 percent.)

Nemo

[pfigen]

I actually prefer to do my uprezzing as soon as the file gets to Ps. Retouching and masking are all much more effectively and accurately carried out at the higher rez even if it does take a bit longer. Invariably, if you retouch and then rez, you will end up fixing what you missed working on the native file.

[01af]

When I want to upsize an image then I always start with the raw file. Usually I shoot RAW+JPEG so I have an in-camera JPEG file and a raw file of each shot. I often end up using the JPEG ... but not for upsizing.

My preferred raw converter is Adobe Camera Raw which has a resample feature built in so it can do up- and downsizing at raw conversion time. However I don't use it as it introduces tiny ringing artifacts around high-contrast edges. Apart from those artifacts, the resampling result is the same as Photoshop's resampling method Bicubic Smoother.

So I convert to the camera's native size into TIFF format. I don't let ACR apply any sharpening but I do use some of ACR's color noise reduction (strength 4 - 8; the default of 25 is waaay too strong). Before ACR 4.1, the sharpening was too coarse so I never used it. ACR 4.1's sharpening now is much better so I may adopt it eventually but have not yet changed my established workflow.

Before upsizing, in Photoshop I always pre-sharpen the image by applying Sharpening For Source as described in Bruce Fraser's book "Real-World Image Sharpening." It basically is Unsharp Mask with a high amount (200 - 400 %, depending on the camera's AA filter characteristics) and a small radius (0.4 - 1 pixel, depending on the camera's pixel count), applied to the medium tones only (using the 'Blend If' sliders) through an extra layer in 'Luminance' blending mode. Note that this is NOT the same as Capture Sharpening which would be Sharpening For Source and Sharpening For Content combined.

Upsizing the pre-sharpened image will retain more detail than upsizing the unsharpened image. That's because the upsizing process will 'flatten' the unsharpened details so after upsizing they won't be accessible to sharpening anymore. Pre-sharpening the details before upsizing will greatly reduce this unwanted effect. For the actual upsizing I use Photoshop's Image Size command with resampling method Bicubic Smoother. After upsizing, Capture Sharpening will be applied.

It doesn't matter whether tone and contrast adjustments are applied before or after upsizing. The majority (if not all) of these kinds of global adjustments should have taken place in the raw converter anyway. It seems to be a good idea to apply delicate local adjustments---in particular those that incorporate paths, selections, and/or masks---to the upsized image, for exactly the reasons pointed out by 'pfigen' (see above).

-- Olaf