Digital BW, The Print

What would happen if an image file were first upsampled 200% using GF,
then montaged with a second file of higher native size....and then
finally GF used again to downsize to a smaller extent the montage
after it is saved as a psd or tiff and flattened?

 Final file size after second iteration would be greater than original
smaller file size. It would be downsampled(GF) or downsized(Pshop
image size dialog) the second time so as to enable different sizes
(small and large) to be printed on Epson 2200. Sizes around 8x10 and
11.5x17.

Also: Andrew Rodney believes that GF is not better than Pshop bicubic
for printer output unless file is at least 20MB to start out with. I
assume that is RGB---> a 6MP digital camera is bare minimum needed.
Any thoughts? I know many in the group think that the MP count should
be way higher before they buy one.
TIA,
Jim H.

--- In DigitalBlackandWhiteThePrint@yahoogroups.com, "jim hayes
<jimhayes@f...>" <jimhayes@f...> wrote:
> What would happen if an image file were first upsampled 200% using GF,
> then montaged with a second file of higher native size....and then
> finally GF used again to downsize to a smaller extent the montage
> after it is saved as a psd or tiff and flattened?
> 
>  Final file size after second iteration would be greater than original
> smaller file size. It would be downsampled(GF) or downsized(Pshop
> image size dialog) the second time so as to enable different sizes
> (small and large) to be printed on Epson 2200. Sizes around 8x10 and
> 11.5x17.
> 
> Also: Andrew Rodney believes that GF is not better than Pshop bicubic
> for printer output unless file is at least 20MB to start out with. I
> assume that is RGB---> a 6MP digital camera is bare minimum needed.
> Any thoughts? I know many in the group think that the MP count should
> be way higher before they buy one.
> TIA,
> Jim H.

Correction: Michael Reichmann, not Andrew Raodney. Aplogies.
Jim H.

--- In DigitalBlackandWhiteThePrint@yahoogroups.com, "jim hayes 
<jimhayes@f...>" <jimhayes@f...> wrote:
> What would happen if an image file were first upsampled 200% using 
GF,
> then montaged with a second file of higher native size....and then
> finally GF used again to downsize to a smaller extent the montage
> after it is saved as a psd or tiff and flattened?

The simple answer is: it depends on the image. From what I've seen, 
GF 'preserves' edge information when scaling up better than PS 
bicubic interpolation. That is, a source image that depends on hard-
edge definition and smooth, flat tones (e.g. architectural and 
industrial subjects) is probably going to benefit relatively more 
than, say, highly textured, organic subjects.

I have experimented with it on such subject matter of quite low 
resolution which I brought up to 12 x 16" @ 240ppi. There was 
extensive color artifacting (notably bloom) quite similar to jpeg 
along (pictorial) red edges, but this was easily corrected by running 
the desaturation 'sponge' tool along these edges. Noise areas turned 
quite blotchy, and had to be subdued via careful selections and 
application of the median filter. 

I am generally satisfied with this result, though in some respects it 
resembles the PS watercolor filter. Others coming from a more 
traditional photographic background may feel it is not acceptable to 
attempt to 'overdrive' low-resolution photography in this manner.

All I can suggest is to experiment and decide for yourself.

I'm not sure why you would want to use GF to downsample your finished 
montage though, I don't see any benefit there. A bicubic resample 
with unsharp mask (in stages, if necessary) would obtain far better 
results, I would think.