Digital BW, The Print

After reading the various posts, I decided to make another (and final 
for me)  test of the various methods using an image of similar size

I took an original 8,230KB tiff file (produced by Capture One from a 
larger raw image) as my starting point, and used PS7 to produce a 
version that was 1,320KB to be used as my test file.

I then upsized this test file back to the original size so I could 
compare it with the original, both on screen and as a print (of a detail.)

SizeFixer (55s) produced what was probably the best print. On screen it 
showed crisper detail but also some more obvious sharpening edges, and I 
think the sharpening is really what made the print look better. I had 
USM off and it still seemed to sharpen considerably. There were several 
areas of the screen image where it seemed to have slightly better detail 
than the others, but the differences were slight.
My second preference was for ACDSee8, using its Lanczos algorithm. This 
image was possibly the one that looked best (just) on screen. Slight 
sharpening would probably have given a better print, but the resize 
didn't give an option.
PS7 bicubic also gave a good result, and was a close third. I tried 
upsizing to 120% and then sizing down using PS7 bicubic and the result 
was identical..
QImage was also good, but softer than the other results. When a small 
amount of 'deblur' was used it was as good as the rest on screen. Too 
much deblur produced obvious artifacts.
The methods other than SizeFixer were all faster, I think all under 10s. 
All too short for me to accurately time. The deblur filter (from 
SizeFixer) I used on the QImage result was slow - perhaps again around a 
minute, but I didn't time it.

The results conform to the conclusion in my feature - if you already 
have decent graphics software it will do the job reasonably and you 
don't need SizeFixer for such moderate enlargements, though it does do a 
decent job for making prints. I'd prefer to apply the sharpening myself 
though..

All of these results seemed pretty usable to me, and quite difficult to 
distinguish. On screen none was quite as good as the original, as the 
detail lost on sizing down can't be recovered. Because of the 
sharpening, some people might have preferred the upsized print from 
Sizefixer - the loss of detail is hardly perceptible in the print. But 
really this is all at the level where it is difficult to tell if you can 
really see the difference so far as prints are concerned. I suspect if I 
had applied different levels of sharpening to each result I could well 
have given different ratings too. Really too close to call. So the good 
new is save your money, use what you have. Almost everything is good, 
and there are no miracles.

Regards,

Peter.

Peter Marshall
petermarshall@...     
_________________________________________________________________
My London Diary	              http://mylondondiary.co.uk/
London's Industrial Heritage: http://petermarshallphotos.co.uk/
The Buildings of London etc:  http://londonphotographs.co.uk/
and elsewhere......



Clayton Jones wrote:

>Hello Bruce,
>
>  
>
>>Thanks for letting us know your results!  Were all the comparisons 
>>done using the SAME image?  
>>    
>>
>
>Yes, it's on the web site now, here's a link.  It's the 2nd one down,
>called "Panamint Range" 
>
>   http://www.cjcom.net/cal05-a.htm
>
>It's a mass of tiny detail with a bit of sky at the top.
>
>
>  
>
>>It might be good now to do the experiments over again using a 
>>totally different image (different type contrast, subject matter, 
>>focal length) just to see if the results hold true in general & 
>>not just for that one image.  
>>    
>>
>  
>