Test of Vuescan 8.3.15 running a Canon 9950

[Ed Richards]

www.public-health-law.com/photos/

This is a two pass scan to allow 4800 dpi scanning of a color negative. The first pass is used to generate a raw (unprocessed) scanner file that is down sampled to 2400 dpi. This is done because the full size file is more than 2 gigs, which causes memory issues in Windows because processing the file can push past the 4 gig limit. Vuescan is then used to load the raw file for processing. It is flipped, rotated, inverted, and color corrected, and saved out. There is no sharpening or post processing - it is not spotted, as it obvious. The file was shot with a Fuji 250mm f 6.3 on a Technika IV, Kodak Portra in a Readyload. It is Quebec in July, 2005.

[paulr]

the copy i downloaded looks crazy ... completely posterized, pixelated, and full of random noise. not sure if that's what it looks like on your end.

[Mick Noordewier]

Ed, this looks very nice. Paul, Ed has put a low-res jpeg version so that you can see the entire image without downloading a gigabyte file. However, it's not going to look that great if you view it at 100%. Look at the full-res TIFF crops to see the actual detail. I'm sure a drum scan would bring out more, but this makes me think I should consider replacing my 4870.

[Kirk Gittings]

"the copy i downloaded looks crazy ... completely posterized, pixelated, and full of random noise. not sure if that's what it looks like on your end."

Same here-it looks terrible. Must be some Mozilla wackiness.

[Kirk Gittings]

The details though look good, but I'm not sure what the point is here. It is a good scan from a scanner at this price level, moderately sharp and with a fair amount of shadow noise. Nothing unusual. There is none of the banding that I saw on the ones I returned, which is good. What are you trying to show?

[Kirk Gittings]

Actually I take that back about a fair amount of shadow noise. I hadn't looked at the tree detail. The noise is actually very good.

[Ed Richards]

> the copy i downloaded looks crazy

As the instructions on the page say - do not try to view these in the browser, download them and open them in PS.:-) They are not optimized for the WWW.

> What are you trying to show?

The interesting part is how I got here. To get the best quality, you have to scan at 4800 and downsample. This was impossible in the past because of memory limits, at least in the window's world. The newest Vuescan has improved the memory handling, and I combined that with Vuescan's ability to read in a raw file and process it as it were data directly from the scanner. (I think Siverfast HDR studio will do this as well.) This let me scan at 4800 and downsize it in the first pass, then load it into Vuescan to do the color and rotations and save the final file.

Being able to do this at 4800 really helped with the noise and the resolution. Remembering that this is color negative material and not Tmax100, the detail is pretty good. Good enough for at least a 16x20 print, and I think perhaps a 20x24.

As for the noise, while I have not tried it on 4x5, I bet Noise Nija would really clean up color noise. Anyone tried that?

[John_4185]

Sorry, but the jpeg looks pretty bad. The clouds and river, for example, look to be like GIF compression - gradient colors are lost. And I used CS to load it. That is, I did not use a browser at all.

(WindoZe people, you can paste the URL directly into CS's 'Open' dialog box to bypass browsers.)

Looks the same through a browser nonetheless.

Perhaps you could post a full resolution of a couple small sections.

[John_4185]

S O R R Y - You DID post detail sections!

My fault.

Looks pretty good.

[Kirk Gittings]

I own Noise Ninja and have worked with it extensively. Again the problem is that we are larger format guys who want detail. Noise reduction programs on scans all destry detail to some extent. The only place NN works for me is a very limited bit painted into the skies where detail is not important, but a slight Gaussian Blur will do that almost as well for free.

"and I think perhaps a 20x24"

That statement suggests that you haven't tried it. Ed, here is where we part company. You do very well at extracting a very good scan from the 9950 and Vuescan. But there is no way that that scan would satisfy my personal requirements at 20x24. I can tell by looking at the detail images. It would be marginal at 16x20. I have been around the block with this 50 times with my personal work and show prints. I have tried everything to get around this to avoid the expense of drum scans. That scan would not get me there. You shoot 4x5 and end up settling for a digital emlargement that is about the quality of a MF when you settle for anything less than a drum scan from a 4x5 at 20x24. Get a drum scan from someone who knows how to do them (like West Coast Imaging) from that negative and make side by side 20x24's the difference will be remarkable. I know, I have taken the time to do this side by side testing trying to avoid the expense on my recent retrospective show. I ended up having about 30 drum color scans done and did 20 B&W Imacon scans myself. I just finally had to bite the bullet and do it, but that was after having tested 4 9950F's, a 3200, two 4870's and two 4990's with numerous software and consulting with friends of mine like George de Wolf who is one of the leading experts in this business.