CanoScan 9000F vs. CanoScan FS 4000US

Discussion in 'Digital Darkroom' started by jdm_von_weinberg, Dec 22, 2012.

  1. CanoScan 9000F vs. CanoScan FS 4000US

    In a recent query about how the CanoScan 9000F flatbed film scanner compared to other scanners, the OP (Al N. at http://www.photo.net/digital-darkroom-forum/00b9Qj ) asked for comparisons and actual examples. I started to answer, but the answer grew into what would have been a annexation of the original post, so I decided to do a new post, instead.

    All I personally have to compare the 9000F with in terms of film scanners is my ancient CanoScan FS4000US which I operate on an equally ancient 400 Mhz PowerPC G4 (the infamous and rare Yikes! machine using a super SCSI interface and VueScan software. The speed of the scanner is fine, but the SCSI interface is rather slow. I have detailed my scanning adventures before at http://www.photo.net/casual-conversations-forum/00arR1 .

    00b9l6-509595584.jpg
     
  2. On the other hand, my CanoScan 9000F is working through USB2 on a iMac 2.7GHz Intel Core i5 machine. Somehow, not only is the 9000F faster, but everything is just peaches so far as handling the images goes.

    00b9l7-509595684.jpg
     
  3. So here are some images.

    First, I chose an old (1973) Kodachrome slide for scanning. This one has a 3rd Dynasty Egyptian equivalent of the famous "brick wall":
    00b9l9-509595884.jpg
     
  4. Here is the 700 pixel max of the entire slide - scanned with the CanoScan FS4000US at 4000ppi, Pretty much at the default settings on VueScan:

    00b9lA-509597584.jpg
     
  5. Here is a 700x700 crop (100% view) of the FS 4000 scan:
    00b9lC-509597684.jpg
     
  6. Here is the CanoScan 9000F scan as done by its own software at 4000 ppi with unsharp mask on but all other options off (no cleaning, etc.):
    00b9lE-509599584.jpg
     
  7. Here is the 100% crop
    00b9lJ-509601584.jpg
     
  8. Here with the unsharp mask off. as expected the result is less 'crispy' but not bad at internet sizes, I think. Still 4000 ppi

    00b9lM-509601784.jpg
     
  9. The 100% crop here shows the lack of the Canoscan software's unsharp mask.
    00b9lO-509603584.jpg
     
  10. Here is a CanoScan 9000F scan of the same slide at 9600 ppi. Reduced to 700 max for display here, of course; but the original file is 34.2 MB and 12736 x 8640 pixels in size. The unsharp mask is on as well.

    00b9lR-509603784.jpg
     
  11. Here is a crop (obviously a much smaller crop of such a large image) at 100%. Looking at this you can see why pixel-peepers think everything is unsharp.
    00b9lT-509605584.jpg
     
  12. I think that I am justified in using the older dedicated (not flatbed) 4000 for my best scans. However, as you can see in the examples, the 9000F is quite good looking with the Unsharp Mask turned on, and OK for internet images like these even with the Unsharp Mask off.

    On the other hand, aside from showing the small amount of CA on a slide shot with a PC-Nikkor 35mm f/2.8 lens on Kodachrome, there is no real reason to push even Kodachrome scanning up to the 9600 ppi that is possible.

    I emphasize that nothing was done to these images except for "spotting" out a couple of dust motes. Otherwise the images are as they came out of the scanners.

    I'd personally love to see some Nikon dedicated film scanners results at 4000dpi and unmassaged. I've always heard how good they are, but I've never had the chance to try one.

    That's it.
     
  13. As far as I know no flatbed scanner does better than 1700dpi, you might like to read this JDM:
    http://www.filmscanner.info/en/CanonCanoScan9000F.html
    Why don't you connect the FS4000US trough USB, at least mine can.
    I remember that the Canon 9000F delivered more natural colors after some warming up and calibrating, otherwise the scans get a red tint.
     
  14. Now I see, your Mac only supports USB 1.1?
     
  15. My new iMac supports USB 2, but the old CanoScan FS 4000 scanner only has USB1 - that's why I went to Fast SCSI on the old machine 400MHz PowerPC tower.
    You could die of old age waiting on the CanoScan with a USB1 connection. I can't imagine that anyone found it usable at that speed.
    I have looked for SCSI>USB2 or Firewire converters, but although some were once sold, none seemed to work worth a damn - at least according to everything I could find - and even then there are no NEW ones that I could find the last time I looked. Besides, it is not inconvenient to have a separate machine behind me to chug away while I work on the iMac (my main computer) See the earlier post linked to at http://www.photo.net/casual-conversations-forum/00arR1.
    Thanks for the link, but I do wonder if the 1700 limit is true. The 9000F certainly seems to produce better images than my CanonScan 4000 does at 1700 ppi.
    Give me a little bit, and I'll post a 1700 scan from the FS 4000 of this slide so we can see.
     
  16. Did you try multisampling/scanning with Vuescan as well? The 9000f software is pretty bad and I would turn all sharpening and adjusting off during the scan.
    You will also find reviews of the FS 4000US scanner there .
     
  17. Hmm.
    Here is the CanoScan FS 4000 scanned at 1700 ppi - reduced for display:
    00b9m1-509609584.jpg
     
  18. Here is a 100% of the above. Like the 9600 ppi scan, the original scale is different so 100% is larger. I have to admit, it looks pretty good.
    00b9m3-509609684.jpg
     
  19. To adjust for the smaller original size - here is a crop of 350pixels blown up.
    00b9m5-509609784.jpg
     
  20. I have to admit, maybe the review at http://www.filmscanner.info/en/CanonCanoScan9000F.html has a point.
    In any case, it's what I have for larger negatives, and even 1700 ppi equivalent works OK for posts like these on P.net. Perhaps I can speed up things even more by now scanning to start with on the 9000F at lower resolution. :|
     
  21. is fare and noise reduction off in vuescan by default?
     
  22. Did you see that filmscanner.info recommends scanning around 2500dpi with the 9000f for slides and printing purposes?
     
  23. Oh yes, I never, ever use FARE or noise reduction, I always turn them off as a result of bad luck with introduction of artifacts on Kodachrome. They are not used in any of the above scans on either machine with either the CanoScan software or the VueScan.
     
  24. I saw it, but I was looking for other things. I'll certainly try that resolution. Thanks.
    If you look at the results in the first pair (FS 4000 @4000) and CanoScan 9000F @4000 with unsharp on, they really aren't to far apart, however. Of course, that may just reflect the true resolution of the FS 4000, for that matter. :)
     
  25. Your welcome and I hope you also enjoy my Carl Zeiss SO lenses photo sample series in the coming days in the Pentax forum ;-)
     
  26. Looking at both at 4000dpi, the Canon FS4000US is showing you texture in the surface of the bricks that isn't there in the 9000F scan at any resolution. There's a pretty big difference between the two, which would be more obvious using a subject that had more detail to begin with (like a field or trees with leaves.) I've looked at a lot of slide scans and have never seen a FS4000US scan that didn't have more detail or at least film grain at a 100% crop at 4000dpi.
    I own the FS4000US and the Nikon Coolscan 5000. The Nikon has a modest edge in resolution, but much less shadow noise when scanning slides, faster times, better Dmax, and better color accuracy. You can get great scans from the Canon but it takes some post-processing and work-arounds. It struggles with contrasty slides.
    I recommend IR cleaning on for color negs and non-Kodachrome slides. It has negligible impact on image quality and saves a lot of time.
    I put some thoughts on noise with the Canon here:
    http://jingai.com/scanningguide/slide%20noise%20reduction.html
    Vuescan may well behave differently now- I haven't scanned anything in a few years.
     
  27. I do occasionally use the IR in VueScan for non-Kodachrome, but if you really clean well before scanning (which I really didn't bother much with here), I still feel manually "spotting" (as it was called in film days) is better.
     
  28. JDM,
    Thanks for doing this test. I know these can be tedious.
    I think Roger brings up a good point.
    Seeing actual texture in your scans with the FS4000 is what some other tests miss. I have seen other tests that appear closer in results but that is only from testing very high contrast scenes. They show detail in say a suspension bridge against a light background. Your scene shows a medium to low contrast image where detail can get lost with the flatbed scanner.
    I have a Nikon Coolscan V and an Epson 4490 flatbed. I will look for some slides to scan and post them tomorrow.
     
  29. That "1700ppi" claim was debunked by an optical engineer online, I can't seem to find the link at the moment... but Canon's own documentation and website claim the stated ppi is optical, not software created. I guess they could be lying by omission.
    For me, the end result is what's important, and maybe I have low expectations, but I seem to get great prints (8.5x11) and onscreen images from my 35mm film scans using the 9000F. I uploaded an example (with the white statue) to my portfolio. This was scanned to TIFF using the 9000F at 2400dpi, then sharpened in Aperture (Intensity 0.5, Radius 2.0), and exported as a jpeg. No other changes. How much sharper does it need to be? I don't print walls.
     
  30. I agree, Joel.
    For archive purposes, I think it's still worthwhile to use the FS4000, but I have only the 9000F for larger negatives, and frankly, it's still pretty good even for the smaller formats. Doing strange formats (24x24mm, stereo pairs, etc.) is much easier on the 9000F, as a rule.
     
  31. JDM, I scanned a few kodachrome slides with my canon FS4000US using Vuescan and infrared spot cleaning, and it worked.
     

Share This Page