Scanners for negatives

Posted by: Arye_Gur on 23 October 2004

I must confess that although I wrote here several times that I don’t see a difference between pictures that were taken by a Digital cameras and between pictures that were taken by SLR cameras, I leaved the Minolta z1 and I continue to use my Minolta 700 si SLR camera. This is most of all because that the act itself of shooting the pictures is much better to me with the SLR than with the Z1.
When it comes to viewing the pictures, I like the idea of viewing them on the computer screen rather than printing them on a paper.
The problem is, that scanning the negatives at the shops that developing them, don’t bring the expected results. I have a relative who is expert in the subject of pictures on computers and TV, he examined the scanned pictures and told me that it is obvious that the scanning is poor. He showed me that the dynamic range is poor and that the Black and the white are not as black and as white as they should be. He could easily improve the pictures, but still, this is not this.
The solution, as I understand it, is a negative scanner. From reading articles in the internet, I see that there are many problems involved in negative scanners – some are because of the nature of the negatives and some are because of the technical requirements needed from the scanners.
What do you know about the subject? Do you have experience with it? What do you think is better – going for an expensive DSLR or staying with the SLR and going seriously for negative scanning?

Arye
Posted on: 23 October 2004 by Geoff P
I disagree with both comments in some part.

I had an Olympus dedicated negative scanner that did a far better job than my HP flatbed with a negative adapter in terms of resolution in the resulting file. The scanner had softwrare which allowed pre-set color / contrast recipes for different film makes (eg Kodak & Fuji) and it was possible to tweak indivdual scans beyond that. The drawbacks to the negative scanner were supersensitivity to dust which is easy to pick up when loading negatives, which is also a pain BTW, it was slow at high resolution and the color recipes were very standard. Any photography in unusual lighting conditons or with unusual contrast ranges needed quite some tweaking in post processing on the computer

I now have a CANON 300D which is a very reasonably priced Digital back for CANON EF sereis SLR lenses ( which I have already from my Canon SLR Negative camera). This has a 6.2 Megapixel working area and generates RAW images if you want. IMO the RAW images have a least as good resolution as anything I could get off a negative scanner and it is easy to change white balance and color/contrast etc on RAW images in a Batch mode .
The typical storage cards available now go up to 1 Gb which is plenty enough for a RAW format photo session.The RAW files also contain an imbedded jpeg which can be extracted and processed as a good A4 picture.

I have even used a good widefield MACRO lens I have, to re-photograph old color prints. On the 300D this is easy and gives files which have better resolution than I can get by scanning on the flatbed at high res.It is about 10 times faster aswell and you can post process the RAW images on your computer then once happy ditch the RAW files so you don't eat up unnecessary dsiskspace.

regards
GEOFF

"Just trying to make a NAIM for myself"
Posted on: 23 October 2004 by Don Atkinson
Pixels v DPI (Dots Per Inch)

I have had difficulty understanding the relationship(s) between pixels (as in 6 mega pixel cameras) and scanning density (as in 2400 dpi).

I have also wondered whether each pixel or dot is a simple binary (on/off) characteristic, or whether each pixel or dot is a more complex (byte?) characteristic with (say) 8 or 16 (bits?) of data.

Any definitive answers?

What I have noticed is that my flat-bed scanner (Epson GT-7000) with or without the transparency adapter is naff. i.e. scanning either negs or prints is a waste of time.

I take my film (Kodak/Fuji ex Canon EOS100) to Happy Snaps (this is probably the limiting factor!)

I normally get (Happy Snaps) to make a CD at their 4.5 Mb density and use this to generate A4 prints at home (Epson Stylus Photo 950)

If anything looks like its worth enlarging more than A4, then I take the neg and get it photographically (rather than digitally) enlarged and printed.

I had thought of getting a Coolscan negative scanner. But realised I would still need to get others to create anything bigger than A4. I am unclear whether a Coolscan for say £400 would be any better than a 4.5 Mb density image from Snappy Snaps?

Cheers

Don
Posted on: 23 October 2004 by Joe Petrik
Don,

quote:
I have had difficulty understanding the relationship(s) between pixels (as in 6 mega pixel cameras) and scanning density (as in 2400 dpi).

I have also wondered whether each pixel or dot is a simple binary (on/off) characteristic, or whether each pixel or dot is a more complex (byte?) characteristic with (say) 8 or 16 (bits?) of data.


I was going to write a detailed reply, but a quick search turned up this site, which does a good job explaining the difference between dpi and ppi, as well as pixel depth -- 8-bit vs 16-bit, for instance.

Joe
Posted on: 23 October 2004 by Julian H
Ayre

If all you are doing is viewing on-screen you might as well buy a digital camera (the 3MP Canon A75 is an excellent choice).

It will pay for itself within a very short period of time because of all the savings you will make on processing.

Julian
Posted on: 23 October 2004 by Geoff P
quote:
I have had difficulty understanding the relationship(s) between pixels (as in 6 mega pixel cameras) and scanning density (as in 2400 dpi).

I have also wondered whether each pixel or dot is a simple binary (on/off) characteristic, or whether each pixel or dot is a more complex (byte?) characteristic with (say) 8 or 16 (bits?) of data.

Any definitive answers?


OK here goes.

The size of the 6.3 million pixel imaging area in my Canon 300D is 22.7 x 15.1 mms. By a bit of mathematics that gives the approximate size of each pixel (or "dot" if you like) is 7.2 microns square. There are 1000 microns to the millimeter so there are 25,400 microns to the inch/ 7.2 gives 3528 pixels. So a rough comparison is that the Canon camera I use "scans" pictures with 3528 dots to the inch on your scale of 2400 dpi.

The pixels are actually light sensitive transistors arranged in cells in a chip that is very similar to a CMOS Memory chip. Each cell also has amplifying transistors. The light intensity falling on each pixel is graduated electronically by a "grey scale" algolrithm so there are levels or "bits" if you like.

There is one important thing that applies to both scanners and pixel imagers (and also to color negative film for that matter) which needs to be considered. Just as in a TV the dots are matrixed so that there are Red, Green and Blue dots which combine to create colored images.Typically there are 2 Green dots for every 1 Red and Blue dot because the human eye needs more intensity in the Green to see a balanced set of colors (known as a Bayer pattern).

So in theory the resolution of a colored image scanned with a certain number of dots per inch is really only a thrid or a quarter of the quoted number. Example 2400 dpi is really more like what would be about 800 dpi in black & white. In practice because of inetrpolation between pixel densities the probable resolution to the eye is only halved to about 1200.

I think I've got that right but I welcome correction.

regards
GEOFF

"Just trying to make a NAIM for myself"
Posted on: 23 October 2004 by Geoff P
Don & Joe

I will now read your link to see how wrong I was when I drafted the above reply.

regards
GEOFF

"Just trying to make a NAIM for myself"
Posted on: 23 October 2004 by Don Atkinson
Many thanks Joe,

Its going to take a while to read and understand, but the first couple of pages have helped to refresh my memory and cleared up a couple of important points as well. It seems to be a comprehensive exposition.

Glad you didn't finish typing first, THEN find this site....

Cheers

Don
Posted on: 23 October 2004 by Don Atkinson
Geoff,

What you wrote about pixels/dpi is more or less what I thought I knew, but I wasn't sure. Joe's link seems to confirm this, more or less, so you won't feel embarrased (on this forum?) The bit about 2xgreen to each red/blue was new to me.

My video camera is a 3ccd model ie each ccdcharged couple device? is dedicated to either the R G or B colour. The salesman said it was better this way.....

Cheers

Don
Posted on: 23 October 2004 by Geoff P
Hi Don

I'm in the semiconductor industry which is why I'm a bit of a "nerd" on the IC side of this.

Most (not all) of the SLR and larger format digital camera backs are CMOS Image Sensor chips which allows image processing electronics to be integrated. It is also possible to correct for small variations in the electrical sensitivity of individual pixels since they are individually amplified in the CMOS chips.

CCD (charge Coupled devices as you rightly say) use columns of pixels and store the electric charge generated by light in charge wells. Stronger light signals are sent further in, into deeper wells. The individual charges are read out at the bottom of each column and passed to a seperate processor chip. It is harder to recalibrate CCD image arrays so they are more fussy to make, but they have higher light capature capability than CMOS IS chips.

Video cameras typically use CCD's because they need the higher light sensitivity from them than from CMOS chips. Using one for each color is a nice way of enhancing image resolution. The output of the individual colors is probably "balanced" in the processing electronics.



regards
GEOFF

"Just trying to make a NAIM for myself"