fooling the Epson scanner for the better?

Hi

 

recently I was thinking about how my Epson 3200 works, and thought that since

there is that area at the top which is used for 'sampling' the light I wondered

what the result would be if I covered it with a section of plain cleared

(un-developed) film. Sort of giving it a base line.

 

The results seem to be that I get a wee bit more out of the scan in what will

become the 'shadows' of the negative. Please refer to the image below for a more

detailed representation of how much is "a wee bit"

 

I haven't tried it with positive, but seeing as it makes no difference to the

dense areas I guess that it won't really make a significant difference.<div></div>

That is an interesting idea... one wonders, thus, what would happen if you placed a neutral

density filter on there. would it trick the scanner into being able to draw more out of the

highlights?

 

t

T:

 

presumably it depends on whether doing so would cause changes in the hardware (i.e. an increase in light output or increase in CCD gain) or whether the change would just be in the software (magnifying noise along with the signal). My guess is it's just software, but I'm welcome to being proved wrong.

I'm inclinded to agree with Jerry. The scanner CCD should have no trouble recording all the thin values on a negative. I know Imacon scanners produce somewhat better results with thinner negs. Something else is going on. One reason I like to shoot and scan negative film is for the wonderfully noise free shadows. I find it is easier to deal with noise in the highlights (e.g. sky/clouds) of negatives (assuming they aren't completely blocked) than pull shadow detail from a dense slide. I'm actually surprised you didn't get colored lines all over the scan by covering the calibration area!

Even Vuescan which manages to enable single-pass multi-sampling on Epson scanners (which not even the Epson drivers can do) doesn't seem to show any effect from changing exposure. I don't think these scanners have a true variable exposure.

Aaron

 

"I'm actually surprised you didn't get colored lines all over the scan by covering the calibration area!"

 

I think its because I used a sheet of even density material over the calibration area.

 

I know that there are no commands that can be sent to the scanner from an external source (such as vuescan), but I thought that the calibration area might work like 'auto gain control' in some tape recording systems.

 

I don't know if this is operating at the analog calibration phase or just altering the numbers, but the scan was still linear (gamma = 1)

 

I really don't know either, but it seems to work

That's actually a pretty old technique and I think one that only affects the scanner software- do a search here or at comps.periphs.scanners for more.

 

I'm doubtful you're going to get any benefit from doing this.

Roger

 

searching google groups, and confining to comp.periphs.scanners did not yeild so many things that covered the calibration area.

 

can you point me at something?

I'm not sure- it's been a while.

 

One thread that doesn't clarify much:

http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=00Bq10

Chris,

 

Can you post sample comparison images, one with and without the calibration tip? Is noise

and apparant grain reduced and/or more shadow detail derived?

Hi

 

ok ... I scanned a thin negative to see if I can see any differences in banding, but I can't find much differences.

 

posted a few samples below<div></div>

and a comparison of the image as it appears scanned with full range and gamma 1<div></div>

samples of the images for those who wish to play with them

lastly a comparison of the result of applying the same range of levels to each and some severe curves

 

I feel that there is a little better result in the covered scan<div></div>

upon an email request for trying this with colour, I put a strip of the same neg material (the leader, unexposed but developed) over the calibration area. I got surprisingly good results

 

again, I scanned as positive and this time colour. Comparing in with images scanned in 16 bit letting the epson software do everything itself (although using professional mode) I get images that are close to the same (after manipulation).

 

I am not sure if this is because the scanner software is applying changes in 16 bits based on the calibration or if there is some alteration of the electronic analog gain in the system.<div></div>

"I am not sure if this is because the scanner software is applying changes in 16 bits based on the calibration or if there is some alteration of the electronic analog gain in the system."

 

If you really want to play with exposure (analog gain) to find optimal settings I'd try downloading Vuescan and run some tests as it should give you direct control over exposure.

 

My guess with the above is that the scanner software is saying "calibration= white) and bumping up brightness slightly in software to compensate for the covering up. There don't seem to be any clear benefits in the above examples.

Thanks for posting this, Chris. Appreciate the hard work.

 

I wonder what the different shape and placement within the tonal scale of the two

histograms suggest. Squash and stretch usually indicates a gamma correction.

 

The output histogram is still in negative form since you scanned the neg as positive,

correct? Or is this just Auto Correction applied? Are you using color management-(ICM) or

just Epson's Color Control both selectable in Configuration dialog box?

 

I swear Epson's interface is still a mystery to me on what it does to the image.

Roger

<P>

you said:<i>My guess with the above is that the scanner software is saying "calibration= white) and bumping up brightness slightly in software to compensate for the covering up.</i>

<P>

and I'm inclined to agree with you

<P>

but vuescan does not provide any additional controls over the gain of the scanner in this instance (as I mentioned above)

Tim

<P>

you said:

<I>wonder what the different shape and placement within the tonal scale of

the two histograms suggest. Squash and stretch usually indicates a gamma

correction.</I>

<P>

I agree I think that something is happening. Now, its possible that

the 'calibration area' feeds into the same ADC stuff as the software

adjustment tool does (be it vuescan or epson).

<P>

Alternatively it could be that the calibration is a much simpler

'auto gain' control for the system providing measurement and feedback

to measure and cope with variations in the lighting which is just

a little too variable (not being LED) and optimise the scan. Epson do

adivse not covering this area after all.

<P>

<I>The output histogram is still in negative form since you scanned the neg

as positive, correct?

</I>

<P>

yes, that is correct, the histogram is as a negative.

<P>

<I>Are you using color management-(ICM) or just Epson's Color Control

both selectable in Configuration dialog box?

</I><P>

ICM

<P>

<I>I swear Epson's interface is still a mystery to me on what it does to the

image.</I>

<P>

I've never found it to be so strange. Since

moving to a 16 bit / channel work, I normally scan with the minimal adjustments

and then do the rest in photoshop. I scan into the working profile of proPhoto

and go from there. There seems to be no problems (or if there are I'm unaware

of them)

<P>

I have the feeling that once using a 16 bit workflow that the differences

between analog alteration and software alteration (especially for the range

of negatives) is minimal; particularly as the 'quantisation' of the

light levels by the ADC might be sufficient as to make less of a difference.

Bill

 

I think I misinterpreted your question. As I understand it the epson scan software doesn't tag the space of the scan, it has only the ability to show monitor colour (which is why I guess it gives you sRGB, colormatch RGB and Apple RGB which are all monitor spaces) As I understand it, it just sends to the application as however you've fooled with the data.

 

I disable all this with the settings in my configuration as in the screen grab below. Then I can use the histogram setting to keep the thing as 0 -> 255 and gamma 1.0 on all scans. I then fiddle (or not as the case in this exersize) in photoshop.

 

hope that helps explain my 'work flow' better<div></div>

Does moving that auto exposure level between low and high do anything useful? If scans take longer then it is doing something at the hardware level.

Roger

 

the auto exposure level seems to adjust how much the software tweaks the curves away from linear for you.

"the auto exposure level seems to adjust how much the software tweaks the curves away from linear for you."

 

Oh well, that's not very helpful.

<p>hello,<br>

i'm very sorry to bring up an old thread, but this seems to be close to what I want to ask.<br>

Lately I've been using V500 with Epson software to scan lots of negatives. I would normally do NO adjustments in Epson software, except for ICE, and do the levels adjustment in Photoshop. But now I've been thinking - does the level adjustmen in Epson software work on raw file from scanner, or does it send different commands to the hardware and thus produce different raw file? I'm asking this, because if it's the latter, then pointing software to correct range would produce much better output, then just using software method.<br>

Does anyone have any experience with this?<br>

Thanks!</p>