Workflow for Color Negative scanning with Elite 5400 - comments?

[mendel_leisk]

Hi Marcos,

 

Well just by chance I checked on this thread the same day you posted. I agree we've run this thread to ground. I'll download your latest version of that action, and give it a try (though right now I'm getting really sick of color neg film).

 

Also, on the subject of color neg conversions, have you seen this thread:

 

http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=00D21a

 

I'm back to scanning old slides, which is a lot easier. Per the negs., I'm outputting 16 bit linear thru MSU, but just with autoexposure.

 

My biggest difficulty here is optimizing focus, with my old mounted slides. I've ended up doing manual focus on each slide, just prior to scan, with the manual focus knob. There are some benefits to this. You can do several focus attempts, and see where the notch on the knob is landing, for each.

 

I then use these as a Vuescan Raw File. I profiled my scanner thru Vuescan, with a Wolf Faust slide target, and am getting very nice results using the Vuescan workflow, coupled with the MSU 16 bit linear for source file. I'm on roll 2, only about 55 to go, so this will take a while.

 

I'm getting familiar with my 20D, taking a lot of pictures. Definitely the future for me, I love the immediacy. And the 24-70 f2.8 makes a very nice combo. I have a Pentax MZ-5n with a 50mm f1.4, but I don't see it getting much use now. Maybe some Velvia100F for fun, once in a while.

 

Hey, thanks for all your advice! You obviously run circles around me in the Photoshop department, and appreciate your guidance.

[marcos_rodriguez]

Hi once again, Mendel,

 

Well, looks like my quest for the perfect scan is far from over, but more about that in a minute.

 

First some comments about scanning linear with autoexposure on, as you've been doing. I made some tests and found it can actually worsen things --and this without taking into account that, when autoexposure is on, scanning times can almost triple, going from two and a half minutes per frame (at 1350 dpi,) to nearly seven!

 

More importantly, however, scans made with autoexposure on showed some bad shifts in color. For instance, I scanned some landscape images, and rocks had a nasty magenta cast, while green vegetation showed an unreal blue/cyan cast. When scanned without autoexposure, rocks had a nice neutral, perhaps a bit warm tone, and vegetation showed in a bright, yellowish green hue. Scans of other subjects showed similar results.

 

About scanning negatives, I share your feelings. Unfortunately for me, all the photos I've taken in my life are in negative film, so I'll just have to endure and keep on scanning.

 

I saw the thread you mention, and even downloaded the NegPos Photoshop Plug-In. I gave it a try and it looks promising, however I found it a bit awkward to use. Moreover, settings had to be tweaked individually for each image in order to get good results, something which, given the large amount of files I need to process, for me it's a "no go."

 

Now to the bad news: unhappily (I say "unhappily" because, just when I thought I was finally getting acceptable results with the Photoshop action, this makes me re-think the whole thing) three days ago I tried SilverFast DCProStudio --and loved it.

 

A friend of mine who recently got it had been recommending it to me, and I finally listened to him and downloaded a demo. To wit: The results were fantastic, especially in the gamma correction department, where SilverFast's sophisticated curves provided the best results I've seen.

 

Now, Silverfast alone didn't do the trick, but it provided excellent "basic" files for final correction in Photoshop where, among other things, I applied the Minolta regular profile, which gave the perfect touch to an already fine scan (the profile lightens the shadows a bit, makes the skies bluer and gives greens and yellows a fantastically warm, pleasant hue.

 

So, as I said, having tried SilverFast made me rethink the whole workflow, and now I'm doubtful that the action I've been using (and recommending) is the best method --actually, I think it is not, at least for most exposures.

 

So, lacking SilverFast (I don't plan on spending that kind of money), I think the best "generic" method is:

 

1. Scan 16bit linear, with autoexposure off

 

2. Apply the linear profile while opening in Photoshop, and immediately convert to ProPhoto (much wider gamut than Adobe 1998)

 

3. Invert

 

4. Crop to eliminate the white edge, so that "autocolor" can work its magic

 

5. Use Curves' autocolor ("Enhance Per Channel Contrast", with "Snap Neutral Midtones" checked)

 

6. Correct gamma with curves again

 

Talk about going "back to basics"!

 

Best,

 

Marcos

[mendel_leisk]

Hi Marcos. You're color shifted 16 bit linears, were they neg scans? I'm assuming so. Minolta actually suggests not to "bother" with auto-exposure with slides. It is "off" by default. I'll look into that. Auto exposure seems to be behaving, but perhaps it varies, slide to slide.

 

Good luck with Silverfast. I'm curious, I believe it does employ proper ICE: does it allow you to decouple the grain dissolver?

[marcos_rodriguez]

Hi Mendel. Yes, the color-shifted scans I was referring to were from negatives. In fact I decided to make the test following your comments:

<i>Per the negs., I'm outputting 16 bit linear thru MSU, but just with autoexposure.</i>

<p>

About Silverfast, the version I tried (DCProStudio) does not do any actual scanning, but rather it works directly with the files from the hard drive --which, in my case, were produced by the MSU scanning as 16bit linear, with ICE on.<p>

 

SilverFast does have a scanning version for the ScanElite 5400, which I tried shortly I first got the scanner. I remember it produced good results (much better than MSU's, with cero clipping,) and with outstanding implementation of ICE --a clear contrast with VueScan. However, given its high price I quickly took it off my mind. Oh, and it does allow de-coupling of Grain Dissolver, but the scans took nearly the same time and grain visibility was greatly increased, so I would not advise this path.

[mendel_leisk]

Hi Marcos. Well I tried a slide scan with/without auto-exposure, outputting 16 bit linear both times. As I'd noticed in past, the main noticeable difference was additional compression of the histogram towards the left, with no pixels higher than 200~225. With auto-exposure, there are pixels virtually all the way to 255 in green and blue, and slight clipping of red at 255.

 

As far as color balance, I could see no difference. Accordingly, with slides, I see no reason to not use auto exposure. For today ;)

 

I think you are right about decoupling ice/gd, and the excessive grain. It's nice to try though. I believe just using grain dissolver tends diminish dust and scratch appearance, by it's nature.

 

I'll look into the Silverfast version you mention.

 

Well, I'm sure this is not the last of this thread. Probably not the longest on photo.net, but quite long, none the less. Lot of ideas!

[gilou]

Hi Mendel, Marco,

 

I read your post with a lot of interrest ... I've a Minolta5400 and many negs to scan.

 

I clearly understand the position of Marco with its last workflow (or using Silverfast), however I would like to know Mendel what is your workflow .... now.

 

On my side I was mainly using vuescan for negs and the DSU for slides.

 

Brgds -Gilles.

[mendel_leisk]

Hi Gilles,

 

This is how I'm doing things with my current slides scanning project:

 

****

 

Minolta Scan Utility settings:

 

*auto expose for slides: ticked

 

*manual focus with knob at time of each scan

 

*output 16 bit linear

 

----

 

Vuescan (used for finished files, scan-from-disk with above):

 

Vuescan version: 8.1.29

 

Startout settings:

 

Input|Media: Image

 

Input|Mode: Transparency (setting to Elite5400 had NO effect)

 

Color|Color Balance: neutral

 

Color|Black point (%): 0

 

Color|White point (%): 0

 

Color|Brightness: 1

 

Color|Scanner Color Space: ICC Profile

 

Color|Scanner ICC Profile:

 

E:/targets/icc profles/Fuji Provia, Sensia, Astia_F040521.icc

 

(Above file produced Friday, April 22, 2005 (10:28pm) by physically scanning the Wolf Faust target in Vuescan. Note, this icc profile seems to work fine regardless of slide film type, so far.)

 

****

 

And if you're still with me, this was my final workflow for scanning color negatives:

 

****

 

MINOLTA SCAN UTILITY:

 

-scan as color slide

 

-uncropped

 

-slide auto exposure off (see exposure control settings below)

 

-16 bit linear

 

-manual focus (with front knob) on smooth, fairly light, grainy area of interest, about 1/3 to 1/2 the distance along diagonal line between center and one of the upper corners of image. Do the focus PER FRAME, and scan immediately. Do not gang-focus all the frames first. Verify focus accuracy is satisfactory before going to next frame, viewing 100% in Photoshop. NEAR Corner-to-corner sharpness IS achievable with relatively flat film strips.

 

Exposure control:

 

-mas: +2.0

 

-red: -1.3

 

-gre: -0.2

 

-blu: +0.8

 

 

Photoshop:

 

*Clean and crop

 

Then, on a copy:

 

*Assign Minolta's posi linear icc profile, then convert to Adobe rgb

 

*adjust. layer: invert

 

*ajdust. layer: levels:

 

find and dark colors

 

snap neutral midtones: no

 

.01bp

 

.01wp

*ajdust. layer (luminosity mode): Curve: (normal mode)

 

198>186

 

*adjust. layer: hue/saturation:

 

saturation: +20 <maybe a bit toooo much, reviewing later

 

*flatten

 

*convert to 8bit

 

****

 

Please comment, and if there's any blanks you'd like filled in (obviously there's a ton more settings), just ask.

[gilou]

thanks for your feedback meindel.

I've not scanned for more than 2 weeks now.... i'm working on my website !

however I've print this threads I will try to follow you process when I will need to scan another time (very soon).

 

Brgds -Gilles.

[mendel_leisk]

Giles, one thing:

 

The object of those settings in exposure control was to get all 3 levels close to 255, even have a few "stray" pixels at 255. This was determined scanning the leader. So some experimentation is necessary to really optimize.

[to-mas]

Hello,

I'm reading this tread for long time course I'm very interested.

I own a Minolta Scan Dual IV, it lower model that yours 5400, but anyway i have the same problems with scanning.

I finally find some time and i tried Marcos work flow (i dont want to spend money on vuescan) and I'm still not sure with some things.

So i would like to ask about some details, to i can write whole process step by step.

 

--

1. Scan 16bit linear, with autoexposure off (by Marcos) or On (by Mendel)

 

2. Apply the linear profile while opening in Photoshop, and immediately convert to ProPhoto (much wider gamut than Adobe 1998)

-with apply you mean assign profile?

I have 2 minolta profiles:

DIMAGE Scan Dual 4 and DIMAGE Scan Dual 4 (Posi LInear) should i use posi linear?

 

3. Invert

 

4. Crop to eliminate the white edge

 

5. Use Curves' autocolor ("Enhance Per Channel Contrast", with "Snap Neutral Midtones" checked)

 

6. Correct gamma with curves again

--How exactly you correct gamma with curves? I was using levels for this and curves just for contrast.

 

thats it?

i dont forget something?

 

I'm looking forward for answer, anyway i would like to thanks for whole this tread course its the most useful what i found on Internet.

[to-mas]

yeah,

i would like to ask why is that photoshop action so complicated when this workflow look more easy?

 

or Marcos do you use this action after that 6steps?

 

sorry if i missed something, i just get out of mind from scanning.

[to-mas]

sorry but i find other thing

do you use corol matching in MSU ?and if,which profile?

[mendel_leisk]

Hi Thomas,

 

Thanks for posting. I'll have a more thorough read tomorrow, and respond (it's late, getting sleepy).

 

One thing, with color neg scans I did have autoexposure off. I've actually just started up again, this time with lots of old films. At the scanning stage, the most important thing for me is to test scan the leader with MSU, and get all 3 channels so there is little or no pixels by 255, but very close. Then output 16 bit linear, of a few real frames, and verify levels in ps.

 

Suprisingly, I'm able to get pretty good color balance with these older films, thru Vuescan scan-from-disk.

 

Your choice, but it's not much to spend, getting a Vuescan Pro License. And it's good for life. With me it's love/hate, particularly the infrared cleaning, which is why I acquire the "raw" file thru MSU.

 

I'm just experimenting still, like a lot of people, I think.

[to-mas]

hello,

and how is the process to get all 3channels close to 255 ?

i was thinking that when you scan on 16linear you can not adjust levels and etc.

i go to play with that.

[mendel_leisk]

You definitely _can_ adjust exposure levels. I was more-or-less following Bart's guidance here:

 

http://groups.google.com/group/comp.periphs.scanners/browse_thread/thread/42c3b2fa6068fc62

 

with the exception of not having "auto exposure for slides" ticked in preferences.

 

So, with auto exposure off, I:

 

1. Prescan a frame having just leader.

 

2. Go to the Exposure control tab, and observe the red/green/blue values while moving the mouse around the image.

 

3. Adjust master, red, green and blues sliders, hit the "refresh" button. Move mouse around some more to see the result.

 

Typically, I would move the master slider up to raise them all, then move the red down, and the blue and green up, some more, aiming for around 250. You can then click "apply to all frames", and also save the settings, for convenience.

 

The down side is increased scan time, usually governed by the longest, blue exposure.

 

Open 16 bit linear scans, with different exposure settings, to verify your levels are getting you just under 255. A few pixels over is ok.

 

You can do similar but with Auto Exposure on. Then the Exposure Control sliders shift the exposure from the auto exposure base.

[to-mas]

Hello,

i already find that i can correct exposition (master, R,G,B), but not levels in 16linear mode.Actually i can but change levels settings in image correction makes no difference.

Anyway, so i was trying adjust RGB exposition to get the best values close to 255 but not too much, in all 3 channels. But i found it very complicated and setting was totally different for each different light .

I was also trying play with auto exposition off and on but i found that autoexpositon makes too big differences so its not good for batch scanning and adjusting what i want finally get to.

So after few hours playing with that i left it and scaned around 12frames of superia 200 with this settings:

corol positive

autoexpo off

correction non

16bit linear

 

Than i was trying to use Marcos method

 

-- 1. Scan 16bit linear, with autoexposure off

 

2. Assign the linear profile DIMAGE Scan Dual 4 (Posi LInear) in Photoshop (i'm scanning directly to photoshop CS), and immediately convert to ProPhoto

 

3. Invert

 

4. Crop to eliminate the white edge

 

5. Use Curves' autocolor ("Enhance Per Channel Contrast", with "Snap Neutral Midtones" checked)

6. I dont correct gamma in curves becouse i dont know how.

 

anyway the results what i get

 

some fine, mostly too saturated (this i think have something with icc profiles, I'm working in adobeRGB, if i assign AdobeRGB saturation comes down), without contrast.

So nothing great.

 

I also had tried that last action what Marcos post,by mostly why are there so many steps and the results was mostly very similar.

Now i quickly tried your method, but i dont understand some parts>

 

And if you're still with me, this was my final workflow for scanning color negatives:

 

****

 

MINOLTA SCAN UTILITY:

 

-scan as color slide

 

-uncropped

 

-slide auto exposure off (see exposure control settings below)

 

-16 bit linear

 

-manual focus (with front knob) on smooth, fairly light, grainy area of interest, about 1/3 to 1/2 the distance along diagonal line between center and one of the upper corners of image. Do the focus PER FRAME, and scan immediately. Do not gang-focus all the frames first. Verify focus accuracy is satisfactory before going to next frame, viewing 100% in Photoshop. NEAR Corner-to-corner sharpness IS achievable with relatively flat film strips.

>>>i use auto,but that is not a point

 

Exposure control:

 

-mas: +2.0

 

-red: -1.3

 

-gre: -0.2

 

-blu: +0.8

 

Photoshop:

 

*Clean and crop

 

Then, on a copy:

 

*Assign Minolta's posi linear icc profile, then convert to Adobe rgb

 

*adjust. layer: invert

 

*ajdust. layer: levels:

 

find and dark colors

 

snap neutral midtones: no

 

.01bp>>> whats this?

 

.01wp *ajdust. layer (luminosity mode): Curve: (normal mode) >>> whats this?

 

198>186 >>> whats this?

 

*adjust. layer: hue/saturation:

 

saturation: +20 <maybe a bit toooo much, reviewing later

 

*flatten

 

*convert to 8bit

>>

excluding that last steps, which i dont understand (i just put adjust layer levels -find and dark colors-snap neutral midtones: no )

i get with this method a little bit better results.less saturated and more neutral colours.

But i tried it just on one frame.I'm going to play with that now.

 

I also tried to scan blank holder now , like slide, autoexpo off, 16bit linear and i get 100perc pure white. R G B 255 without any compensation.

 

 

Also i would like to know if you use corol matching in MSU ?and if,which profile?

 

 

Sorry for so many questions (and my English maybe too), but i'm fighting with that scanner my whole day off (excluding that few hours before, i had this scanner 3months already) and going crazy.

[to-mas]

i now found that i need to corect empty strip of negative not empty holder without everything.i go to play with that.sorry for spaming

[mendel_leisk]

Yeah, that is the key: previewing a piece of the leader. This is how you neutralize the background color of the film. I'll look over your other questions, regarding my excessive shorthand, mostly, when I get home from work. Have to run!

[mendel_leisk]

.01bp = .01 black point, in Photoshop levels dialogue

 

Curve (normal mode) = Setting when initiating a Curves adjustment layer. Marcos advocates luminosity.

 

198>186 = an adjustment point in the curves dialogue, in other words 198 in to 186 out.

 

saturation adjustment : I've been skipping that now.

 

By the way, I'm finding I get much better highlight tones, just taking the 16 bit linear over to Vuescan, experimenting with different color neg profiles it provides. Generic color negative not bad. This is similar to my experience with Tri-X. Working the raw file in Photoshop, I was getting washed-out highlights. Vuescan, on the other hand, recovered a lot of subtle shading there.

 

I suppose there's nothing magical in Vuescan, with more skill I could do the same thing in Photoshop, with curves I think. But in the meantime...

[marcos_rodriguez]

Mendel, Thomas,

 

I happened to check this thread and saw it was still alive, which is probably a record :o)

 

I feel obliged to tell you that, after much experimenting and tweaking, I have given up on trying to achieve good results with Photoshop alone. I'm afraid that the curves, color balancing and gamma corrections involved in successfully inverting a negative image are too complex to be achieved in PS. Therefore, my recommendation (for negatives) is to scan 16bitLinear (without tweaking the exposure tab, and with ICE on) and then post-process the raw images in either VueScan or SilverFast HDR, depending on one's budget.

 

VueScan does not support ICE technology, hence the need to use Minolta's software for the raw image. With SilverFast, considering it is such a hefty investment, I think the best option is to buy the generic HDR version which, albeit more expensive than the ones tailored for a specific scanner, has the added value of working with any number of files (independently of the scanner that was used to obtain them) therefore offering more flexibility for the future.

 

Anyway, those are my two, probably last cents on this fantastic thread,

 

Marcos

[mendel_leisk]

Hi Marcos. Your guidance is what I did lately with Some decades old Kodak Gold: output 16 bit linear thru MSU, then scan-from-disk thru Vuescan. Worked quite satisfactorily. Same workflow with Fuji Reala gave me some problems with color cast, which I haven't resolved, to date. So anyways, works quite well with my old stuff.

 

Haven't done any color neg recently though, I'm pulling my hair out trying to get corner-to-corner sharpness, with mounted slides (or whatever). I've sufficient new mounts on hand to remount the lot, and am experimenting with various mounts that are supposed to help in this regard, with and without glass. Also, considering "home-brew" mount, glass or not.

 

Also considering alternative scanner that might have better depth of focus, but suspect that does not exist, short of REAL high end Imacon/Hasselblad etc.

 

You know, I think there's a few other long-term threads, maybe longer than this one, but still, I bet we're "in the running" ;)

[mendel_leisk]

Did the Kodak Gold 16 bit linear with my usual manual exposure set with scan of leader.

[andrew gundershaug]

Hello Guys...first of all I would like to thank you both for all of the wonderful information that you have both contributed towards my understanding of the Minolta 5400 and the scanning process. I purchased a 5400 II a few months ago and I have been struggling (like everyone I now realize) to obtain high quality results with my scanner. I have read as many posts as I can and this has been one the most helpful. I have adopted a workflow for color negatives that resembles the one that you describe above using a blank frame to adjust the RGB channels in MSU to compensate for the Orange cast. I haven't made the jump to 16bit linear but I plan on giving it a try in the next few days. What workflow have you two adopted for B&W NEGS? I have just purchased VUESCAN and I have begun to tinker with the B&W workflow that VUESCAN describes. Up until now I have been scanning B&W negs with MSU as Color Positives (16bit), Autoexposure:ON, completely desaturated (as described in the manual for chromogenic films). Open the file in Photoshop, Invert, convert to Grayscale, adjust:levels, adjust:curves, and I have come up with a pretty good image. Are you both using VUESCAN for B&W negs...what are some benefits that you see using VUESCAN?

 

Thanks in advance...

 

Andrew

[marcos_rodriguez]

Hi Andrew, many thanks for your comments, it's nice to know someone else might have benefited from this long and interesting thread.

 

The entirety of my archive consists of color negatives, so I'm afraid I cannot be of much help as far as your question is concerned. Nevertheless, based on what I've read on the Internet, VueScan seems to do a great job on B&W film, so that would be the venue I'd take.

 

As for the benefits of using VueScan (or SilverFast), as I've said above, it would be the proper gamma correction, as merely inverting the image in Photoshop and adjusting Curves and Levels is not enough, given the complexity of the algorithms involved.

 

Best,

 

Marcos / www.aukeramen.com

[andrew gundershaug]

I have another question...I have started using Vuescan in my post-processing workflow...when working with slides I am using ICC profiles that I have created using Wolf Faust targets for Provia, velvia, etc...but here is the question: I created the profile in Vuescan by scanning the slide via the Vuescan scanner driver, I then scan my slides in MSU and create a 16bit linear tiff file, then work the file in Vuescan with the "scan to file" option. Would it be better to Scan the target in Linear mode in MSU, and then post- process the "target" file in Vuescan and create the profile that way...I am going to give it a try to see if it makes a difference in my color reproduction. I assume that it may when you take into account the GEM and ICE tools that I am using in MSU and the effects that they will have on color rendition...

 

Andrew