epson 4990 scanner

[John Berry ( Roadkill )]

I just purchased a new epson 4990 scanner. It replaced a 2450. I am also now using silverfast Ai. I can now scan my 8X10 negs. Scan at 1600 PPI 16 bit greyscale, 7.6 X 9.7 inch scan area =359 meg file. Time thru firewire is 4.25 min. My negs are keyed to new Azo #2, so they are pretty contrasty. Seems to handle them ok. Alot of members state thinner negs scan better. My 4X5 negs are keyed to a condenser enlarger. I'll see if they do better. I'm just playing with the stuff I couldn't before. I'm pretty happy so far.

[Sam]

Thanks John, I think Ive made up my mind on the 4990 to scan my 120 and 4x5's. I would like to see what you are capable of with it. posta few images if you get a chance.

SP
www.samuelportera.com

[John Berry ( Roadkill )]

I will later. If I posted some now you would only see what "I'm" able to do not the scanner. Not necessarally what it CAN do. Let me get up on tweeking Silverfast Ai first.

[John Berry ( Roadkill )]

The scanner comes in two versions. The pro version comes with silverfast Ai and monaco color management. Best way to go if you don't have them. I already have color management so I got the non pro version, then upgraded the silverfast se to Ai. saved only $20 over buying the full version. I should have spent another $ 50 and got the pro version then a friend could have gotten the color management at a treat price.

[M Brian Mills]

All of the images on my Portfolio #3 (http://www.brianmills.us/portfolio/index_02.htm) page and all but 4 of the images on my Portfolio #2 (http://www.brianmills.us/portfolio/index_01.htm) page were scanned with an Epson 4990 using firmware. The 4 images on Portfolio #2 that were not scanned with the 4990 were scanned with the Epson 3200. Can you tell a difference? I can't tell a difference and I have the original scans so I can get deep into the detail. All images were scanned from 4x5 chromes at 2400 dpi.

[Kirk Gittings]

John,

I have been using the 4990 since it first came out and the 4870, 3200 and Canon 9950f also before that. The advantage to SF AI is the ability to really tweak the curve and extract as much from the neg as possible in ways that is simply not possible with the Epson scan software. I have been using it and testing it against other software as well and I find it superior. I don't try for a thin negative but a slightly flat negative with substantial shadow detail (Zone III placement) and no blown out highlifghts. With that negative I tweak the curve to bring it in the direction of the tones I want in the final print but leave it a bit flat still and unclipped. For the last month I have been using an Imacon at the School of the Art Institute of Chicago where I teach in the summers. Unfortunately SF is not available for the Imacon and I really miss it, as SF would really improve that scanner as well.

Try scanning in 48 bit RGB and discarding all but the blue channel (which you then convert to grayscale). The blue channel because of the shorter wave length is sharper than the green and red channel and has less noise than the red. The result is a much sharperand less noisy scan than a 16 bit grayscale scan.

[John Berry ( Roadkill )]

Thanks Kirk, Brian, and Sam. I also scan B&W in 48 bit RGB mostly. I select a channel or use the channel mixer before conversion. I just wanted to see what it will do in greyscale. I do really like having a curves adjustment in SF. I have some negs that are dry that I souped up earlier today. Acros in Pyrocat for my condencer enlarger. I'm going to give them a try next.

[Murray Fredericks]

How are you all getting on with the 4990?

I am thinking of buying one for proofing and test printing my 8 x 10 colour negs - it will work out cheaper than contact proofs from a lab in the long run!

I plan to have exhibition quality drum scans made at a later stage when the work is edited. Are the scans sharp enough and generally good enough for making decent 16" x 20" prints? Are there any noise problems?

Cheers

Murray

[Kirk Gittings]

Murray,

The 4990 will get you a decent scan for up to 16x20 UNLESS you have an color image with very large important shadow areas then I spring for a drum scan. Interestingly the 4990 has less noise in color than the Imacon I have been using here in Chicago.

[Brian Ellis]

For those of you who scan b&w in 48 bit rgb, I'm curious as to what film sizes and at what ppi you're scanning. I usually scan 4x5 at 4800 ppi in 16 bit gray scale and my initial file size is over 700 mgs. So an rgb scan at 48 bit would be something over 2 gigs and there's no way I could do that. So what size film at what ppi are you rgb guys scanning ?

I vacilated back and forth for a couple years between scanning in grayscale and scanning in rgb. I'd read something that said one way was better and I'd use that for a while, then I'd read something that said the other way was better and I'd use that for a while. I really didn't see any difference in my scans either way. I finally settled on grayscale because John Paul Caponigro said he didn't usually see any difference either and I figured he should know what he was talking about (www.johnpaulcaponigro.com/faq). Not the best of reasons but since I wasn't seeing any real difference either way I had to pick a method on some basis and that's the one I chose. Now you have me wondering again.

[John Berry ( Roadkill )]

For a sample here is a post from another site I visit :
http://groups.msn.com/Asktheoleproaboutphotography/general.msnw?action=get_message&mview=0&ID_Message=210124&LastModified=4675533776554029691

[John Berry ( Roadkill )]

I sacn 4X5 at 2400 PPI 48 bit RGB = 560 meg. I have JPC's master class book and he recomends scanning in RGB and doing what Kirk said earlier in this thread or dropping each channel on the image as a layer and masking to use different channels for different parts of the image. I did that with one B&W neg and had a 3.4 gig file to start mixing with!

[Julio Fernandez]

John: Silverfast 6Ai is also in my experience great software. It enables you to extract maximum information before you edit in Photoshop for after all, Photoshop can only work on what is in the scan. That said, Silverfast can only extract what the scanner sees. What any scanner will see in dry mounted film is considerable scattering of the light which emphasizes imperfections and diminishes contrast and brilliance. That gives drum scanners the optical edge, because they always are used with wet mounting. If you want to take the 499o or any other flatbed scanner to the max use it with mounting just like a drum scanner. You can learn more about wet mounting in the yahoo group
http://groups.yahoo.com/group/SCANMAX/

[David R Munson]

I just bought one a few weeks ago and so far I couldn't be more pleased. I've been getting great results from it on both color and B&W, both with thinner and denser negs. Everything here (http://flickr.com/photos/70117898@N00/sets/638629/) was scanned on the 4990. I just wish I had more time for scanning right now, because it does a great job. It's putting new life into some of my old work that, until now, I only had mediocre scans of (at best).

[Brian Ellis]

John - Thanks for the response. I have JPC's Master Class book also and don't remember reading anything in it like Kirk suggests (as always, my memory certainly could be faulty). But rather than my searching the entire book could you point out where he suggests scanning grayscale in rgb as Kirk suggested? I'd like to read what he has to say because on his web site he certainly says he usually sees no difference between scanning in grayscale and scanning in rgb. Please understand, I'm not in the least questioning Kirk's method, in fact I plan to try it, nor am I arguing with you. I'm just trying to gather information so that hopefully I can settle on one method and feel comfortable with it.

[Eirik Berger]

I have also purchased this scanner resently, The purpose is to scan my 8x10" BW negatives in a decent way. I have realized that I need to do some researching/testing to get the best results possible.

I tried both Silverfast AI and the Epson software, but lately I have done some testing with Vuescan with great results.
I have experimented with multisampling (from 2-16x, but I guess i will stick with 8x in the future) and with "Long exposure pass".

Krik and Brian, the idea to scan BW negatives as 48-bit RGB and then choose the blue channel sounds very interesting. I agree with Brians concerns regarding file sizes. The files will be too large for my hardware to handle.
An selection in Vuescan is "Make grey from". You can choose between Auto, Red, Green, Blue and Infrared.
Will this do the same as scanning it in RGB and choose the blue channel? I will definetly test this.

[Ed Richards]

> An selection in Vuescan is "Make grey from". You can choose between Auto, Red, Green, Blue and Infrared. Will this do the same as scanning it in RGB and choose the blue channel?

Yes, and it will reduce the file size problem dramatically. Using vuescan to reduce the file size before writing the file also helps. I have found that sampling at 4800 DPI and then having Vuescan reduce the file by 2 (effectively 2400 DPI) is the best bet. This is better than multisampling for me because of the alignment issues - the scanner light heats the negative and causes it to move during long multisampling passes. Since this gives me about 180 meg file from 4x5, you are going to end up with about a 700 meg file. My guess is that 8x10 is not a pure scale up in resolution (movement, film flatness, etc.), so you might try resizing these files to about 500 and see if there is any difference. Keep us posted - I suspect this will work very well, as long as Newton's rings do not become a problem. Handing the 500 meg files is another story, unless you have a very fast computer and lots of ram, or a LOT of patience.:-)

[Eirik Berger]

Movement because of multisampling I havent thought of, but that may be a problem of course. I use Scotch tape to ensure that the negatives lies flat on the glass. This should solve movement problems. I have also flipped the negative with the emulsion side down to reduce newton rings, and flip it back in vuescan (Mirror). I have not seen newton rings ever since with Fuji Acros (with Tmax I still see newtonrings from time to time).

How do you make Vuescan reduce filesize by 2?
Is it "TIFF size reduction" under the "output" tab?

I have learned a lot today, I will keep you posted.

[Kirk Gittings]

Movement during multisampling is an issue with SF and my 4990. SF "Studio" version has an alignment program that works pretty well but some times the movement is too great for the software to solve.

Three points:

Make sure your scanner is on a very firm surface that will absorb the vibrations.

Make sure that an air conditioner is not blowing directly on the scanner.

This sounds weird but it works. Tilt the scanner on its left side and tape the lid down. That takes the weight off the center of a 4x5 and gives some improvement in sharpness.

Do a batch scan from the samenegative and settings. The first one will take the heat and settle down. the second or third will be best. Pick the best one and throw the others away. It works.

[Ed Richards]

> Is it "TIFF size reduction" under the "output" tab?

Yes. Just set it to 2, or even 3, if you want to see the effect of different downsamplings.

> Movement during multisampling is an issue with SF and my 4990.

Silverfast makes multiple scans. The latest Vuescan takes multiple samples with a single pass, so scanner movement is not an issue, at least within reason. I am not sure what taping may do if the negative is trying to buckle from the heat. Are you sure you need to multi-scan at all? Unless your negative is really dense or really thin, it may not matter.

[Ted Harris]

There is a second issue related to "movement during multisampling (or multiscans for that matter) and that is the accuracy of the stepping kotors in the 4990 and the similar machines from Microtek and Canon. At the $500 price point they just do not ahve the engineering for the rigid percision to keep the scanning array 100% perfectly aligned from pass-to-pass. Good enough that it works some of the time or even most of the time, sure since we are talking about very small measurements but all of the time, no. In fact, one of the things you pay a lot more money for in the higher quality scanners is the added precision of the stepping motors achieved through both mechanical and firmware differences with the lower priced units. It's somewhat like we used to say in the television broadcasting industry 35 years ago .... a quad (broadcast video) recorder is $50,000 - to $100,000 worth of recording equipment with an additional $300,000 to $400,000 of error correction mechanics and electronics attached. A lot of that applies to scanners as well.

[Ed Richards]

I agree with Ted. I decided that one pass, one sample gave the least room for error, and that offset the the value of multiple samples. I might make a different call with color where noise is much more of an issue, esp. with chromes because of the high density. B&W negatives are really not much of a challage, even for a consumer scanner.

[Kirk Gittings]

Multi-pass is useful when you are getting allot of noise which it reduces. The Vuescan single pass is superior, but as a scanning program it is really inferior to the controls available from SF AI 6. I have been using both lately and SF wins hands down for overall control. I have also been extensively using a 800 series Imacon which is a good scanner but would greatly be improved by SF software were it available.

[Gregg Cook]

I am a fan of scan RGB and convert to b and w.

If you use channel mixer as a ps adjustment layer you can get the max out of each channel really easily. Just check monochome and use sliders until it looks best.

Simple.

[ajkiwi]

hi all,

just how much diffence is there between drum and a v700 pro wet mount?

[Chris Strobel]

Make sure that an air conditioner is not blowing directly on the scanner.



Sorry to resurrect such an old thread, but I'm curious, why would this make a difference?

[paul stimac]

This goes against what everyone's saying but I usually get better black and white scans with the epson software. ????? Maybe I'm doing something wrong? Silverfast beats epson hands down for color though.

[Zach In Israel]

anyone know how well this one will work with a mac? I've been thinking of getting a 4990.

[ljb0904]

Kirk, I've been using a 4990 with Vuescan for that last few months. I've only been scanning E6. How does Silverfast make scanning better? Does it actually get different raw bits out of the scanner than Vuescan? Is there something about Silverfast that makes it different from scanning with Vuescan and tweaking curves in PS?

Zach, I use a 4990 on a Powerbook G4 and have no problems. I use Vuescan as I've said.

[PMahoney]

Paul-
I've tried a variety of scanning methods with B&W output on my 4990 and consistantly come up with better B&W and Color-to-B&W scans with the Epson software as well. Then again, I don't have the upgraded Silverfast or Vuescan software which might offer a greater degree of control, but comparing the standard Silverfast with the epson software, epson comes out on top in my workflow.
Peter

[Ted Harris]

Peter, when you say standard Silverfast do you mean SE? If that is the case it really isn't the standard software but is a crippled version with very limited functionality.

Kirk may add more but IMO Silverfast makes setting white and black points and making other adjustments easier than the manufacturer's software or VueScan. Secondly it pushes the hardeare capailities of the scanner much farther than the others do.