PDA

View Full Version : Digital camera?


jdadson
03-10-2006, 06:47 PM
[Moderator: If there's a better place for this question, feel free.]

I need a better digital camera. The one I have (a Kodak EasyShare) thinks it's smart, but it's not. If there's something light in the background, it thinks, "Ah ha. Sky! I'll zap it up with some vivid sky blue." I guess it's automatic white balance that's the culprit. Whatever. In order to take pictures of my paintings, I have to experiment around with different backdrops until I can fool it into not adjusting the picture all out of whack. You can actually see the colors changing in the view-finder as you move the camera a little this way or that. Even after all the fiddling, I usually have to adjust things further in PhotoShop.

I won't even go into how treacherous the USB driver for the acursed thing is.

Does anyone here have a cheap digital camera that takes good pictures of paintings?

Is there such a thing as an inexpensive digital camera that just records as best it can what's actually in front of it?

FriendCarol
03-10-2006, 07:37 PM
There are both photography forums (with a thread on digital cameras, what's best of) and a technology for artists forum, but I can't get an answer, either. If you get an answer (and this thread gets moved), please share!

jdadson
03-10-2006, 07:58 PM
I did mangage to get a better picture finally. See "green orange" thread. What I did was to put the painting against a white backdrop. The camera did not make as big a mess of the color. Plus, in PhotoShop, I had a reference to work with. I fiddled until the white backdrop was white, then cropped it out.

Einion
03-10-2006, 08:57 PM
Is there such a thing as an inexpensive digital camera that just records as best it can what's actually in front of it?
No, I don't think so. The best results are unquestionably to be achieved in RAW format but I doubt you want to go to that much trouble or expense (don't know what the cheapest cameras are that have RAW but I don't think any would be classed as cheap). Other that getting consistent lighting sorted out - diffuse light ideally - being able to set a custom white balance is a good thing so I'd certainly recommend looking for that as a feature. My go-to reference in case you don't already know it is Imaging Resource (http://www.imaging-resource.com/).

FWIW the posted photos of your paintings look pretty good to me, although I don't know how much you have to tweak them before posting. I realise that having the originals on hand will show you how far away they actually are which is frustrating but quite a few professional shots of paintings don't do much to capture the look of the actual work!

I think you could cross-post in Photography, there's bound to be some knowledgeable folks there.

BTW, if you want image-editing tips just shout, I've worked as a professional photoretoucher.

Einion

FriendCarol
03-10-2006, 09:40 PM
That's a great site, Einion! I just input my preferences and one suggestion was the Fujifilm FinePix E550. Online (average?) price is about $270, and it supports RAW format. I don't like that it's silver (might want to take macro shots, don't want the camera body reflecting light into the shot :rolleyes: ), but looks quite decent.

I've been relying on reviews & Editor's Picks from PC Magazine (online) to identify potential purchases. This Imaging Resources site looks better informed.

rroberts
03-10-2006, 09:55 PM
RAW format is the best because it includes ALL the digital information. Most of the low price range digitals don't include it -- the default is JPEG, but there is a lot of information lost because of the compression algorithm, even with high-res jpeg.

However, RAW format images are large and you should be sure that the camera includes software for working with RAW -- Photoshop CS2 includes support for RAW image workflow, but of course that's pretty expensive.

Tip : always save the original RAW images. Copy/Save them as TIFF or native Photoshop (PSD) and work from those.

Once you get into working with RAW images, you will discover the wonderful world of color correction. I would suggest you also get a Digital Gray Card. You place the gray card in at least one of the photos you take of a series. Then, in the image manipulation program, you can opt to click on the gray card -- bingo! -- the color correction is almost instantaneous. You can then apply those changes to any or all of the other photos in that range.

The best digital gray card I've ever used is available from http://www.rmimaging.com/products/graycard_index.html

I hope this helps.

jdadson
03-10-2006, 11:24 PM
Apparently ALL inexpensive digital cameras mess with the color on purpose. Or is there an exception?


The E550 has the slight oversaturation that consumers prefer, and that most consumer digicams therefore produce. As usual, strong reds are the most highly saturated, other colors are much more modestly so. It has a bit of a cyan-to-blue hue shift that I've also seen in Canon models, apparently designed to produce more appealing sky colors.

FriendCarol
03-18-2006, 01:31 PM
I think you want one that allows preset formulae for corrections for situations like backlighting or portraits, but also allows the 'manual' setting. That's what I've been checking, anyway -- that all settings have 'manual' overrides.

FriendCarol
03-19-2006, 09:56 PM
Jive, it's not that they mess with the color, they are apparently unable to make it accurate. I can't remember why, but they aren't merely redirecting the light (as with mirrors), they're somehow reproducing it, and they can't do it with 100% accuracy. The imaging-resource site has an indepth review of the camera I've decided I want -- the new Fujifilm 900... Among other evaluations, it shows actual vs. displayed colors. It mentions how the fact that the vectors showing the differences are not only small but tend to run outwards in a circle around white is a good thing. Have a look!

Einion
03-20-2006, 01:50 AM
Yep, both types of common systems - CMOS and CCD - have a problem with capturing colour with fidelity in the first place and there are a number of post-capture processes that make this worse in the finished shot. The final colour can be a sort of 'house style' for each maker, although needless to say it's a lot better with some than others; often it could just come down to finding the style that best suites one's own taste, something like it was with colour print film.

FWIW I have a friend who recently upgraded from one Canon to another and the improvement in the images is quite amazing so even within a single maker's products there can be quite a range. I'd be a lot happier with his camera rather than my own (a Konica Minolta Z3) if I had a choice but his is an SLR so in addition to the much higher initial cost he has to buy lenses as well so it's apples and oranges. And honestly, although I'm very fussy about noise because of a long history using film and my professional background, for what I use the camera for the images are good enough and it's light, comfortable to hold and use (good menu organisation), flexible in use and reasonably fast. And I like how it looks too :)

Einion

bigflea
03-20-2006, 10:11 AM
jdadson,

to your question, I would like to ask one more, which is, is there a way to convert 35mm slide reproductions to a good facsimili jpg. My attempts have been miserable failures in terms of getting a reasonable facsimili of either the original artwork or slide. The 35 mm slide does a very reliable job if I manage the conditions of the shot well.

With digital cameras, on the non technical side of the question, I have had to make the shots outdoors under a white canopy, or diffused bright shade conditions, to get a reasonable facsimili of an art work from the digital cameras I have used. They seem to pic out a color for an area and then convert all the surrounding colors into one mass of saturated color. By working with the manual adjustments and the directions for the controls/conditions, I can improve the results somewhat.
Ken

tk04
03-20-2006, 10:29 AM
is there a way to convert 35mm slide reproductions to a good facsimili jpg.
Ken

I have a HP scanner that is sold with an extra input-device, in addition to the scanner itself. You put the negative into this "input-device" and scans it. You get a jfp-file from it, the same way as with other scans. I have used it on slides and the quality is amazing.

bigflea
03-20-2006, 07:55 PM
Karin, when I have seen slides and scans of slides it seemed each scan lost in quality overall? Any comment about that from your experience?
ken

tk04
03-21-2006, 06:12 AM
You have to scan the slide with a high resolution since it's to be enlarged. Professionals might have an optimal formula for this. I have just used a resolution that has appeared OK. The "sharpness" is no problem.

The main problem would be the color-balance. Also because if the original slide, or negative, is slightly off in some way or another, that "offness" might become higher when you scan. But it's possible to correct these things in Photoshop. You can ask someone to do these corrections professionaly if you are to use it for public purposes.

If you in the end will have a result that you find acceptable - I don't know. My scans appear to be quite close to the original negative or slide, but as a reproduction of original artwork even a small "offness" is a problem of course. And then if the slide is "off" from the beginning, you have to be very good with Photoshop and photography to be able to correct it.

The alternative to scanning will be a new photo of the original.

Einion
03-22-2006, 12:00 AM
I would like to ask one more, which is, is there a way to convert 35mm slide reproductions to a good facsimili jpg. My attempts have been miserable failures in terms of getting a reasonable facsimili of either the original artwork or slide.
I presume you're using a scanner with a built-in lit top, what we call a trannie hood?

Karin, when I have seen slides and scans of slides it seemed each scan lost in quality overall? Any comment about that from your experience?
Well since I've done more of this than anyone else here maybe you should ask me :lol:

All scans lose information, the amount is based on the quality of the scanner (and to some extent the software used). A photography rule of thumb I was taught a long time ago is that if you imagine a negative has tonal information of 0-100% a straight print of it will have about a tenth of that; a good scan of a transparency will yield much much more than this. To get high-quality results you used to be talking a $$$$ pricetag for the scanner, this could well have changed with the drop in CCD price and I have seen sub-$500 scanners with trannie hoods. Professional-level drum scanners are just ridiculous prices as you'd expect but the scan quality (especially since the transparencies should be mounted wet) is amazing and it can really be worth it if you need a large image size - like magazine size at 300ppi from a 35mm transparency.

If you want maximum quality when you scan it's best to use the scanner's limit of resolution or at least much higher than you need at the end of the day, it gives you the most information to play with, but this is only practical if time is not a constraint. So you'd:
scan;
save a copy to work on;
adjust/retouch;
reduce;
sharpen if necessary;
save as JPEG.

Einion

FriendCarol
03-22-2006, 12:21 AM
I don't know if you have the equivalent over there, Einion, but in the States there are franchises of (formerly so-called) print shops (Kinko's is one). That might be a better way to go if it's a matter of a few slides.

Ken, unless you intend to continue making slides with your camera, it might be cheaper to have the professionals transfer specific slides, into either a CD-ROM or separate jpg files. Otoh, if you want to paint from your slides, I heartily recommend a 'rear-screen projection' setup with slide projector (the latter often available cheap from Ebay). You get a nice, bright, big reference, even in daylight. If you're interested, I made a rear-projection screen recently that works really well, very cheaply and easily.

Einion, as soon as I buy my camera (about a month, I'm guessing), I'm going to have lots of questions about using software to 'fix' color. To start with, what does one use as a reference? :confused: The actual scene won't be there, and my memory for color isn't that good! Or would I set up a routine to fix color based on photographing paintings, then use the same values?

tk04
03-22-2006, 04:37 AM
If you want maximum quality when you scan it's best to use the scanner's limit of resolution or at least much higher than you need at the end of the day, it gives you the most information to play with, but this is only practical if time is not a constraint.

The optimal resolution depends on the size of final jpg-picture. It's not the case that more is necessarity better. But you will need a high resolution, yes. That's because the jpg-copy will be larger than the negative. Experts in the field would know the optimal resolution based on what resolution the final jpg-picture is to have and the size the jfp-picture is be presented in. I have never bothered to calculate this since I'm not responsible for a photo-lab and therefore in no need to develop quidelines for employees to follow then they rutinely scan pictures.

tk04
03-22-2006, 04:51 AM
I don't know if you have the equivalent over there, Einion, but in the States there are franchises of (formerly so-called) print shops (Kinko's is one). That might be a better way to go if it's a matter of a few slides.



Photolabs enlarge pictures all the time by digital means. It might be possible to get a jpf-copy of a slide from one of these labs. Problem is that they do this fairly "mechanically". Most of them produce thousands of pictures. Of course he can try that and see what he gets, but he might lose some money on it if the result isn't good enough. Therefore, he might need to contact a photographer or a printshop, it he wants it done professionally, to have it done as a "special commision". Gieclee-producers might also be able to help out. But I find it quite likely that he would know this himself.

What Ken talks about here are photos of his own work. Artists have traditionally used slides of artwork to apply for competions, exhibitions and so on. Ken probably wants to convert a slide to a jpg-file to use on a website, or for some other purpose involving computers. If he has sold his work it might not be possible for practical reasons to take a new photo of the work. Or he might have had his slides made by a professional photographer. Then he want so avoid taking new pictures because it's costly to have artwork photographed professionally. Or something like that. Problem here is to have a copy of the slide that is good enough to represent his work.

So no - it's somewhat unlikely that a projector will solve this problem. But of course, I might be wrong about my assumptions here.

tk04
03-22-2006, 05:32 AM
As an addition to my post nr. 14. A too high resolution gives as much distortions to the original as a too low resolution, even if it's different types of distortions.

If I was to have someone do this professionally, I would have taken care to specify my needs so that the scan is done for a specific purpose, for instance scan it to be used on the web. And also discuss how it will work out if you want to present it on the web in different sizes - like a small picture and a an enlargement. In the last case, you might have to make two scans, if you won't compromise on quality. You can of course then use the same file for other purposes too, like taking a print of it, but then used for other purposes the copy is likely to have a lower quality, it means with more distortions, than the web-version.

Or if you do this yourself - The important thing is to remember that the web show pictures in a standard resolution and you have to judge the quality of the scan based on how it shows on the screen.

I appriciate you posts here so I hope some of this might be helpful.

bigflea
03-22-2006, 09:47 AM
Thanks everyone, for all the feedback and information. I am learning something I think.

Usually I am able to make good facsimili 35mm slides, and that is how I document work. I take the shots and have, in the past, sent roles out to a professional processor, who has recently gone out of business. I have not found a processor for slides that is reliable in terms of producing a facsimili of works. But that is a different problem.

Based on the information Einion and Karin have presented I see it means I will have to learn more about scanners to make some headway in solving the problem. I had been taking slide film and print film to a processor for all purposes, including making jpgs. from slides and photos with no successful results. Also a friend's attempts to convert good slides to jpgs.( on their scanner) was less than hoped for.

It is also helpful knowing more about what to look for in a digital camera, ie. what format it is capable of. Knowing nothing or next to nothing about the technology makes it difficult to ask the right questions.

Anymore information that can be posted would be welcome. Thanks.
Ken
ps FriendCarol, I don't paint from slides, but had not thought to look on ebay for a good projector, which I do need. Can the set up you are referring to be used simply for a slide show ?

FriendCarol
03-22-2006, 01:57 PM
Hi, Ken -- sure! Basically the slide projector is in front of me, in my case about 8 feet away (the distance determines size, of course), and shines onto a screen positioned in front of me, made simply of mylar frosted on both sides. The main difference is just the position of the projector: Normally it's behind the audience, and projecting onto the front of a screen which (for some reason) has to be viewed in an otherwise dark space; the audience is between screen & projector, or in front of the projector (shining towards their backs).

Btw, I bought the mylar at an architectural print shop (normally produces copies of blueprints); roughly a 3' length, probably about 30" wide, was about $7.50. I made a frame of 1/2" PVC piping to support the mylar 'screen.' I taped the mylar to one pipe, which can be easily removed from the frame; therefore the screen can be rolled up. Pipe and fittings are available in Home Depot, and cost less than the mylar. :)

The only problem I could anticipate is that you'll almost certainly need a remote to manage your presentation, since you won't be sitting next to the projector. ;)

Karin, the shops I'm talking about aren't photolabs -- yes, those do very routine processing for the most part, and wouldn't be very helpful. This other type of shop (Kinko's being a prime example) started out as print shops, then graduated to large specialized workspaces with very high-end equipment plus (usually) some workstations avaiable to users. So if you need some very fancy output or file, you go to one of these places and rent a PC networked to a piece of equipment (scanner, printer, whatever) that you would never be able to afford to have in your own office. These workstations are rented by the hour.

Or, you take your problem to one of their specialists at the counter, who uses a workstation to take care of the job. They will even provide training or assistance at a workstation you rent (for an additional cost), if you like. It's not at all like a photolab, though the ones I've been in do also have equipment to handle slides and negatives.

Btw, it won't hurt to have too much information (too high a resolution scanning) unless time (or memory) is a constraint. One can always discard extra information; PhotoShop for one provides for many algorithms among which to choose to reduce bits to a smaller number of bits. But one cannot add information... except by extrapolation/interpolation which is not always helpful. Unfortunately, however, each step up in resolution squares (at least) the time required for processing (just because from, for example, 300 to 600dpi is obviously applied not just to a linear measure, but to an area). When I try to scan w/c paintings even just at 300dpi, it can take 45 minutes for a small scan (less than 8"x11") to reach the manipulating software!

Having a lot more memory would help. But the first thing I do, always, when it reaches PhotoShop (even my 12-year-old limited edition version) is to change the image size, setting resolution back down to 96dpi. The resulting image is far superior to one I might have scanned at 100dpi and then reduced to 96dpi, without even specifying which reducing algorithm it should use. The results are still very poor, but I'm using a cheap scanner (printer) I bought on Ebay 3-4 years ago for under $100. :D

tk04
03-22-2006, 04:42 PM
Btw, it won't hurt to have too much information (too high a resolution scanning) unless time (or memory) is a constraint.


Yes, in principle. And never mind.

Since negatives are small they are not that slow to scan even with resolutions over 1000.

FriendCarol
03-23-2006, 10:11 AM
Anyone not interested in technical issues re scanning should skip this. :)

The disagreement seemed to turn on whether too much information could be a bad thing. That is a compression question, afaic. Well, it's a sampling question, really; data compression is a lower order question. Of course, Jpeg is itself lossy compression, but that's irrelevant, since resizing the image can be non-lossy and we're really talking about resizing.

In time series analysis, the natural cycle or periodicity of a process (assuming there is a cyclical process, and its period is unknown) can mean that the specific quantity of (sufficiently large) data is not very relevant, because when it is sampled matters. I.e., worst case: you just happen to keep sampling the data at the peak of its values each time (or at any part of it's cycle -- the sampling happens to be in tune with the cycle). Even if the sampling is not completely synchronized, unless the data is large (covers many cycles), it can be very misleading, regardless of how many values you've gathered.

I was just wondering yesterday if there could be an analogous situation for sampling of color on a 2-d surface. One major problem I often have, apart from the poor quality of my scanner, is that watercolor paper is not a 2D surface. It's possible I get very poor results in part because the sampling interval does not match the paper's irregularity, but of course that's not a 2D problem. (More data, in this case, tends to give me a better result. The deepest depressions are less represented, perhaps, and the top surface better represented?) Anyway, I haven't thought of any situation regarding sampling of a flat surface where fewer data could be better, yet.

bigflea
03-23-2006, 10:52 AM
So the quality of the scan, or its effectiveness as a facsimili, depends on the resolution that is selected, which is determined by the intended purpose or use, which includes the size of the image that will be shown? Larger jpgs. may need a different scan than smaller jpgs., to attain a reasonable color balance, ie. a facsimili of the original artwork or slide of artwork. (Color balance has seemed, to me, to be the problem in jpgs. from scans of slides.)
Please correct me if I am mis-paraphrasing what has been stated re. scanning slides and the relationship between resolution and size needs.

The term "trannie hood" is commonly understood, or is there another term that refers to scanners with a built in light box? I wonder if I use the term if it will mean the same thing here as everywhere else it has meaning?

The visualization of the scanning problem re. watercolor paper and the natural cycle is interesting. Are you saying that the scan misses some of the information because it only takes in info at the high points of the paper?
Ken

FriendCarol
03-23-2006, 03:01 PM
So the quality of the scan, or its effectiveness as a facsimili, depends on the resolution that is selectedYes...which is determined by the intended purpose or use, which includes the size of the image that will be shown?Not really.
Think of it this way: A newspaper photograph is composed of many tiny dots, which are not so tiny that you cannot see them. If the dots were smaller, and also closer together, that would be higher resolution. Higher resolution would be better visually, yes? Would look smoother, etc. But other factors -- primarily cost, one assumes -- determine the 'best' resolution for this purpose (newspapers).

The size of the photo in this case does not change anyone's mind about the resolution -- the purpose is minimally acceptable daily print coverage of news. Size has nothing to do with it in this case.

Now, assume instead you are an artist with a huge painting to submit to a juried show via e-mail. Assume they only permit you to submit a 1mg file. Now we are dealing with 2 sizes: Size of the painting, and size of the file. Three factors relate these 2 sizes, but one is something we haven't discussed. The two we know a bit about are resolution and compression.

The third is bit depth. Let's sideline this, since it's generally fixed: Basically, your PC monitor can show a certain number of colors based on how many bits (binary digits) are allowed to be combined to code the color. In the very old days (early 70s :D ), an addressable dot on your screen (a pixel) could only be black, or white. One bit was enough to describe that, although 8 were available (or 7... another complication we'll ignore).

Some older monitors, however, are limited to 256 colors. But most PCs today can show millions of colors. That is, either 16, 24 or 32 bits are devoted to specifying the color of EACH PIXEL.

Related but not at all relevant: Monitors are larger these days in terms of the number of pixels that can be addressed. The 'largest' image in pixels (not inches) that can fit on your screen might be 600x800, or 1024x768, etc. Whether you have a 17" monitor or 21" monitor is a different kind of size -- the number of pixels for a given resolution is the same.

Iow there is no relationship between pixel (one dot on the screen, which can be 'addressed' or take on one color) and size, generally. The screen size and bit depth help set that relationship for a particular monitor.

Those are many different kinds of 'size' that we might be speaking of! What the artist wants to do is prepare the best scan of a particular sized piece of art (dimensions of the surface), where the result of the scan fits in a particular sized file. The size of the file is measured in the number of bits needed to convey the information.

To give you an idea of what a raw file size means: Pretend that the file is made up of each bit in the first row, followed by each bit in the second row, etc., down to each bit of the last row. In the front of the file, a couple characters (made up of 8 bits, or 16 bits, or 32 bits, whatever!) specify how many rows and columns follow, so whatever reads the file knows where to cut the first row and start the second.

If you understand this (above) about file size, we can go on to look at what compression and lossy compression can do to the file size. :)Larger jpgs. may need a different scan than smaller jpgs., to attain a reasonable color balance, ie. a facsimili of the original artwork or slide of artwork. (Color balance has seemed, to me, to be the problem in jpgs. from scans of slides.)Second point first: I'm not sure whether you mean 'color balance' or just 'color!' If a whole scan looks somewhat yellow, that's actually much easier to correct (in PhotoShop, et al) than if only green is too yellow... if you can imagine that.

In the first case, it's as if there's a yellow filter over the whole thing. All you have to do is compensate for that filter on the whole file.

But when only bits and pieces of the whole thing are off... Einion can address this much better than I can, actually... The only thing I know how to do is select the bad areas and try to fix them. I think someone who knows what they're doing (someone like Einion, or perhaps WFMartin, though he's a print specialist) can alter a profile somewhere to selectively alter colors.

Going back to what you said earlier, no, a larger work doesn't require a different scan than a smaller work. It's hard to state precisely the best way to grasp this, but I'll try to give you an overview: If someone wanted to scan some 'art' that was entirely one color (one hue, value, and chroma), no matter how big that art is, the (compressed) file will be tiny. That's because all the file has to say is what color it is, and where the color is (in this case, how many rows and columns of pixels to make that same color).

Even if the artist used only 2 colors, it's not a big file. We can even use compression that loses no information, and it still won't be very big.

If a piece of art is stripes, again, we can make a file that describes that work fairly small. Think of it like this: If I wanted you to prepare a copy of my art which is stripes, I only have to tell you how big (relatively) the stripes are, and what color they are.

Notice I said 'relatively' because the electronic file doesn't know how big the original is. It never does. ;)

The point is, as the art becomes more random or chaotic, even if only a few colors are involved, the file needed to describe it becomes bigger. Does that make sense to you?Please correct me if I am mis-paraphrasing what has been stated re. scanning slides and the relationship between resolution and size needs.We're getting there, slowly. I'll wait for your response before I try to explain further.
The visualization of the scanning problem re. watercolor paper and the natural cycle is interesting. Are you saying that the scan misses some of the information because it only takes in info at the high points of the paper?I wish! :D No, it is including wrong information, because it is measuring the color on the recessed paper. That either looks darker or black, or is in some other way not measurable, depending on the color. So if I painted a nice bright blue, the scan apparently sees it as speckled light and dark blue, at best. It's worse than that IRL, because often my scanner doesn't know if a painted area is light blue or light green or aqua or purple. :rolleyes:

I didn't address some points here (about scanners). Someone will, I'm sure. ;)

Einion
03-23-2006, 04:04 PM
Einion, as soon as I buy my camera (about a month, I'm guessing), I'm going to have lots of questions about using software to 'fix' color.
Sure, ask away, although this might not be the best forum for it so maybe via PM.

To start with, what does one use as a reference? :confused: The actual scene won't be there, and my memory for color isn't that good! Or would I set up a routine to fix color based on photographing paintings, then use the same values?
You can use values (and sometimes colour) within an image to adjust it but it's better to have a control strip within the photo which will give you fixed areas to correct toward. You might have seen these in some photos of pictures - a strip with a value scale, which would include your black and white points, a perfect middle grey and CMY swatches + their mixtures.

This is the ideal, you don't have to go this far if you don't have any expectations of getting perfect reproductions - which is an illusion anyway as far as on-screen representations go - for reasons I'm sure you'll already appreciate.

And as I joked about above even professional photographs of painted works sometimes don't capture them with much fidelity (this even includes images for national galleries, although the standard for these is much more consistent than it once was).


If you want maximum quality when you scan it's best to use the scanner's limit of resolution or at least much higher than you need at the end of the day, it gives you the most information to play with, but this is only practical if time is not a constraint.The optimal resolution depends on the size of final jpg-picture.
That's the optimal resolution for the size of the final image, this does not necessarily bear on the original scan - when scanning transparencies too much resolution is always better than too little if one has the time.

Experts in the field...
Since you wouldn't know that would be me ;)

As an addition to my post nr. 14. A too high resolution gives as much distortions to the original as a too low resolution, even if it's different types of distortions.
Would you like to explain how exactly? Since this bears on what I said above please restrict yourself to the two examples I gave - maximum scanner resolution or much higher than the required size.

If I was to have someone do this professionally, I would have taken care to specify my needs so that the scan is done for a specific purpose, for instance scan it to be used on the web.
Since you don't know this said professional would do stuff in the background that you'll be entirely unaware of ;)

And also discuss how it will work out if you want to present it on the web in different sizes - like a small picture and a an enlargement. In the last case, you might have to make two scans, if you won't compromise on quality.
No. You scan large and reduce as necessary. Scanning to a specific finished resolution - except when using hardware of top quality and when the trannie is wet mounted as I mentioned above - is not the best way as apart from anything you have to pay to have it done each time.

And if you did ask for two scans I can tell you that plenty of scanning houses would do exactly as I've suggested above - scan at a high resolution and then make two copies at the requested sizes :D

Einion

bigflea
03-23-2006, 08:54 PM
One clarification,
by "larger" I was meaning a "big picture" on a website, as opposed to a thumbnail. This in regard to selection of a higher resolution, or a lower resolution, in a scan, for either a " big picture" or a thumbnail. IOW, I understand, the original image/artwork size is not pertinent to the resolution needed. So, if we wanted to post a larger jpg or other digital image, we may require a higher resolution, then required by a smaller digital image, in the original scan of a slide?

Yes I understand that as the complexity of color variation increases, in the original work to be scanned, the higher the resolution needed.

In general, higher resolution is selected, depending on purpose, cost, and time required/available, then the image is reduced or manipulated for multiple uses?

"Color balance"- I was thinking of scans that did not resemble the original works, and appeared to be scewed toward particular color limitations and saturations, and also some that appeared to be monotoned in a way the original works were not. I used it as a general symptomatic description, but is probably not.
Ken

Einion
03-23-2006, 11:17 PM
So, if we wanted to post a larger jpg or other digital image, we may require a higher resolution, then required by a smaller digital image, in the original scan of a slide?
Resolution is an area that often gives people trouble partly I think because it can be seen from two different perspectives, which aren't actually incompatible but can seem that way.

What matters is the number of pixels involved, not the actual image resolution per se - a 500 x 600 image at 72ppi is the 'same size' as an image 500 x 600 at 300ppi, it's just dimensionally larger (6.9" x 8.3" compared with 1.7" x 2").

Yes I understand that as the complexity of color variation increases, in the original work to be scanned, the higher the resolution needed.
It's more about capturing detail than colour itself, although of course the detail is in colour the scanner's ability to reproduce colours accurately isn't really tied to its size within the image (with some exceptions).

In general, higher resolution is selected, depending on purpose, cost, and time required/available, then the image is reduced or manipulated for multiple uses?
Yep, that would be normal practice. For example you might scan a trannie at 3000ppi, which would give you an image that is the dimensions of the 35mm transparency at 3000 pixels per inch.

Say you want two versions of it, 1, to make a print of the image at 10" x 8" and 2, a picture for your website that's 500 pixels wide.

For 1, you'd need a 200-250ppi version at those dimensions, for 2 you'd just save it at 500 pixels wide (ideally at 72ppi but that's not necessary, only the number of pixels is).

"Color balance"- I was thinking of scans that did not resemble the original works, and appeared to be scewed toward particular color limitations and saturations...
This is quite common regrettably. Many cameras have problems with the blue/violet/red-violet area for example and the same is true of cheaper scanners.

...and also some that appeared to be monotoned in a way the original works were not.
This can also occur, although if it's fairly pronounced it's a very bad image.

You can still do more than you might expect though; compare the photos in posts #1 and #10 in this thread for example:
http://www.wetcanvas.com/forums/showthread.php?p=4426908
Bear in mind this was a small JPEG to begin with and the correction took about five seconds to do - it would be hard to guess that there was still enough colour variation within the image to get that much out of it.

Einion

bigflea
03-24-2006, 09:04 PM
By "capturing detail" could you mean, any variations that occur in the 2 dimensional surface (as opposed to something that is a peak or valley in the surface) ?

More pixels per inch increase the scanned information and that is what you are referring to as a higher resolution scan? By image resolution you mean something different than how many pixels per inch?

Ain't this fun?
Ken

Einion
03-25-2006, 03:49 AM
By "capturing detail" could you mean, any variations that occur in the 2 dimensional surface (as opposed to something that is a peak or valley in the surface) ?
All kinds of detail you might want to capture from brushmark texture to canvas weave to fine painted elements. Assuming they're visible in the original transparency of course the higher the resolution the better they will be reproduced in the scan.

More pixels per inch increase the scanned information and that is what you are referring to as a higher resolution scan?
For fixed image dimensions yes. Remember you can have two images at 72ppi and 300ppi that contain precisely the same amount of information.

By image resolution you mean something different than how many pixels per inch?
If it helps maybe try to think in terms of pixel dimensions; take two images of the same painting, one considered small and the other large (low resolution and high resolution). The low-res one might be 300 pixels wide while the high-res one is 1250 pixels wide.

This can be considered irrespective of the ppi - it's the amount of pixels that go to make up the image information that is really important - although you do have to keep the formal resolution in mind for some applications.

Ain't this fun?
:D

Einion

Richard Saylor
03-25-2006, 04:13 AM
For Ken:

An image stored in a computer doesn't have linear dimensions like inches or millimeters; it has pixel dimensions. If a 24x36mm trannie is scanned at 3000ppi, then simple arithmetic shows that the scanned image is approx. 2820x4200 pixels. When I save this image, I may specify ppi (linear pixel density) if I wish, but this by itself doesn't change the amount of information in the image or the size of the image file (number of bytes).

If I use all the information in my 2820x4200 pixel image, I can make a 14x21in print at 200ppi, a 9.4x14in print at 300ppi, a 2.8x4.2in print at 1000ppi, etc. (Most printers won't actually crowd 1000 pixels into each inch, but don't worry about that.) To compute the size of the printed image, just divide the pixel dimensions of the image file by the number of pixels per inch.

By the way, the image won't even fit on an average computer monitor without being resized. Just for fun, the "resolution" of my monitor is 1024x768pixels, which works out to about 72ppi since its width is about 14.25in: 1024pixels/14.25in = 71.9ppi. At 72ppi, the size of my scanned image would be 39x58in.

bigflea
03-25-2006, 09:20 AM
Thanks Richard, that helps alot.
So the amount of ppi does not effect the amount of information a scan duplicates, but it does effect the quality?
Ken

FriendCarol
03-25-2006, 01:26 PM
Think of it this way: Here's your picture. Should you show it to us by showing us 5 dots of it in every inch, or 300 dots of it in each inch? Usually (normal art, not stripes or solids!), yes, ppi (or dpi in the States) affects the quality. It's just sampling more of the image.

Richard Saylor
03-25-2006, 05:21 PM
So the amount of ppi does not effect the amount of information a scan duplicates, but it does effect the quality?The ppi of a scan actually does affect the amount of information, but beyond the scan, the ppi alone does not affect the amount of information. Let's say I have a 1"x1" picture which I want to scan.

If I scan it at 5 pixels per inch, I will get 5x5 pixel scan. I.e., the original picture will be represented by only 25 pixels (5x5=25). That's not very much information. If I try to make, say, a 5"x5" print with this information, the print will look terrible, as there will be only one pixel in each inch of the print. (There are only 5 pixels of information for the 5" width (or height) of the print, and 5pixels/5" = 1ppi.)

If, on the other hand, I scan it at 1200 pixels per inch, I will get a 1200x1200 pixel scan. That's 1440000 pixels, which is a lot of information compared to the previous 25 pixel scan. In fact, if I make a 5"x5" print using this scan, there will be 240 pixels in each inch of the print. (There are 1200 pixels for the 5" width or height of the print, and 1200pixels/5"=240ppi.)

Briefly, the ppi of a scan does determine the amount of information in the scan, but thereafter changing the ppi alone does not affect the amount of information in the image file.

bigflea
03-25-2006, 09:33 PM
Hence the recommended higher resolution for the original scanning, in order to avoid having too little information for a facsimili?
Ken

Richard Saylor
03-25-2006, 09:54 PM
Hence the recommended higher resolution for the original scanning, in order to avoid having too little information for a facsimili?
KenYes.

TessMurdock
03-28-2006, 02:09 PM
I used to be a product manager for a Hitachi Digital MPEG/JPEG camera so I went to a lot of trade shows for digital photography products. I spent a lot of time looking for a camera that married the best of analog photography while giving the convenience of digital at a good price. I looked at cameras raniging in price from $150 to 10,500.

The one I purchased is a SONY Cybershot DSC H1. It is about $450. An important feature is that it has an extremely good optical lens for a digital with a 12X optical zoom. It also has a 2.5" LCD monitor, and super steady shot. It is packed with features for such an inexpensive camera. I like that I can store thousands of photos on the 1 GB memory stick, but I can also store photos on the camera. It has an ok flash, but most of these features have manual overrides. It has a great digital capacity with photos as large as 5 megapixels, so you can edit and print very large photos.

If you would like to learn more, I suggest you visit sony.com for a list of features and accessories. Consumer Reports also published a favorable review about it recently. I also puchased a close-up and zoom lens. These after market products are available only from SONY, but I thought it was a good investment. It has extended the capabilities greatly. I take lots of photos of landscapes I usually start on site and finish in my studio from 8 X 10s that I print from the camera. It is not a professional quality camera, but just short of it, in my opinion. I think it is the best camera for the price available. When I put together my portfolio I will definitely be using this camera. Don't but your memory stick from SONY, too expensive.

Good luck to you.

Tess :wave:

TessMurdock
03-28-2006, 02:28 PM
When discussing dpi remember that most computer monitors are capable of displaying photos at 72 dpi, hence many of your photos are blown up greatly. A minimum dpi for printing purposes is about 300 dpi. Also, when printing a photo, think about the dpi capacity of the output device (printer) you are using. It is hardly worth the trouble to edit a 10 megapixel photo that is going to be printed on a 1024 dpi printer at 8X10". Most computers can't handle a 10 megapixel photo anyway.

Best wishes,
Tess

Richard Saylor
03-28-2006, 04:22 PM
.....It is hardly worth the trouble to edit a 10 megapixel photo that is going to be printed on a 1024 dpi printer at 8X10". Most computers can't handle a 10 megapixel photo anyway.However, if the optimum input for the printer is 300 dpi, then for an 8x10" print your image should be at least 2400x3000 pixels, which is 7.2 mp. The input dpi (computer image resolution) and output dpi (linear ink dot density) of a printer are not directly related. For the highest quality print, one must provide the optimum recommended input resolution.

Most modern computers should have no trouble handling a 10 mp image.

FriendCarol
03-28-2006, 04:28 PM
Hmmm. And there I had just decided, finally, after months of hemming and hawing, on FujiFilm 900 (at just over $300 online). *sigh* Thanks, Tess. Now I have to start my evaluation over again. Macro is a must for me, so I suppose first I have to figure out how much for the closeup add-on. :D

I wish they rented digicams the way they rent cars!

TessMurdock
03-28-2006, 08:59 PM
Hi Richard,

Maybe one might be able to open and print it, and if you have a 1024 MB of RAM, but editing it in Photoshop CS, as I do, my 64-bit processor becomes as slow as a sick dog.

Tess

Richard Saylor
03-28-2006, 10:56 PM
Hi Richard,

Maybe one might be able to open and print it, and if you have a 1024 MB of RAM, but editing it in Photoshop CS, as I do, my 64-bit processor becomes as slow as a sick dog.

TessYes, same here. I'm using a dinosaur 32-bit processor with a mere 256 mb RAM as well as an extinct film scanner and digicam. For those reasons, I usually work with 4mp files and print 5x7". If I want to do an 8x10", I use a sophisticated interpolation algorithm (in Qimage Pro) to boost the dpi to 300 (which takes forever), resulting in a 7.2 mp image. Surprisingly, my 8x10" black & white prints on an HP Photosmart 7960 are quite good, and my color prints are good enough for friends and family. ;)

tk04
03-29-2006, 07:41 AM
Would you like to explain how exactly? Since this bears on what I said above please restrict yourself to the two examples I gave - maximum scanner resolution or much higher than the required size.

Einion

An image on web has a given resolution. It's quite low, 72. You can't change that. This is different from a printer. My home-printer don't mind resolutions at 1000.

If you scan an image for publication on the web, with a resolution much higher than the one needed for a certain size, you will have to reduce the resolution to use it on a web-site. The computer will then add together pixels, it means the color-information, after certain mathematical formulas - averages in general - and you will lose information during this process. How much you lose will depend on the image, of course. I find that one loses less by "scanning to target", not with a resolution of 72, but the resolution needed for a certain size.

tk04
03-29-2006, 07:47 AM
And if you did ask for two scans I can tell you that plenty of scanning houses would do exactly as I've suggested above - scan at a high resolution and then make two copies at the requested sizes :D
Einion

Well, for the sake of the "plenty" part of the above mentioned houses, let's hope I'm not one of their financial important customers, or a customer that otherwise brings prestige, since it undoubly would be the last the "plenty" part saw of me.

TessMurdock
03-29-2006, 09:14 AM
I used to be of the "Scan for Higher Resolution the Better" school. Putting out a quarterly publication sobered me up real quickly. Working with print bureaus taught me this rule of thumb: do your scans at twice the dpi of the final printing. In my particular case, I produced a four color cover that was printed at 300 dpi. I delivered photos and other images at 600 dpi on CD. 1200 dpi would be overkill, a waste of space on my computer (unless I was going to blow up or use the images for some other purpose that required 1200 dpi imagery).

Another note, when "grabbing" an image off of your computer (screen resolution -- 72 dpi) you can improve the appearance of the image in Photoshop by first increasing the screen resolution in a Dialog Box in the program then shrinking the image by at least half.

Have a wonderful and colorful day.

Tess

FriendCarol
03-29-2006, 12:55 PM
The computer will then add together pixels, it means the color-information, after certain mathematical formulas - averages in general - and you will lose information during this process.While this is true, it is also somewhat misleading: The part that's missing is that the information 'lost' would have been lost anyway (and possibly more) if the scan had taken the information at a lower sampling rate originally.

The distinction is between an analogue image (in the world -- a painting, perhaps), and the digital reproduction. If you never sample the full information in the analogue image, it can't appear -- even as partial input to an averaged value in a reduced size -- in the digital image.

So the question become a practical one: either of time and expense (which I assume is where Tess is coming from), or a balancing of the precision of full information against precision of perception. When we try to do the latter (the balancing act), if we know the final size and the final resolution we want, as well as essentially some measure of the smallest meaningful unit (i.e., how big/small could a paint speck be? or perhaps how many molecules are required to display 'color?'), we can make good decisions. But if we might want the image (in future) for multiple purposes, it can only be good to sample it (and save it) as fully as current technology allows.

Iow: the information you don't sample with the lower resolution scan is also 'lost.' If you ever needed to blow up the image to see a tiny part of it, that would not be possible, because the original information (in the original image) is no longer there. These are practical considerations these days, since we know that satellites in space, hundreds of miles above the earth, are fully capable of sending sufficient digital information to allow voyeurs to read the license numbers on cars! Do artists need this much information? Probably not. But technically, if we're talking about loss of information, a lower resolution scan has already lost it.

Btw, there are algorithms of compression that are 'lossy,' and others that allow us to fully recover all the original information. Jpg (or jpeg) is lossy, but there are plenty of compression formats that aren't.

Richard Saylor
03-29-2006, 04:23 PM
An image on web has a given resolution. It's quite low, 72. You can't change that.Actually the resolution of an image on the web depends on the resolution of the monitor. If I set my monitor at 640x480p, then the resolution of any image viewed on it is a mere 45ppi. It I set it at 1024x768, then it is up to a respectable 72ppi.
This is different from a printer. My home-printer don't mind resolutions at 1000.Most printers are equipped with resampling software, which will change the resolution of any input to whatever is appropriate for the printer.
If you scan an image for publication on the web, with a resolution much higher than the one needed for a certain size, you will have to reduce the resolution to use it on a web-site. The computer will then add together pixels, it means the color-information, after certain mathematical formulas - averages in general - and you will lose information during this process. How much you lose will depend on the image, of course. I find that one loses less by "scanning to target", not with a resolution of 72, but the resolution needed for a certain size.This is true. It is therefore best to save the original high resolution scan and work with a copy for web use. Downsizing an image can seriously degrade the quality. The trick is to do it in such a way that it doesn't appear degraded. I've found that sometimes a very simple resampling algorithm (such as bilinear) works better than more sophisticated methods, and with a touch of the unsharp masque, one can produce a really nice looking image.

Richard

Richard Saylor
03-29-2006, 04:54 PM
Karin and everybody-

I agree that it may be a waste of resources to scan at an extremely high resolution if all one needs is a web image, however.....

If I want a 5x7" image on a 72ppi monitor, I can obtain directly it by scanning at a resolution which will give me a 360x500p image. However, I've found that better results are obtained if I scan at 2 or 3 times greater resolution than I need, process the image to my satisfaction, then downsize to 360x500p. If done carefully, this will produce a better image than a lower resolution scan. Of course, an absurdly high resolution scan is ridiculous for such an application.

Richard

Einion
03-30-2006, 02:47 AM
When discussing dpi...
Hi Tess, welcome. Image resolution is expressed as ppi. Printer resolution is given in dpi. It can be important to differentiate between the two.

most computer monitors are capable of displaying photos at 72 dpi, hence many of your photos are blown up greatly.
Not really. 'Blown up' would imply that the image is actually enlarged in pixel terms, when in fact you're changing the resolution and the actual dimensions.

A minimum dpi for printing purposes is about 300 dpi.
Printing how? If one is printing an image digitally the ideal resolution varies from output device to output device (200-250ppi isn't uncommon).

If one is making separations where the image becomes screened (lpi - lines per inch) the rule of thumb is 1.5-2 times the output line screen; although you can get away with less than 1.5x for some images. If you include stochastic screening you have to do things a little differently again.


An image on web has a given resolution. It's quite low, 72. You can't change that. If you scan an image for publication on the web, with a resolution much higher than the one needed for a certain size, you will have to reduce the resolution to use it on a web-site.
Yes, this is covered above.

The computer will then add together pixels...
Essentially yes, it's called interpolation.

it means the color-information, after certain mathematical formulas - averages in general - and you will lose information during this process.
Yes. But to echo what Carol has said above you also lose information in the scanning process itself and if you scan for an image that's 300x500 pixels for example that's all you have.

I find that one loses less by "scanning to target", not with a resolution of 72, but the resolution needed for a certain size.
As the saying goes 'individual mileage may vary' ;) One needs to try direct comparisons to get an idea of what works best (just like chefs in training who may cook two dishes the right way and the wrong way to actually see {taste} the difference it makes).

And a lot depends on the interpolation process one uses, and on what's available, Adobe just added two new bicubic options for example to tailor the result to what's in an image). Remember that with any digital image just resampling is not the only thing you'd do - as mentioned above a couple of times you may want to sharpen for example, which can improve the appearance of something markedly (and there are multiple ways of doing this, even with just a single sharpening filter). This is just like what happens within most digital cameras and scanners where there are processes applied to the raw image data according to the manufacturer's chosing.

Well, for the sake of the "plenty" part of the above mentioned houses, let's hope I'm not one of their financial important customers, or a customer that otherwise brings prestige, since it undoubly would be the last the "plenty" part saw of me.
But one would not have any way of knowing.


I agree that it may be a waste of resources to scan at an extremely high resolution if all one needs is a web image, however.....

If I want a 5x7" image on a 72ppi monitor, I can obtain directly it by scanning at a resolution which will give me a 360x500p image. However, I've found that better results are obtained if I scan at 2 or 3 times greater resolution than I need, process the image to my satisfaction, then downsize to 360x500p. If done carefully, this will produce a better image than a lower resolution scan.
Exactamundo.

Einion

tk04
03-30-2006, 04:22 AM
If I want a 5x7" image on a 72ppi monitor, I can obtain directly it by scanning at a resolution which will give me a 360x500p image. However, I've found that better results are obtained if I scan at 2 or 3 times greater resolution than I need, process the image to my satisfaction, then downsize to 360x500p. If done carefully, this will produce a better image than a lower resolution scan. Of course, an absurdly high resolution scan is ridiculous for such an application.

Richard

I find that scanning to a slightly higher resolution, like 100 as the final aim, gives the best result - if it's to be converted to web. All this of course based on knowing the sixe of the final image and what it is to be used for. It's also an adaption to the software I have available in Photoshop. Essensially this software adds adjacent pixels fairly mechanically, but in somewhat different ways.

tk04
03-30-2006, 04:33 AM
Most printers are equipped with resampling software, which will change the resolution of any input to whatever is appropriate for the printer.
Richard

Tess says she used to scann professionally at a somewhat higher resolution than the final print, but I believe she said the final print was 300. (Can't see the post at the moment).

But at higher resolutions, it's nothing to gain on scanning for instance a photo at 2000 if the printer takes 1000 - of course, as long as the photo is to be printed on the 1000 printer and you know the final size. I'm not talking about enlargements.

TessMurdock
03-30-2006, 10:17 AM
Thank you for the additional information. I have posted some additional comments.

Quote:
Originally Posted by TessMurdock
most computer monitors are capable of displaying photos at 72 dpi, hence many of your photos are blown up greatly.

Not really. 'Blown up' would imply that the image is actually enlarged in pixel terms, when in fact you're changing the resolution and the actual dimensions.

Hi Einion -- Good discussion. Sorry about my lack of technical jargon. What I meant was that the image would display all of the pixels available, and since the monitor generally displays at 72 ppi, then an image with a lot of pixels would blow up the display to a very large dimension. In reality, the image may be quite small but have a very dense ppi.

Quote:
Originally Posted by TessMurdock
A minimum dpi for printing purposes is about 300 dpi.

Printing how? If one is printing an image digitally the ideal resolution varies from output device to output device (200-250ppi isn't uncommon).

If one is making separations where the image becomes screened (lpi - lines per inch) the rule of thumb is 1.5-2 times the output line screen; although you can get away with less than 1.5x for some images. If you include stochastic screening you have to do things a little differently again.

Tess: Well, I didn't know we were going this technically deep. I am just speaking from my experience with print bureaus, specifically, ones that use your basic Heidelberg Press. I learned as much as I needed to ensure I delivered a good product to the pre-press staff and they advised me to submit work at twice the resolution of the final printing.

When we are talking about your basic HP photo printer, the same rule of thumb holds, twice the resolution of the final output in dpi. I think the additional technical information is great but I have found a good formula that works, is supported by the technical professionals that I work with, and my final printed products have won awards.

I appreciate your in depth clarifications.

Tess

Einion
03-31-2006, 06:17 AM
Hi Einion -- Good discussion. Sorry about my lack of technical jargon. What I meant was that the image would display all of the pixels available, and since the monitor generally displays at 72 ppi, then an image with a lot of pixels would blow up the display to a very large dimension. In reality, the image may be quite small but have a very dense ppi.
Ah sorry I misunderstood; yep that's exactly what happens - a 300ppi image 3" wide will display a great deal larger than this on a monitor.

Well, I didn't know we were going this technically deep. I am just speaking from my experience with print bureaus... they advised me to submit work at twice the resolution of the final printing.
I used to work in one of those :) In case it's of any use you can get away with quite a lot (for instance if there is a surcharge for output time, as in some London bureaus).

For critical images you want to aim for twice the line screen (or close to it, you don't have to be too concerned if it's a little off or if you have to enlarge an image slightly in XPress) for anything less important, even within the same document, you can drop down to 1.5x the line screen. For something like a ghosted background image you can actually go down to close to 1x and actually get away with it.

There should still be a section on resolution and its effect on image quality in the manuals that come with Photoshop if you want to check.

When we are talking about your basic HP photo printer, the same rule of thumb holds, twice the resolution of the final output in dpi.
But the resolution of some inkjet printers is 1200dpi (and up)...

Einion

tk04
03-31-2006, 11:56 AM
I found an interesting article online that explains some of the tradeoff with scanning pictures. This article is probably some years old. It seems like the scanner in this example scans optically up to 300 dpi and thereafter interpolate. This technology develops very fast. But since it hasn't been mentioned in this debate - it's important to aware that the scanner too is a digital medium, not just the computer.

Still, this article explains some of the trade-offs very well. It's absolutely worth while to notice the conclusion in this article, even if the available scanning technology is better now. My private scanner for instance scans up to 2400 dpi optically.

"We should say that interpolation is very desirable for line art. If your output device is a 1200 dpi imagesetter, then scan line art at 1200 dpi. This reduces jaggies considerably on line art. All interpolated pixels are either white or black in Line art, there's no tough decisions.

But for color or grayscale scans, interpolation is far less suitable. There are so many more colors possible, and we don't need nearly that much resolution for color anyway."

http://www.scantips.com/interpol.html

Einion
03-31-2006, 12:41 PM
But since it hasn't been mentioned in this debate - it's important to aware that the scanner too is a digital medium...
I must be reading a different thread :lol:

Thanks for the link to that page on Wayne Fulton's site but just to make sure there is no confusion, here:
"We should say that interpolation is very desirable for line art.... But for color or grayscale scans, interpolation is far less suitable."
Wayne is referring to interpolating upwards, not downwards.


Ken, this page from the same site may be of some help with regard to scanning transparencies:
http://www.scantips.com/basics12.html

Einion

FriendCarol
04-01-2006, 01:04 AM
Wayne is referring to interpolating upwards, not downwards.Plain English translation for those of us not quite so conversant with the concepts: Interpolation most commonly (secondary school mathematics) means putting a point between other points: adding a point one hasn't actually measured between points one has measured. (We usually run across it in graphs of straight lines, or curves.) This could be called interpolating upwards: it projects or estimates information we didn't start with.

If one has scanned at higher resolution, and reduces the resolution, the new projected points are in some way (many different ways, potentially) averaged from many more points that were actually measured. This is like interpolating downwards.

Just translating for any readers who may happen along later. :D

tk04
04-01-2006, 04:36 AM
Thanks for the link to that page on Wayne Fulton's site but just to make sure there is no confusion, here:

Wayne is referring to interpolating upwards, not downwards.

Einion

Sorry, it wasn't my intention to cause confusion. There will two steps in a process where you scan a slide to present it as an image on the web.

Point is that if you are to scan a slide to show it on the web you will normally want to scan at a higher resolution than 72. That's because you need a higher resolution to enlarge the image. But if you scan with a much larger resolution than necessary it might imply upward interpolation by the scanner. That depends on the scanner. My scanner is optical up to 2400 dpi so it won't interpolate at high resolutions.

The difference between optical and digital scanning is the same as the one you have in digi cameras with optical and digital zoom. Digital scanning implies upward interpolation by the scanner. After a certain resolution, say 300 dpi, the scanner doesn' read the color information directly from the original(the pixels), but adds color-information by putting in pixels after a mathematical formula. This adding of pixel will be based on the color-information from the original, but it will be done after a mathematical formula. That's the first step in this process.

If you scan with a much higher resolution than "necessary", you will then have to reduce the resolution of the scanned image in the computer - from for instance 1000 to 72, to adapt it to the standard resolution on the web. It means that the computer interpolate downwards by adding together pixels. That's the second step.

So in worse case, the color-information of the original will be digitally slightly changed twice.

As I started out with - it's quite complicated to decide where the optimal scanning resolution will be. That also depends on the scanner, the size of the memory of the computer, the software and the original.

bigflea
04-01-2006, 09:27 AM
Is it correct then to assume that with a scanner capable of high resolution scans, it is then possible to scan for "target", and avoid the problems that occur from interpolation? This seems to be Karin's original point about potential problems with both high and low resolution scans, ie. the difference between the scan resolution and the intended use of the image creating interpolated information that is either wrong or of little visual value? And when we see a jpg. that is produced from a 35mm slide, in which the color is generalized or exaggeratedly skewed toward selected colors, is that a result of interpolation because of the scan resolution? In interpolating down, wouldn't there potentially be the same problem as up, since the coloring between scanned points could change in the original image?
Ken

Einion
04-01-2006, 02:38 PM
Is it correct then to assume that with a scanner capable of high resolution scans, it is then possible to scan for "target"...
If you are sure you only need just a single image at a specific size then you could do this but it is typically very difficult to scan to an exact size from a small original like a 35mm transparency. It is a good idea to scan larger and make smaller copies as necessary - this is routine in the industry.

...and avoid the problems that occur from interpolation?
The problems with interpolation are primarily with interpolating upwards, not downwards (with some exceptions as stated earlier).

In interpolating down, wouldn't there potentially be the same problem as up...
Short answer: nope.

Einion

Richard Saylor
04-01-2006, 03:30 PM
It can be difficult to get an accurate color print from an inkjet printer. However, there are factors other than interpolation involved in accurate color reproduction. (Actually, I have never had a color problem which resulted from interpolation.) The main problem is getting the little nozzles to squirt out the right amount of the right color of ink at the right place in the print. The printer has the formidable task of translating digital information (which is better suited for a computer monitor and additive RGB color mixing) into tiny dots of colored ink.

A modest amount of resampling (upward or downward) is not going to mess up the color. Don't worry about scanning to a target resolution. Unless there are computer resource issues, it will do no harm to scan at the highest possible optical resolution, process the image, then resample to a resolution appropriate for your print. If your printer is smart enough, you can even skip the last step.

bigflea
04-01-2006, 10:41 PM
So far it seems there is no debate about the need to scan a 35mm slide to a higher resolution then is needed for the intended purpose, which in the case of my question was a facsimili jpg. Still it seems unclear re. how high a resolution is optimal for the intended purpose.

Re. interpolation, the color problem is one of blurring only, as a low optical resolution scanner interpolates upward?

If there is no color problem in interpolating down, from a high resolution optical scanner, then why is the problem of downward interpolation a concern at all ( in color facsimilization,if that is a word)?

In terms of a facsimili of a 35mm color slide, are the color skewing and generalizations a problem from interpolation with a low optical resolution scanner, or are they from some other cause, assuming the slide is a good facsimili of the original work?

Is is accurate to think of high optical scan resolution as the truest digital facsimili possible re. a 35mm slide?
Ken

tk04
04-02-2006, 05:13 AM
So far it seems there is no debate about the need to scan a 35mm slide to a higher resolution then is needed for the intended purpose, which in the case of my question was a facsimili jpg. Still it seems unclear re. how high a resolution is optimal for the intended purpose.

Re. interpolation, the color problem is one of blurring only, as a low optical resolution scanner interpolates upward?

If there is no color problem in interpolating down, from a high resolution optical scanner, then why is the problem of downward interpolation a concern at all ( in color facsimilization,if that is a word)?

In terms of a facsimili of a 35mm color slide, are the color skewing and generalizations a problem from interpolation with a low optical resolution scanner, or are they from some other cause, assuming the slide is a good facsimili of the original work?

Is is accurate to think of high optical scan resolution as the truest digital facsimili possible re. a 35mm slide?
Ken

If you scan at a low resolution you are very restricted on how much you can enlarge the picture later on. That's the rationale for this idea that "more information is always better than less information". It means that if you scan an image at a low resolution planning to use it as 2x2 you can't change you mind later on and use it as 4x4. It will become "grainy" - that's inevitable. In that sense it's always better to have a higher resolution than a lower resolution.

The changes to the image as a result of interpolation downward - in means reducing the resolution of an image - is very subtle. Whether you as an artist bother about this, or not, is a subjective thing. It's like putting in a extra brush-stroke in a picture. Lots of people don't understand why those few brush-strokes matter.

Two colored images - one scanned optically at 100 and the other at 1000 - and both reduced to 72 to be used on the web - won't be identical, and the one scanned at 100 will be closer to the original. But if this was about putting famility photos from the digi-camera out on the web, I don't believe anybody would notice, or bother about, the difference.

The bias you get by scanning optically at 600, and thereafter reducing it to 72, compared to scanning at 100, isn't huge. It's larger if you scan at 1000 or 2000 and reduce it to 72, but still an untrained eye by someone who is unfamiliar with your work, will probably not notice.

FriendCarol
04-02-2006, 09:09 AM
Two colored images - one scanned optically at 100 and the other at 1000 - and both reduced to 72 to be used on the web - won't be identical, and the one scanned at 100 will be closer to the original. (emphasis addded)This could only be true if the reduction algorithm were very badly chosen. Averages are, on average, better data. That's why all students are taught to take 3 measurements the first time they take any science course -- weigh everything 3 times, find its length 3 times, etc. Then take the average of the 3 measurements and use that as the accurate one.

The algorithm of setting a pixel to match its neighbor is, of course, not an averaging algorithm. It certainly would not be a great choice for reducing the information by a factor of 100 under most circumstances.

tk04
04-02-2006, 02:47 PM
This could only be true if the reduction algorithm were very badly chosen. Averages are, on average, better data. That's why all students are taught to take 3 measurements the first time they take any science course -- weigh everything 3 times, find its length 3 times, etc. Then take the average of the 3 measurements and use that as the accurate one.



Is this a statement that you will like to withdraw?

FriendCarol
04-02-2006, 03:07 PM
Why? That averages are, on average, better data is certainly true. In the U.S., chemistry students, physics students, etc. are always taught in their very first lab to take 3 measurements of everything. So what have I said that isn't true?

In this case, by 'data' we mean a measurement of the value (color value -- RGB or whatever) of a point (not mathematical point, , but very small area) on something being scanned. If we want to display a data point that represents a corresponding area on the original, it is still better to oversample the data and find an average (using bicubic or whatever algorithm) than to simply sample it at the resolution of the final, or less than twice that resolution.

Even as I type this I am waiting for the laptop hooked to my cheap scanner to finish processing an image I just sampled at 600dpi. My eventual goal is to put it in a forum here, so I'm oversampling by a huge margin. The reason I'm doing that is watercolor on watercolor paper doesn't scan very well at all. I'm doing the best I can -- though it will very likely take an hour -- to get a decent file. When it finally transfers the file into my ancient PhotoShop (version 3, Limited Edition), at least I'll have a fair amount of data to manipulate.

When I reduce the file size by altering the resolution and reducing the number of pixels, that will also take a long time. But I'm far more likely to get colors approximating my original if I do it this way. And I'm just lying here typing on my old online PC waiting for it to happen; it's not a problem for me, today, if it takes a couple hours.

I recognize that there is a waste here in time and efficiency, but that only matters commercially -- if I (or my client) were paying for time or computational power, or if I were in danger of missing a deadline for the end result. But if we're only talking about image quality, oversampling results in data that can be manipulated to give a better result. Always... except in time series analysis when period is unknown, or where the image is something simple like stripes of given color(s).

FriendCarol
04-02-2006, 03:18 PM
The image is in PhotoShop now, and it's good... In fact, I can easily make out the actual texture of the paper at this resolution! The problem, of course, is that the file is 5.5Mg at this point, and physically (as a display) would be huge. PhotoShop has reduced it by a factor of 8 to fit it on the screen.

If I had scanned this at 100dpi, I would certainly not be able to make out the texture of the paper on the result. Maybe I'll scan this twice, and show the results. Examples might be more persuasive than statistician's arguments. ;)

Be back later.

Richard Saylor
04-02-2006, 03:57 PM
If a scan were perfect, then one could easily determine the exact resolution needed for an intended purpose, provided no processing of the image is anticipated. However, most scans are imperfect, needing at least some color correction, and the intended purpose may have a different aspect ratio than the scanned object. I.e., some processing/manipulation of the scan is usually needed. For this reason, if for no other, it is desirable to work with a greater resolution than the minimum needed for the application. There are enough variables which can affect the scan that it is impossible to determine in advance the minimal additional resolution needed. Fortunately, scanning at a resolution higher than the minimum needed does no harm whatsoever. While it is true that downsampling an image causes some deterioration (mostly blurring or fuzziness), this is easily corrected with a sharpening utility.

FriendCarol
04-02-2006, 05:01 PM
Since this thread is about cameras (or the title is :D ), I started a new thread:
http://www.wetcanvas.com/forums/showthread.php?p=4496075

Einion
04-03-2006, 04:41 AM
Ken, say you owned a Porche and posted a question on a forum about how to do some maintenance on it. Which opinion would you be more inclined to accept: that of a Porche-certified mechanic or someone who works on their Chevy Nova in their spare time?

Einion

bigflea
04-03-2006, 10:10 AM
Don't know if I could make an equation between mechanical troubleshooting, and the problems that seem to be a part of digital technology. Mechanical problems do no defy logic the way digital systems SEEM TO . For example, I was assuming what I thought to be a logical conclusion in thinking that the higher resolution scans, with the highest amount of detail information, when reduced for a jpg., would not present any problems in terms of a facsimili. Yet it is unclear to me how interpolation down comes in to the resolution problem, or if that is the cause of color biases, etc . Interpolation up or down seems only to produce problems in blurring which seems a different problem than color saturation and bias. I certainly know more now about the potential problems and possible solutions for scanning from 35mm slides, and appreciate the debate over different points of view. Same with the digital camera technology.
Ken

Richard Saylor
04-03-2006, 05:38 PM
.....For example, I was assuming what I thought to be a logical conclusion in thinking that the higher resolution scans, with the highest amount of detail information, when reduced for a jpg., would not present any problems in terms of a facsimili.....Your assumption was correct.
.....Yet it is unclear to me how interpolation down comes in to the resolution problem, or if that is the cause of color biases, etc . Interpolation up or down seems only to produce problems in blurring which seems a different problem than color saturation and bias.....For the last time, downsampling causes no problems which can't be corrected in a decent photo editing program.

tk04
04-03-2006, 06:27 PM
I found an informative debate on the net discussing compression - it's two - three years old. So it might not be fully updated about software. An extract from the debate:

"Michael Lowe , feb 04, 2003; 10:31 p.m. Garry, are you in possession of the scanner from which the files were made? If you are, why not just rescan at the lower resolution required for web use? It can't take more than 30 seconds for a low res scan. I initually make two scans, one for web use and one at 4000dpi, thereby avoiding any compression loss.

Garry Edwards (Bradford, U.K)  , feb 05, 2003; 04:04 a.m. Michael, I don't have a scanner. I agree that it would make sense to re-scan at a low resolution, but I simply don't have that option. My scans are outsourced and are produced on a drum scanner.
I simply tell them how big the final print is likely to be and they scan at the resolution needed. This can result in extremely large scans, as in this sample. Unfortunately, when reducing to a small scan most of the fine detail has disappeared and so has the subtle gradation.
However, thanks to you all I now understand the process of re-sampling a bit better and hopefully will be able to do better in the future. "

http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=004Te3

Just for the record, "rules" have to be used with some common sense. Line-drawings for instance seem to be better scanned at higher resolutions.

TessMurdock
04-03-2006, 07:55 PM
Always more to learn about all this stuff. Good discussion all.

Tess

bigflea
04-03-2006, 09:10 PM
What did you say Richard?

And before I forget, what is the cause of color bias in scanning?
Ken
ps. thanks Karin, that link clarifies the problem of compression, which sounds like the cause for color bias in scanning from a good facsimili slide to a jpg.

Thanks everyone for your input which is all helpful.

Richard Saylor
04-03-2006, 09:27 PM
What did you say Richard?Nevermind.:)

And before I forget, what is the cause of color bias in scanning?
KenThe scanner.

Richard Saylor
04-03-2006, 09:44 PM
.....An extract from the debate:

"I simply tell them how big the final print is likely to be and they scan at the resolution needed. This can result in extremely large scans, as in this sample. Unfortunately, when reducing to a small scan most of the fine detail has disappeared and so has the subtle gradation....."Either he or whoever does his scanning is screwing up, or else his photo-editing software is no good. Or perhaps conversion from a lossless format to JPEG is being done too early in the editing process.

FriendCarol
04-03-2006, 10:10 PM
When I scan line drawings, I use the lowest reasonable resolution, usually 100dpi. It's just not necessary to get the color right, so I don't bother to oversample the drawings. It's only if I'm trying to get semi-accurate colors from a painting that I scan at 600dpi, and then wait patiently for an hour for the file to make its way into PhotoShop. But then I'm using a 6-year-old laptop with 64mg RAM. :D Can't really expect anything fast.

Einion
04-04-2006, 06:39 AM
Don't know if I could make an equation between mechanical troubleshooting, and the problems that seem to be a part of digital technology.
Ken, the important part of the analogy wasn't about what was being asked, it was about which type of respondent you'd be most inclined to accept advice from in a certain context... I had hoped that would be clear.

Mechanical problems do no[t] defy logic the way digital systems SEEM TO .
I'm know what you mean but often it's not a flaw in logic, it's simply that one doesn't know enough to draw a worthwhile conclusion. I'm reminded of the phrase 'a little knowledge is a dangerous thing' - you can build a house of cards that appears, with one's limit knowledge of a given subject, to be solid but is in fact on very shaky ground. I did the very same thing the other day when interpreting some stats; I won't bore you with the specifics but quite simply if I had been correct in what I thought the stats showed my conclusion would have been sound. However I wasn't, so my argument was completely spurious and even though based on a perfectly reasonable line of thinking the conclusion was of no value whatsoever.

Anyway, back to the matter at hand - your question was about scanning from transparencies so we need to stick to specifics, and while Carol should be applauded for making the effort to do a comparison, it's not an exact parallel. You need to try two scans from one of your trannies on the scanner you'll be using and then compare the results; if you're happy with the results of scanning for a specific size then that's fine. But as I mentioned above you will find it extremely difficult to scan to a specific pixel dimension from a small original with any accuracy (it's actually difficult even with large originals and higher-spec scanning software that allows you to zoom in on the scan preview).

Interpolation up or down seems only to produce problems in blurring which seems a different problem than color saturation and bias.
That's correct. And that is why sharpening filters are included.

One point I want to make here about resizing being in some way inherently a bad idea in case it isn't completely obvious. Let's forget scanning for a moment since a lot of people these days don't take photos of their work and then try to scan the prints, the negatives or the transparencies. You cannot take a digital photograph at the exact size you might require for posting online, so what do you do? You take a picture at a certain resolution, you crop it and then you reduce it to the size you want. This, as a rule, won't give you as good a result as when scanning because most digital cameras save JPEGs while when you scan you don't have any initial image changes to worry about (what are called compression artefacts - the little blocks you see most clearly on very poor images on web pages) but despite this it is absolutely routine and is done every day. In fact every picture posted on this site that was taken with a digital camera had to be resized and the vast majority of them were JPEGs to begin with and yet the quality isn't seen to be poor overall, is it?


This thread really has gone nowhere useful in the last few pages and while the 'discussion' might appear to be useful it's not helping anyone who might be unfamiliar with the subject. For anyone who is unsure: if you do a scan and it's close to what you need but you have to resample it a little (e.g. you need a width of 300 but you have 342) that's going to cause much more evident problems as a rule than scanning at a higher resolution, cropping and then resizing as I recommend at the foot of page 1 of this thread (e.g. you need 300 but have 500 or more). If you don't believe this then please, by all means try comparisons for yourself. I stress the plural here, don't think a single comparative test will show you anything that you can apply universally to all situations - scanning original art, photos and transparencies are all quite different and the scanner you're using and the software that runs it will all play a large part in the results, as will the software (and the resampling method) you might use to manipulate the image afterwards.

Einion

tk04
04-04-2006, 08:32 AM
The reason why interpolation downward causes smaller colorshifts is that each pixel contains color-information - red, green, blue. When you compress the image, the computer will add together this color-information from surrounding pixels and make a new pixel from this information - and at the same time reduce the number of pixels. How this is done depends on the specific allogaritm used.

This "blurring" that arise from downsizing is due the allogaritm used - bicubic. Not all allogarims will cause "blurring". The reason why bicubic is preferred, in spite of the blurring , is that it gives better color-registration than other possible methods. But this is also something that is discussed all the time. In the thread I provided there is a reference to this - about using bicubic for upward interpolation and a bilinear (? - I don't see the article at the moment) for downsizing.

When you then downsize with a bicubic allogaritm you can get rid of some of the "blurring" by using sharpening tools in Photoshop. It's the color-shift that gives the real trouble. Essensially because the color-information in the image has changed slightly by the compression, even with a bicubic allogoritm. And it's not easy to correct it because it's not caused by an overall bias in the whole image, but due to smaller changes in individual pixels. Some areas of the picture will also have larger changes than others, depending on the original color-composition.

Photos give additional color-problems - due to the original photo.

That said - this color-shifts are very subtle. You will have to know the original to see them. If this was about reducing the size of 8 MP digigal familiy photos to show it on the web noone will bother. And it's also true that you will have a fairly descent representation of your original, even if it's compressed. But you get the best representation of your original if you can avoid compression. I find that scanning slightly higher than the target is necessary because the web-compression uses some pixels - if you have too low a resolution you get "jagging".

Richard Saylor
04-04-2006, 11:21 AM
Ken,

If you plan to share your scans with other people using different computers, then you will have far greater color inaccuracies to be concerned about than Karin's minute color shifts due to resampling. http://www.colormatters.com/chameleon.html

bigflea
04-04-2006, 11:45 AM
Einion,
Ofcourse you are right about the problem of attempting to logically problem solve when there is little or no working knowledge or familiarity with the technology. And that was my meaning in saying digital technology "seems to" defy logic. Because I posted the question as someone who has no real technological knowledge re. digital scanning etc. I can only listen to what has been said and try to learn what may be basic information, without trying to come to any clear conclusions.
For me the debate has been very informative re. what is basic to the technology, and that is a starting point. At this point I would have to begin making comparisons of good facsimili slides at different resolutions, and put the results into the context of the information as it has been presented. I have tried to rely on professional scanning, and been disappointed to find that there are no comparisons between slides and jpgs. produced by the same processor. Without understanding the original color work, it seems the processing of slides to jpgs. is unpredictable at best. Because the work is often about color information vs. line information, it seems to be an additional problem which processors are not aware of or concerned with.

Thanks again for all the input.
ken

tk04
04-04-2006, 01:04 PM
Ken,

If you plan to share your scans with other people using different computers, then you will have far greater color inaccuracies to be concerned about than Karin's minute color shifts due to resampling. http://www.colormatters.com/chameleon.html

It's several factors that might influence how an image appears on screens around the world. In general, these qualities have improved with the technology, as everything else concerning digital technology. Norwegian consumers are very much on the frontier when it comes to having the newest technology because of high incomes, but in other national markeds there might be more "outdated" eqipment. That's of course something to be aware of if your website is meant for a worldwide audience and you have good equipment yourself. If it's a family site meant for friends and uncles that have more or less the same equipment as your own, you don't have to bother about it.

But this discussion is about the starting point. If the starting point isn't correct, it won't get better later on. The guestion from Ken was something like: I still don't understand how compression influences colors, does anybody know? That's what I have explained. If you first grasp the logic of it, it's not that difficult. But you do have to understand the logic - Ken is right about that - and most people find these things quite confusing. So it's perfectly normal to be confused about it and not to know it all.

bigflea
04-04-2006, 08:42 PM
thanks Richard,
that link described what might be the worst case scenario for work that is attempting to rely on color as communication. It is easy to see what the author means by comparing the directness of Van Gogh's color works to the apparently circuitous pathway of web color communication.
In a color work, where the note of color between two forms is lost, which was an entirely different note than either adjacent note, the context of the massings of notes can be entirely changed. What effect takes place visually, for the viewer, can be entirely altered by the context of notes together. So the question of how to attain a true facsimili is to some degree an all or nothing proposition. In other kinds of works a loss of a color division may not be critical to the visual meaning of the work. That is the substance of my concern about how to produce facsimilis in scanning.
Ken

tk04
04-05-2006, 05:52 AM
Why? That averages are, on average, better data is certainly true. In the U.S., chemistry students, physics students, etc. are always taught in their very first lab to take 3 measurements of everything. So what have I said that isn't true?


Just as a sidenote to Carol about compression. If this is some sort of basic understanding how it works, it might be the reason why you haven't grasped compression fully on the theoretical level. The example you refer to will be similar to taking 3 identical scans of the same picture - and then take the average of the SAME pixel. It's not directly comparable to taking the average of 3 pixels in the same scan.

Otherwise - it's a lot of factors that can affect scans and the appropriate resolution - so it might be factors that make a given resolution better in some cases than others.

Einion
04-05-2006, 07:10 AM
I can only listen to what has been said and try to learn what may be basic information, without trying to come to any clear conclusions.
For me the debate has been very informative re. what is basic to the technology, and that is a starting point.
Actually you could have learned a lot more if there hadn't been any debate on the issue as many of the points raised were irrelevant tangents I assure you.

At this point I would have to begin making comparisons of good facsimili slides at different resolutions, and put the results into the context of the information as it has been presented. I have tried to rely on professional scanning, and been disappointed to find that there are no comparisons between slides and jpgs. produced by the same processor. Without understanding the original color work, it seems the processing of slides to jpgs. is unpredictable at best.
So you've gotten professional scans done and they have been poor reproductions of the transparencies? Have you tried discussing this with them to try to get a closer result? There are a lot of technical issues here, not the least of which is the hardware but the skill of the operator can play a large part too. If you aren't happy with the scan you should be able to have it redone at no charge, although of course there is a lot you can do with the colour after you get the scan yourself.

Because the work is often about color information vs. line information, it seems to be an additional problem which processors are not aware of or concerned with.
You're trying to get the scan to look right on your screen right? But your monitor won't be accurate to begin with if it's at all typical of PCs, so even if it does look 'perfect' on a calibrated monitor it might still look off when you get it home; I think this may be part of the problem here.


But this discussion is about the starting point. If the starting point isn't correct, it won't get better later on.
I presume you're talking about resolution here? I do hope it's not in relation to colour, which it could easily be interpreted to be about.

The guestion from Ken was something like: I still don't understand how compression influences colors, does anybody know? That's what I have explained.
No you haven't explained anything, you've given your opinion on the subject.

Just as a sidenote to Carol about compression. If this is some sort of basic understanding how it works, it might be the reason why you haven't grasped compression fully on the theoretical level. The example you refer to will be similar to taking 3 identical scans of the same picture - and then take the average of the SAME pixel. It's not directly comparable to taking the average of 3 pixels in the same scan.
We're muddying the waters more here. It's not my intent to embarrass you but you've said outright above that you're not a professional in this area and it's pretty clear to me - and I am a professional in this area - that you don't fully grasp this topic on a theoretical level, resolution and interpolation to name two areas, so why are you making this point in this way? One has to be very careful about using words like compression when it has a very specific meaning in this context and it's not something we've been directly talking about here. I don't want readers, who may be swayed by your insistence or the confident tone with which you state your opinions, to be mislead about things and much of what you've posted above is misleading.

Einion

bigflea
04-05-2006, 08:58 AM
Einion,
Yes I have tried discussing results with the professional processors, and that has been part of the difficulty. One, my lack of knowledge of technical issues re. scanning processing makes discussion limited, and two, the apparent disinterest on the part of processors, who are very comfortable in their use of technical language, which is used to explain circumstance is such a way that discussion cannot proceed toward a possible new conclusion re. an improved result.

Re. the monitor, ofcourse you may be right that it could be a contributing factor. However, I have not had any problem with digital shots themselves, taken with a good quality camera, under right light conditions. If I want to use a digital camera to photograph my works, I use a white canopy in outdoor light, and take the shot under the diffused bright light. This gives a very good result, and is reliable from painting to painting, in most cases. Another painter friend has used a digital camera with a tif file that has given very good results, which when emailed to me, has very good quality in terms of a facsimilis to the original works(paintings using complex colorings). So I have not had a general problem on the monitor that I can see. It is only in the scanning from slides to a cd for use on the internet that I have seen very unreliable results. Unforntunately I have not been able to resolve the problems with the processors I have used in the past. That was disappointing because this particular processor had always done such an outstanding job with the 35mm slides.
Ken

Einion
04-05-2006, 09:24 AM
Yes I have tried discussing results with the professional processors, and that has been part of the difficulty. ...the apparent disinterest on the part of processors, who are very comfortable in their use of technical language, which is used to explain circumstance is such a way that discussion cannot proceed toward a possible new conclusion re. an improved result.
I know what you mean I suspect. This is a bad sign - they're hiding behind their terminology. Even though it can be hard to get some things across to a non-specialist you can easily put most concepts into normal English with only a little effort if they're bothered to try. Without knowing more I can't be more help but if we assume the scans aren't great then they should be able to easily recognise this by looking at them (do they ever show you the scan on their monitors with the trannie on hand to compare?) If we assume the scans are of reasonable quality in terms of colour rendition then obviously they would want to insist that they're fine, which I would do in their shoes, although of course individual standards do vary so 'acceptable' isn't going to be universal by any means unfortunately.

Re. the monitor, ofcourse you may be right that it could be a contributing factor. However, I have not had any problem with digital shots themselves, taken with a good quality camera, under right light conditions. If I want to use a digital camera to photograph my works, I use a white canopy in outdoor light, and take the shot under the diffused bright light. This gives a very good result, and is reliable from painting to painting, in most cases. Another painter friend has used a digital camera with a tif file that has given very good results, which when emailed to me, has very good quality in terms of a facsimilis to the original works(paintings using complex colorings).
Okay, that is a good start - at least you know that images from two different sources show with reasonably accuracy on your computer.

Unforntunately I have not been able to resolve the problems with the processors I have used in the past. That was disappointing because this particular processor had always done such an outstanding job with the 35mm slides.
Shame, it does sound like more their problem than anything then. BTW, do they give you .PSD documents, TIFFs or JPEGs on the CD? I presume they are scanning in RGB mode and not giving you CMYK files?

Einion

tk04
04-05-2006, 12:30 PM
I don't want readers, who may be swayed by your insistence or the confident tone with which you state your opinions, to be mislead about things and much of what you've posted above is misleading.

Einion

There are more trade-offs involved in scanning photos. So with photos there could be reasons for scanning at high levels.

A mislead reader of this thread will just take another scan. So in this case, I can't say I understand the risk of it - whoever represents the risk. It's much more of a problem to give bad advice about oil-painting for instance - since those effects won't show until many years later. But then - the basic idea behind WC is that each person is responsible for his or her own statements, and nobody can blaim anybody if they listen to something that is wrong.

I consider this discussion about me, and my qualifications, somewhat off topic. And I don't feel comfortable about it. In the future, I will prefer you to ask me for permission first before you start to express opinions about me publicly in the forums.

Not that I am obliged to answer this sort of qustions. It's even forbidden under the user-agreement to reveal personal info about other members in the forums that they haven't given themselves. But since you have brought it up. I don't claim any kind of specific skills on WC because I'm not interesting in displaying them. At least that's one reason for it. If I for instance said that I was a web-designer, and used that as an argument in a debate like this, I would have felt obliged to both document my claim about these qualifications and give examples of my work related to it - so that people could see what sort of level I'm on. Of course, everybody can participate in debates and put forward their opinion in a normal manner - without showing off credentials. But if I referred to qualifications as an argument in my favor in a debate, I would have felt obliged to document them.

tk04
04-05-2006, 12:40 PM
I presume you're talking about resolution here? I do hope it's not in relation to colour, which it could easily be interpreted to be about.


Could you be more specific, please.

Einion
04-05-2006, 02:20 PM
There are more trade-offs involved in scanning photos.
Compared to what? What exactly does this comment relate to?

A mislead reader of this thread will just take another scan. So in this case, I can't say I understand the risk of it - whoever represents the risk. It's much more of a problem to give bad advice about oil-painting for instance - since those effects won't show until many years later.
Granted, but people often don't cross-check and they may labour under misapprehensions that they pick up along the way. This can happen frequently in an online-forum context and I'm trying to limit the impact here because there's something I can do about it.

But then - the basic idea behind WC is that each person is responsible for his or her own statements, and nobody can blaim anybody if they listen to something that is wrong.
If you want to think of it that way that's fine.

We're now on the fifth page of an off-topic discussion largely because of your repeated posts about certain points to do with scanning, scans and resolution issues, in the face of more than one person's disagreement on some of the key issues. One side is right about each of these and surely it matters which side it is... it should to Ken and any other readers hoping to learn the best way to deal with practical scanning.

I consider this discussion about me, and my qualifications, somewhat off topic.
Yes I'm sure you would but I don't see any 'discussion' about you. You're the one who mentioned indirectly that you're not a professional in this area earlier in this thread or have you forgotten? So nobody asked you your qualifications although they would be free to and you don't have to answer of course; I was actually seriously thinking of doing exactly that to get some idea of your level/expertise with regard to the hardware and software issues.

In the future, I will prefer you to ask me for permission first before you start to express opinions about me publicly in the forums.
Oh I'm sorry, how exactly do you think that what I said above is not in the same vein as your comment to Carol "it might be the reason why you haven't grasped compression fully on the theoretical level"?

It's even forbidden under the user-agreement to reveal personal info about other members in the forums that they haven't given themselves.
:confused:

Could you be more specific, please.
It's quite simple, are you talking about resolution or colour here:
But this discussion is about the starting point. If the starting point isn't correct, it won't get better later on.

Of course, everybody can participate in debates and put forward their opinion...
This part of your post is quite right: members are free to participate in discussions of this nature, but as you should be fully aware of by now other members are allowed to shoot holes in information they see as wrong and just to be absolutely clear on this point to anyone reading it doesn't matter if they are right. However, if a discussion goes back and forth on the same ground a couple of times it's perfectly reasonable for someone to ask for some proof to be posted (just like on the issue of magenta in a spectrum) instead of just getting the same view reposted in different words - that is precisely what has led to some of the endless disagreements in this forum and the solution is really quite simple.

Einion

tk04
04-05-2006, 05:51 PM
In Norwegian this "thingy" that is used for scanning slides is called a "transparent adapter" - and to scan from transparent adapters. I don't know if that would be the english term - probably possible to check that by a seach on the net. It was a question about it somewhere here.

Richard Saylor
04-06-2006, 01:10 AM
In Norwegian this "thingy" that is used for scanning slides is called a "transparent adapter" - and to scan from transparent adapters.Is this something to use with a flatbed? A film scanner requires no special device for slides unless they are mounted, in which case a slide holder is convenient.

tk04
04-06-2006, 07:00 AM
Is this something to use with a flatbed? A film scanner requires no special device for slides unless they are mounted, in which case a slide holder is convenient.

Several flatbed scanners have this device. I went somewhat up in price because I wanted a good one. I bought it mostly because I wanted to able to use slides in the computer in the same way as photos. But the quality is quite amazing. If you know enough about photography to make appropriate corrections of photos in Photoshop, the obstacle to making private paper-copies of slides by the computer, within a reasonable size, is the quality of the ink in the printer, not the quality of the scan.

LEIGH RUST
04-15-2006, 04:16 AM
As far as colour reproduction goes I'm a big fan of the Panasonic Lumix range. They have some great features such as Mega O.I.S, making it easier to capture things as you see them by turning the flash off and using an internal floating lens whichit focuses through to remove camera shake. I really love that they use the Leica lens, giving wonderful reproduction of colour. Even their entry level models are fantastic, they use the lumix lens which has been modelled on the leica. Funnily enough I've found the lumix lens has better control of the red-violet spectrum that the leica has a little trouble handling perfectly. I work in electronics retail and 90% of my customers buy panasonics once they see the feature sets. You should find the current entry level unit for about $350 AU.