Disc rez inconsistency

[dadsm3]

Hello: When I get my 35mm films developed I usually get a set of prints made and a disc as well, just to play with on Picasa, email around and print the odd one for fun at home. My problem is that the quality of the resolution on the discs varies widely....from 200-300KB's to over 6MB's. It sure makes a big diff when I make a large print. My lab doesn't make the discs in-house, they just don't cut the negs and send them out. I only get shrugs when I ask why.....I always pay the same (about $10CDN, or $8USD). Ususally the b&w's are a little higher rez than the colour, but even they vary 1.5MB's to over 6MB's.
Anybody have similar issues? Should I be asking for high-res scans? I really don't want to pay big money for them, I'm not advanced enough for photoshop and my printer isn't up to snuff anyways.
Mike

 

[jlw]

I'm not sure whether you're concerned here about pixel count, resolution, or file size.

Pixel count is the number of pixels vertically and horizontally, and is the best measure of the actual data size of the scan. For example, a scan of a 35mm negative made with a 2000 ppi scanner will be roughly 2000 pixels high x 3000 pixels wide. This number is determined by the settings used when the film was scanned, and it shouldn't vary subsequently unless you "resample" the scan to change its pixel count.

To find out an image's pixel count, you need an application that can read it out. Sometimes your operating system can show it directly as part of the file's property information; Photoshop and most other image editors can doisplay it; I don't know anything about Picasa, since it's Windows-only, so I don't know where to tell you to look for this info.

"Resolution" is pixel count applied to a specific output size. To choose an easy-math example, if you took the above scan and printed it on 10x15-inch paper, the resolution would be 200 pixels per inch (2000 pixels divided by 10 inches, and 3000 pixels divided by 15 inches.) You can work this out for any output size if you know the pixel resolution, as above. So, a scan of a given pixel size will have different resolutions for different output sizes.

File size is the number of kilobytes or megabytes that your scan occupies on a disc. This is what I suspect you're seeing when you say the image varies from 1.5mb to 6mb. This is not necessarily any thing to worry about -- it's quite common for scans with the same pixel dimensions to have different file sizes.

The reason is that most digital-image file formats, such as JPEG, use mathematical compression techniques to make the file size smaller. An image that has large blocks of similarly-colored areas will compress much more than one that has many areas of tiny, intricate detail. So, for example, if your first picture shows mostly blue sky and the next shows a crowd of people's faces, the first picture will compress to a much smaller size than the second one, even if the scans have exactly the same pixel count.

So, to find out what's going on, the first thing you need to do is examine the images you get back on the disc and see whether or not they all have the same pixel count. If they're being scanned at different pixel counts, something really weird is going on in the lab and you're entitled to find out what it is. It actually would be easier for them to scan everything at the same pixel count, so if you're getting different counts, it's probably a malfunction or a mistake.

Once you know that you're comparing images of the same pixel count, you can look at the file sizes and see if they're realistically varying because of compression. If you've got two files with the same pixel size, and one of them has a file size of 200kb and prints poorly, while the next has a file size of 6mb and prints well, it suggests that your lab is using compression settings that are much too high for some of your images. This would degrade image quality and might account for your poor results. What you can tell the lab to make them stop doing that, though... well, that's an entirely different problem!

 

[aad]

I've seen that as well. If the same lab is running at different resolutions (not necessarily file sizes) it means they have several options, and you're getting whatever one was last run. If I get a CD, ometimes I get DeLuxe, sometimes standard.

 

[dadsm3]

Many thanks for your thoughful response. Yes, you're right, they're all the same pixel count, 2059x1373. However, the file sizes are approx 464KB. Thats seems very small to me. The scans look pretty bad. What do you think?

 

[jlw]

To find out what file sizes are appropriate, I fired up Photoshop CS2, opened a 24-bit color image, resized it to 2059x1373, and saved it at several different format and compression settings. Here are the file sizes I got:

TIFF, uncompressed: 8.2mb
JPEG, maximum quality setting: 756kb
JPEG, high quality setting: 440kb
JPEG, medium quality setting: 280kb
JPEG, low quality setting: 232kb

This suggests that your scans are being saved at something similar to Photoshop's JPEG High quality setting, which should be capable of producing good results. So if your scans are of poor quality, there must be some other reason -- such as inferior scanning hardware, or incorrect settings of the scanning software.

How about posting one or two worst-case examples, so we can take a look at them and suggest what might be wrong?

 

[dadsm3]

Here's one.....let me try a colour next.

 

[dadsm3]

Here's a colour....

 

[dadsm3]

The colour is at 795MKB, the B&W is 415KB. OK I guess....but I thought I should blow my my kids digitals out of the water with my M6, and I'm not. Shot at 6.0MB follows (with the M6 and 50mm DR 'cron follows....

 

[dadsm3]

This is one one with 6.0MB....no comparison. I'm not looking for artistic critiques cuz I know I suck. I just want to get the best value for my buck quality wise.
But Italian-Canadian girls are beautiful, aren't they?
Mike

 

[jlw]

These look good to me in terms of amount of detail (in the hair etc.) relative to the pixel dimensions, so I'd say the scans you're getting are sharp. The tonality is a bit harsh, though, and the color image seems to have a slight pattern that suggests that the scanner is applying a little more unsharp masking than you really need.

Still, not bad for a generic scan. (Scans are like prints: you can get high-volume ones done on an automated machine that provide adequate quality for a low price, or handmade custom ones that take longer and cost more.) It would be an interesting exercise if you could hand over one of the same negs to a friend who owns a dedicated film scanner and knows how to use it, and let him/her take a crack at it; then you could compare it to your processor's scan.

Don't feel too bad that your results don't "blow away" those of your kids' digicams. Basically, at the scan resolution you're getting, you're producing roughly the equivalent of 3-megapixel images, so a lot of the extra image quality of your hardware is getting lost.

It's kind of like comparing an excellent lens to a mediocre lens on the basis of machine-made 4x6 prints -- you really won't see a lot of difference at that size. Once you start pushing up the enlargement ratio (or pixel count), though, that's where the difference would start to show.

A 4000-pixel consumer film scanner could produce 24-megapixel images from your 35mm film frames, and at those dimensions (assuming your film and camera technique are good enough) you'd be able to see a difference vs. the average consumer digicam.


PS -- Nice-lookin' family you've got there! I envy you people who live in a houseful of models...

 

[dadsm3]

Thanks jlw, you're very kind. I've asked questions on photonet and have been very rudely responded to.....thanks for the helpful advice. But don't you think the 1st two shots don't measure up? If I'm paying hard coin don't you think I should be getting premium quality? Really, if guys like us are going to stick with film, someone should be looking after us. We shouldn't have to shoot a roll of film and settle for what's 'acceptable'.....
After all, we're still the only ones that print anything. Why should we accept all that loss of information because that's all the digital labs will give us?
I don't want this to devolve into a useless digi-vs-film thread, but if film is going to survive they better give me the results I earned with my Leica optics, especially if I'm willing to pay for it. Am I being naive?
Mike

 

[jlw]

Well, again, it's a bit like printing. Generic 4x6 machine prints are basically a commodity item, and at the price you can't expect much more than a usable result from "average" negatives. If you want higher-quality results, or special handling of "difficult" negatives, you need to go to a custom lab where your negatives are printed individually by a technician rather than on an automated machine.

Scanning works the same way. For a processor, scans with processing are a commodity service that they handle as quickly as possible on automated equipment, so they can give you a low price. I suspect that they assume that most people getting scans want them only for onscreen viewing or emailing, so anything more than 3-megapixel resolution would actually be a handicap for them. (No doubt you've been emailed some snapshots from a yutzy relative who bought an 8-megapixel digicam and doesn't realize he should be downsampling the shots before attaching them to the message...? I've got an uncle who does this all the time!)

I think the most sensible way to handle it would be to treat your 3-mp scans the same way you'd treat proof prints or contact sheets. Use them for filing in an image database and/or for viewing onscreen to select your best shots in a batch. Then single out the best shots for more quality-oriented treatment, in the form of higher-resolution, individually balanced scans.

Your choices for where to get these high-quality scans are the same as for higher-quality prints: You can go to a custom service, or learn to do it yourself.

If you want to use a service, you might check to see if your current lab has a "premium" scanning service that offers higher-resolution scans; provides greater bit depth (16-bit instead of 8-bit; gives you more latitude to adjust your image in software); and uses less compression. Or you might wind up needing to send your film to a trade shop that does professional-level scanning.

(Incidentally, that's one great thing about shooting film vs. using a digital camera -- with a digicam, you're stuck with what you've captured, wheareas with film you can always do another scan when better equipment or technology becomes available.)

The do-it-yourself option involves buying your own film scanner; you can get a good one for considerably less than the cost of a typical Leica lens. You'll need to make sure your computer has the horsepower to handle high-resolution images, and you'll need to do some learning of software and techniques, but there's lots of info available on that. Once you've made this investment of time and money, you'll be able to get your scans and final images to look exactly the way you want -- it's kind of like the difference between buying machine prints and printing in your own darkroom.