Technical questions re: digital photography

Factual Questions
1

I started a different thread in Café Society about an image.

Image: http://i28.photobucket.com/albums/c...Egg03_16047.jpg

Thread: Does this image speak to you? - Cafe Society - Straight Dope Message Board

That thread was concerned with artistic merit (or lack thereof) of an image I took. Many in that thread posted that the images were really grainy, which leads me to some technical questions. Part one of this was posted there without reply, leading me to believe that this is probably a better area to pose those questions.

Part one is an edit from that thread; part two (begins ***) is posted for the first time.

I have a Canon PowerShot A720IS. I had it set on superfine resolution and large image size because that records at 8Mb, which I assumed would be the sharpest resolution. Actually, the images come out at 4Mb for some reason. I was also shooting at “ISO 1600,” which I know with film would mean more grain, but I figured the large capture would counterbalance that.

I recently took a few pictures at work and the grain (when printed) was dismaying, and that was in prints that were only 4x6. An art teacher said I should have shot them with available light instead of flash—which seems counterintuitive to me…when pictures were grainy from 35mm, I knew I should have used flash. In this case, the camera was on a tripod maybe 8 feet away (flash is good to 12 feet). Did the ambient light fool the meter?

I used to shoot a lot with a Nikon FM and a Minolta X-700, so I understand things like stopping down for greater depth of field, faster shutter speeds to freeze action, and so on. But perhaps my film experience is causing me to do the wrong thing with digital.

Another thing I hadn’t known when I bought the camera a couple months ago: it has very limited f/stops. I just assumed it would go to f/16 or maybe even f/32, but it only goes from f/2.8 (4.8 tele) to f/8. It has a full range of shutter speeds (up to 1/2000 sec in most modes). Just a tangent for anybody thinking of adding one to his camera bag.

***The owners’ manual says it has two shutters, when I wouldn’t think it has even one. After all, it probably uses the same sensors to show the image in the LCD as it uses to capture. Taking a picture would seem to be just a matter of switching from viewing to recording.

I made test shots that indicate that the ISO seems to be the biggest factor in generating awful “grain.” I use quotation marks because it’s beyond grain, like the difference between a grain of rice and a rice cake. Hopefully by staying at or below ISO 400 results will be improved.

It would seem to me that making a larger image and having finer resolution are essentially two ways of achieving the same thing. So why are there separate settings? You can easily resize them with Paint, Photoshop, etc. and either the resolution is there or it isn’t.

Here are some technical questions regarding flash.

As with most point-and-shoots, the flash range is limited. From using SLR’s, I supposed that this would operate on similar principles. For example, when buying a flash the guide number would be listed in feet or meters for a given ASA/ISO. If a flash had 100GN with ASA 100 in feet, that meant dividing by the distance to determine aperture. @25 feet, use f/4. @10 feet, use f/10. With more sensitive film, that number would change. For ASA 400 film, the GN would double…for K25 it would be be half.

Issue #1: With this camera the flash distance is 12 feet. Period. It doesn’t matter what ISO is selected. That’s all you get. HUH?

Issue #2: they sell an extra flash for it to get greater distance. It’s designed for a small bracket that threads into the tripod socket. I read in a blog that Canon says it’s “wireless,” which is both true and perhaps misleading: it’s auto-slaved. There are a couple indications that the camera is not compensating for the extra flash.

  1. When pictures are taken at (too) close range, they’re washed out.
  2. If taking pictures at an event like a wedding, flashes from other cameras will trigger it.

So it seems that it’s really designed for when a subject is far enough away, and the flash simply pops, without communication to/from the camera, i.e. no through-the-lens flash metering or anything. Autofocus probably determines the distance, calculates via guide number that it’s too far away and opens as wide as it can, and the hope is that the exposure falls within latitude.

Then it occurred to me that I already have a flash that has a built in slave. I could buy a bracket…cool! Here’s where it gets weird: test photos with my non-Canon flash are uniformly UNDERexposed. Carefully positioning the flash ahead of the camera’s sensor doesn’t rectify the problem. Bouncing the flash off the ceiling helps some but they’re still underexposed at a range within that of the main flash. If the extra flash did nothing, the images would be perfectly exposed, but by adding more light, I have caused not enough light. HUH?

All I can figure is that Canon has found a way to make sure you use their flash, which is both pricey AND primitive, and to ruin any and every other images made with aftermarket flashes.

Any ideas on all this?

2

The ISO setting is the only thing that really matters for digital “grain.” The higher the ISO, the worse the grain will be. You can exacerbate the grain by underexposing the image. but generally, you are better off with a lower ISO and using the flash, rather than cranking the ISO and using poor available light. Also, 8Mpix is a waste with a point-and-shoot camera. The diffraction limt on the lens is less than that (usually). You should pop over to dpreview.com and check out some of their forums.

3

You appear to be confusing resolution and file size here. The resolution of your camera is 8MP (megapixels), where a pixel is an individual picture element (a dot, if you will). The software inside your camera takes the image and compresses it using the JPEG algorithm, leaving you with a picture of approximately 4MB (megabytes). The main factor in file size is the level of compression applied - more compression leading to lower quality pictures.

An independent factor is the resolution at which the picture is captured - it can be the full resolution of the camera, or made smaller to give you a smaller file.

Size does not counterbalance grain - the higher the ISO setting, the more the output from the camera’s sensor is amplified, leading to noise amplification as well. Noise is particularly prevalent in dark areas of an image, since there wasn’t too much signal there to start with. The resolution that your image is being captured at is independent of noise - you can have a small noisy picture or a large clean one.

The ISO is indeed the biggest factor that influences grain - and it’s unfortunate that physically smaller sensors (such as the one in your camera) are more susceptible to noise. A picture taken at ISO 400 with a compact will probably have as much noise as a picture taken at ISO 1600 with a DSLR.

Which two settings are you referring to here? The size of an image can be either the file size or the resolution.

4

I don’t understand this paragraph here… Are you talking about this image that you tried to link to? (Link was broken here, was OK in the other thread) Are you saying you did use the flash, or are you asking if you should have (or shouldn’t have)? I think the exposure is OK on this image - the main problem is the huge amount of noise caused by using such a high gain (“ISO” setting).

I believe most digital cameras have mechanical shutters. I’m not too familiar with CMOS detectors, but with CCDs the image is first stored on the chip, then the electrical charge is transferred to the corner of the CCD to be read out. If light continues to fall onto the CCD while this is happening, you get a smeared image. On my Canon SD800, this smearing is very obvious on the LCD viewfinder - bright light will have a sort of trail below it - because the mechanical shutter is not used in the viewfinder mode. The shutter is used for the actual captured image, so there is no smear there.

Right, it’s not really grain. It’s electronic noise.

The image size setting controls the JPEG compression ratio.
Sorry, I don’t have time to reply to the rest, maybe later…

5

Hmm, this makes my brain hurt. I guess my analogy would be making a Kodachrome 25 on 35mm vs. making a 1600 ASA slide on a 645 medium format, then magnifying to the same size when projecting. The K25 might be a small image size and require more magnification but it’s razor sharp and grainless in resolution…the 1600 medium format might be much more grainy but would require less magnification so the imperfections wouldn’t be amplified as much.

Stated another way, it would seem to me (of little experience with DP) that it’s all about the number of pixels you capture, regardless of what size they’re spread across. If I scatter 4Mb across an 8x10 image size, it’s 1:1 when I have an 8x10 made. If I scatter those 4Mb across a 16x20 image size, it’s a 4:1 ratio but those same 4Mb end up on the finished 8x10. Right?

OK, so it sounds like a practical matter of the small sensor being unequal to the task. I’ll have to chew on that awhile but I have a practical direction to go in. Thanks!

6

Most point and shoot cameras have a “pre-flash” that’s used to help focusing and maybe allow some pre-exposure metering. There might also be some red-eye reduction going on. This pre-flash will trigger your slave strobe prematurely, causing the underexposure that you’re seeing. Slave strobes for digital cameras will usually have adjustments that allow the slave to handle the preflash.
As far as the grain goes, other posts have already nailed it. The high ISOs on digital cameras are mostly for show – you’ll get an image, but it’s going to have a ton of noise on it. Strictly speaking, it’s not actually grain that you’re seeing; it’s closer to static. Each individual sensor on the chip is picking up not only light, but random leakage from adjacent sensors and thermal noise. There is software that can reduce the apparent graininess of the image – e.g. Noise Ninja and Neat Image.
But these software products take a fair amount of computation and often remove some detail, so your best bet is to use the lowest ISO you can get away with.

7

Forgot to mention – the fact that the pre-flash performs pre-exposure metering explains why your slave strobe resulted in an underexposure. The pre-flash set off your strobe, making it appear that you were trying to take a picture in a brightly lit room. So the camera throttled down the exposure (or the flash).

8

Exactly. I find find digital noise very ugly. OTOH, I often find film grain quite appealing and it’s one thing I miss about shooting digital SLRs.

As stated above, ISO is the only factor in digital noise. Increasing the ISO just increases the gain of the digital sensor. Unfortunately, increasing the gain will also increase the interference amongst the pixels in the sensor, creating digital noise. ISO noise is especially pronounced in digital compact point’n’shoot cameras due to the smaller sensor size. DSLRs also have this problem but the noise levels are much lower because of their larger sensors and better noise reduction algorithms.

As for the limited aperture range, this is always the case for P&S cameras due to their physical limitations. For example, f8 on a compact P&S is physically much smaller than f8 on a DSLR. As a result, you’ll get a much greater depth of field with the P&S than you would with a DSLR and you don’t have to sacrifice shutter speed. Higher f-stops such as f22 or possibly even f16 would be so tiny on a P&S that you would get diffraction artifacts. If you need the slow shutter speeds that only f22 can provide, then your only option is to use an ND filter.

9

You’re talking about dpi or dots per inch. The number of pixels you capture will always be the same, it just a matter of how large you want to make the print or how many dots per inch you’ll print out. A 5 megapixel image printed at 300 dpi will be physically much smaller than a 5 megapixel image printed at 100 dpi.

Typically, for an 8x10 print, you’d want to print at least 300 dpi. For larger prints, you’ll probably need more than 300 dpi.

10

No, with regard to the grainy print, I hadn’t posted a link. It’s this one:

But, oops. Going back in and looking at that file, I see that I cropped it because there is a lot of wasted space in the foreground (made it landscape), then reduced the image size…I often put “reduced” images in powerpoints and they look fine at a fraction of their original size. I ended up with a 70Kb image… :smack:

My thinking was that reducing the image size would merely jam the pixels closer together, like a trash compactor and I wasn’t thinking about making an 8x10 or anything. That, coupled with shooting at ISO 1600 (and flash), explains it I guess. I think I’ll go back and crop the raw image again but not reduce the image size, resubmit for a print, and see if the results are substantially better.

“Not really grain…” C’mon, throw me a bone! :stuck_out_tongue:

I’ll have to contemplate that compression ratio thing. Is it basically like zipping and unzipping a large file, with errors creeping in during the process?

Some general questions for the digi-photo-tech-ocracy…for a non-DSLR like mine,

  1. What would you generally need to make a decent 4x6 or 5x7 or 8x10 prints, in terms of Mb in the image?

  2. What settings would you recommend, in terms of ISO, resolution, and compression?

  3. Are there any other benchmarks you might use to determine which prints are likely to be good?

11

Sorry to post 3 times in a row, but I just found my little depth-of-field equivalency chart. According to this, f4 on a P&S will render a DOF equivalent to f11 on an SLR and f8 on a P&S will render a DOF equivalent to f22 on an SLR.

12

So the digital noise might be compared to cranking your car stereo? It gets louder, but so does the distortion…to the point of distraction and possibly blowing your speakers.

Apertures: I don’t get how they make these lenses. Mine has a 6X ratio (optical portion) and back in the day of film SLR, an f/2.8(4.8) with 6X ratio would have been considered pretty fast, weigh a ton, and cost a fortune. IIRC part of the reason these lenses can be made is that they don’t have the same image resolution specs.

Still, I assumed that depth of field wouldn’t change. I thought that was just baked in…in a given format, any 135mm lens at f/5.6 is going to have the same depth of field. Granted, the P/S digitals’ focal lengths don’t correspond to 35mm film but at “normal” lens focal length…?

13

The big distinction is that ZIP compression is lossless while JPEG compression is lossy and once you’ve lost that info, you’ll never get it back. Therefore, it’s in your best interest to compress the JPEG as little as possible. Use a compression setting anywhere from 80% to 90%. Also, keep in mind that compression is not the same thing as reducing the image size or resolution. BTW, what photo editing software are you using? This might help us make more specific comments and recommendations.

I always make sure I print at a minimum 300 dpi.

Always use the lowest ISO setting (if you need more light, use a flash), always shoot at the highest resolution (this gives you more latittute for cropping, if necessary), use moderate compression for JPEGs at the most to conserve picture quality.

Make sure your monitor is properly calibrated to get the best possible colour fidelity. Also, make sure you apply some sharpening before printing. Photos for printing should look a little too sharp for comfort on the screen. Also, NEVER edit your original photo. ALWAYS work on a copy.

14

About a hundred years ago, I read Modern Photography and Popular Photography. They pointed out that f values are really mathematical ratios, i.e. focal length/aperture size.

They said that really, we should have t-values instead of f values, t being for transmission. Two 50mm f/1.4 lenses may be “mathematically” identical but lens coatings, flare, and so on can mean the actually amount of light transmitted to the film can vary, and if your work is technical enough you might the extra precision.

Well if that’s the DOF…although I think I have a “fast” f 2.8 zoom, maybe it’s really three stops slower? I guess there are no free lunches…

15

To understand, it might be helpful to consider in more detail the source of digital noise, because what’s going on is a bit different than the graininess of film.

In order to capture images at higher speeds, high ISO film uses larger photosensitive grains. Because the grain is physically larger, it will get hit by more photons during the exposure, and hence will react to the same overall level of light in less time than a the smaller grains of a lower ISO film would. The tradeoff is effectively lower resolution - grainy images.

When you change the ISO setting on a digital camera, however, you aren’t doing anything to the size of the photoreceptors on the camera’s sensor. What you’re doing is increasing the amount the signal coming off the sensor is amplified before compiling the image. When you snap a picture, each pixel on the sensor reports “I got hit by this much light!” and the camera assembles all the data for you and arranges it into a nice image. By bumping the ISO setting from 400 to 800 (for example) you’re telling the camera to multiply all the pixel reports by 2, allowing you to expose for a half the time or to close your aperture by 1 stop.

The issue is signal to noise ratios. In addition to the light coming in through the lens, there’s various random crap triggering the photosites on the sensor. So long as the random crap is small relative to the amount of light coming through the lens, it’s not a problem. But, if you bump your ISO from 100 to 1600 (four stops), you’ll expose the sensor to just 1/16 the light while the level of random crap remains essentially the same. Now you’ll multiply all the light levels coming off the sensor by 16 to make up for that, and you’re not only multiplying the light that came through the lens but all the random crap as well, and it’s that amplified random crap that makes digital noise.

The solution is to improve your signal to noise ratio. Unfortunately, the current crop of 8 and 10 megapixel point and shoot cameras have crappy signal to noise ratios to begin with, because cramming so many pixels onto the very small sensors of point and shoot cameras have resulted in pixels so tiny that individual pixels see only very tiny amounts of light (low signal). To get any real improvement in digital noise levels at high ISO settings, you have to move to a physically larger sensor, which at the current time means a digital SLR. It’s a damn shame that camera manufacturers seized upon megapixels as the be-all and end-all spec for marketing purposes, because a well-made 3-4 megapixel camera with decent optics could print better pictures at modest sizes than the 8-10 megapixel cameras that everyone is selling that only take low noise shots in ideal lighting conditions. No, you wouldn’t be able to blow them up to 16x20, but who does that? Better performance in sub-optimal light would be far more useful to most users, but to get it you have to step up to the dslr world.

16

I’m not using any editing software except Paint to resize them. Most of the stuff I’ve taken has been just to document events, nothing high art.

So you’re talking about printing from home? I’ve used Walgreens…cheesy, I know, but I only got the camera a few months ago and am still deciding what I want to do with it.

What does 300 dpi translate to in Walgreens terms?

Check. First thing I do is save a copy, then work on that copy. What do you use to sharpen? Photoshop?

17

@Gorsnak: Thanks, I pretty much understand that.

18

Sheesh, we’re really bouncing back and forth here :cool:

Perfect analogy.

I can’t speak for why high quality lenses are cheaper for P&S cameras, but I doubt it has anything to do with the resolution. I’d be curious if anybody else has any insight on this topic. I do know that highly respected optics companies like Leica and Zeiss provide lenses for typical consumer P&S cameras and they still remain affordable. I don’t think it’s just about branding.

DOF isn’t baked in, DOF is dependant on the actual physical size of the aperture and not on the focal length to aperture diameter ratio (f-stop) .

Exactly. What you gain in DOF, you lose in the wide apertures. You’ll find it much harder to throw the background out of focus with a P&S f2.8

That’s your problem. You’re extremely limited with that program. I use Photoshop to do my photo-editing but you don’t need to shell out the big bucks. I highly recommend Picasa as a very capable, free photo-editing program. It’s very easy to use and it will give you far more control than Paint.

19

I was extrapolating that from something I read about video cameras. Essentially if the medium can’t resolve 100 lines per mm, it’s silly to put a lens on the camera that can. So the lenses from Leica, Zeiss, Schneider (on Kodaks?!) et al carry a premium name, but the designers probably didn’t lose a lot of sleep over some aspects of the design. Getting into DSLRs, however, the bar would be raised.

Not sure I’m tracking. I’m saying that if you have some lenses designed for a given format (say, 35mm) built by Canon, Nikon, Pentax, whoever—and say they’re all the same focal length, say prime 135mm f/2.8, all focused at 15 feet… Compare the markings on the barrel for f/8 on all those lenses and you’ll find the ranges are the same for depth of field. I think the chart you mentioned implies the same. And whereas f/8 on a 135mm lens gives a particular depth of field, f/8 on a 28mm or 400mm lens will be drastically different.

I probably just have to remind myself that the minimum aperture isn’t very minimum and it’s designed to keep everything in focus, i.e. idiot-proofed.

The group photo (http://i28.photobucket.com/albums/c222/rhythmace/Madeleine03_10005.jpg) was a debacle. I took that picture with the express purpose of having prints made for each person. But for the most part, I’m just taking snapshots. To give an idea, I posted the link to a bunch I took last weekend, over in the other thread:

http://volunteerism.meetup.com/185/photos/

Not a lot of artistry in most of these and that’s fine. I do hope people will see these pictures on the web and join our volunteer group. Also, I hope to recognize those already in our group, helping out and making a difference. But #35 made me think that as long as I’m at it, when I do snap something I like, I want to be able to make a decent print.

I miss my Nikon FM.

20

MP just means more pixels, not better ones. The sensor size on these P&S cameras aren’t getting bigger, but they’re cramming more pixels on them. This results in more noise in the final image. Changing the resolution of the image won’t fix that since the camera takes the image over the full sensor, then pots it down to the selected resolution.

There are several good software packages out there that help remove noise from final images. Try the free demo of Noise Ninja to start.

P&S cameras are fairly useless above ISO 400 regardless of how high they actually go. I would never shoot the Canon A720 above 200 unless I had no other choice. The only exceptions are a few models by Fuji with the SuperCCD. The F30/31 was the best of the lot, but the follow on F40 and F50 have been a bit of a disappointment.