Slashdot Mirror
Make Your Own Digital Camera ISO Test Target
dpnow writes I run a digital photography site and came across what I thought might be an interesting story. It's about a Cornell university researcher that has reverse-engineered the design of the ISO 12233 resolution test target, used by all the best digital camera testers. These usually cost over $100 but a free pdf download of the target is available. Print it out on a good quality printer and you have your own ISO-spec test target so you can find out how good (or bad) your camera really is! "
I love ideas that make me feel like I got less than my money's worth AFTER THE FACT!
frequent www.dpreview.com and get professional reviews of cameras.
Dpreview carries digital camera reviews dating back to 1996. They are usually very detailed.
Somebody can't read! It said over 100 pounds ($180).
Insert comment here about people of a certain nationality making too many assumptions about units of various things...
-- Steve
the article links to a story which links to the actual content you want. To get to the real content, thus bypassing the advertisements that they wanted you to view in the first place, go here:
. html
http://www.graphics.cornell.edu/~westin/res-chart
Another site with better pictures
Mark
http://www.gpsinformation.org/jack/photo-test/pi cs/lens-tests.html
hope you are able to find to find the site a help
Chris Williams clw7500nc@gmail.com
Or so you can find out how good (or bad) your "good quality" printer is.
-S
--- What parts of "shall make no law", "shall not be infringed", and "shall not be violated" don't you understand?
With description of the optics and details of the resultion measurements is here. He created also his own chart which includes shades of grey for better measurements of MTF50.
Save the bandwidth. Don't use sigs!
Pwned.
Photography - Electronic still-picture cameras - Resolution measurements
ISO 12233:2000 paper version (en) CHF 116,00
116.00 CHF Switzerland Francs = 94.2233 USD United States Dollars
Wow, mine must really be messed up. They are all in black and white! (Have a look at the charts first whydon'tcha?)
Someone obviously didn't even LTFA. The target is black and white.
This is a fun toy to play with, but I'd trust professional reviews.
ISO 12233 Test Chart
Ummm dude... it's black and white.
/.
:P
If you are going to comment at least look at it first... oh wait.. this is
News for nerds... stuff people talk about without reading.
Telcos have alot of dark fibre in the States. Most people assume that's optical fibre...but it's actually moral fibre.
The target has no color.
.. its about resolution and dpi .. most inkjet printers probably won't do .. you'd want to go out and print it on a laser. Hopefully everyone has access to one (in college, at work, or at the local Kinko's print shop).
Anyway
I have no intention of doing what this article talks about, yet it is still very useful to me in another way : instead of telling people I'm rearranging my sock drawer and that's why I can't go out galavanting on Saturday night, I'll tell them I'm testing my digital camera's resolution with a reverse engineered ISO 12233 resolution test target.
Someone should sell t-shirts with this thing on them.
For Sale: ISO Resolution Test Target
$50.
Paypal only, please.
Well, my calibration consists of putting regular black toner in my laser printer and printing on white paper.
How do you calibrate for black and white (not grayscale) ?
May contain traces of nut.
Made from the freshest electrons.
Now, there's interesting:
Westin explains that the ISO specifications can be used without restriction for projects like his, though the copying of a commercially reproduced target is, of course, illegal.
(see, I _did_ rtfa!) So, it's illegal to reproduce the image, but creating a new image from an exact description of the image is legal. Yet that _is_ what 'reproducing the image' is!
The reason for this situation is that the image in question is very unusual, in that it has a freely-usable exact description in existance. But what if an exact (text) description of Mickey Mouse were made? You certainly wouldn't be allowed to create new images from it, and yet it's hard to see how Disney would own the rights to that description... hmm...
Whence? Hence. Whither? Thither.
Whos the man with the biggest pixal camera in his hand! SHAFT! now the IEEE just published a chart to prove it ;-)
Ambient [Servlet Based Webapp Engine]
I think there's something wrong with my eyes. All the lines look crooked to me.
my calibration consists of putting regular black toner in my laser printer and printing on white paper.
Then you seem to ignore dot gain, caused by the mechanical spread of ink or toner across the page and the optical spread of light inside paper.
- Throw away any test charts into the corner of the room
- Toss any rulers on top of the test charts
- Newspapers? - on top of the rulers
- Avoid any brick walls
- Pick up camera (and attach lens if applicable)
- Go outside (yes, it really does exist!)
- Shoot numerous pictures of various subjects, at varying apertures, focal lengths and durations. Using a flash for some of the shots would be a good idea too.
- Make some nice large prints of your efforts
- Do the prints look OK to you? If they do, congratulations - consider the test passed, and you might even have a few prints you can actually use for something as well. If not *now* it's time to retrieve the test charts, rulers etc.
Seriously, the only people that really need these charts are people that are designing or calibrating imaging systems. A charming term that I think was coined over on DPReview to describe everyone else is "measurebator". Believe me, if you've got a lens bad enough to make a difference visible in a print, then you'll know it without any test charts. I had a lens that backfocussed, a Nikon zoom lens I got for my film camera some years ago. I picked up the problem without test charts just fine (I often focus on an eye in portraits), and so out came the rulers, or in my case a newspaper. The largest focussing error in the series of test shots that I took was less than 2mm at a range of 3m.Needless to say, I've never touched a test chart, or any facsimilie thereof, since then. The *only* chart that I do have is a Gretag Macbeth colour chart (it's a grid of 24 coloured squares) to get colour balance correct. I also have a couple of Kodak Grey cards for setting white balance if you want to nit pick and call one of those a "chart".
UNIX? They're not even circumcised! Savages!
Can't anyone write anything tech without feeling the need to throw in a highly-charged buzzword?
If you read the chart's creator's web page, he didn't reverse-engineer anything. He created the chart from the published international specification. That's pretty much the opposite of reverse-engineering: engineering. That is, taking a set of specifications and producing a design that meets them.
But I guess that's not as interesting-sounding.
I don't quite see how taking a picture of "error 500: Internal Server Error" is going to tell me anything one way or the other about my camera. However, it might explain a thing or two about their server...
I'm in a robotics lab working with a CMUcam. This should help us a lot. Those people whining that "real photographers dont use test charts" seem to neglect that some of us are trying to calibrate cameras for sensors and not just pretty pictures.
thanks poster
I loved my inferior camera, and my family snapshots, until I found that they had barely-visible distortions.
Now I have to find an accurate camera and retake all those photos.
for excess traffic. /. just hit him so hard I can't
get anything but "500" errors. Somebody should a warned him.
SLASHDOT: news for people who can't concentrate on work or have no life at all and got tired of yelling back at the TV.
Only look in the mirror to detect a removable blemish.
As for the permanent ones you know enough already.
you can find out how good (or bad) your camera really is!
Or you find out how good or bad your printer is.
This will only work if your printer can reproduce the pattern fairly exactly. The less DPI you have, the worse color management you have the more useless it will be.
Of course, some inkjets will print black as a mixture of colors if you don't explicitly tell them to print as black and white. My HP Deskjet 712 is one of these, although it probably lacks the resolution to produce a faithful representation of this chart and would spit out a warped sheet of paper regardless.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
Testing a digital camera is not that complicated:
:)
:)
1. Take a picture of a red object against white background.
2. Take a picture of a green object against white background.
3. Find a person with zits/razor cuts on his/her face and take a picture of that person.
Compare these pics to what you see in real life. Chances are that your reds will be off and the person will have a skin with more orange in it. If so, throw your camera away, unless you do not care about these things. If your camera is close to the real world, you're lucky
Here are some other things to try:
1. Take a picture at night. Look for purple fringing. This will determine the quality of your lens. If you have an SLR, you should not be concerned: you can change the lens. If you have a point-and-shoot that produces this result, stay away from night shots
2. If you do not have an SLR, try to test the camera for vingetting (dark corners) and barell effects. This can be done by taking a picture of a grid or something like that.
Here's a quick-and-dirty lens test you can do, without all the fancy-schmancy charts:
Tape a newspaper to the wall. Shoot it square-on. Check for flatness of field, distortions, and vignetting at the edges. Shoot it at an angle, at different aperatures, to test for depth of field. Easy!
This link has instructions on how to do this.
-- "You can lead a yak to water, but you can't teach an old dog to make a silk purse out of a pig in a poke" - Opus
.
s amples.html
a chmentid=118010&stc=1 3 1193&page=7&pp=15 a chmentid=180976&stc=1 a chmentid=221673&stc=1
Why does his "the best cameras" aka "pick of the litter" page have 10 (ten) cameras listed for each of the 12 categories? He says the "cameras are not listed in order of preference":
http://www.steves-digicams.com/best_cameras.html
I can see 2-4 choices per, but 10?
Digital camera reviews are no where near as technical and detailed as they need to be to be useful, compare this:
http://www.steves-digicams.com/2003_reviews/exs3_
with this:
http://www.rcgroups.com/forums/attachment.php?att
http://www.rcgroups.com/forums/showthread.php?t=1
http://www.rcgroups.com/forums/attachment.php?att
http://www.rcgroups.com/forums/attachment.php?att
Note the "pre-printed form letter" the one guy gets back from his warranty servicing with the check mark beside the following paragraph:
"Your camera is operating according to factory specification in all modes. The phenomenon you have experienced (an orange halo visible in the bckground after taking some pictures) is not a defect in operation of your camera. It is a function of the geometry of the lens optics. Under certain lighting conditions this effect may be noticeable. Darker backgrounds will minimize or eliminate this effect."
You can no longer find the S3 or anything like it on the market, Casio has probably quietly removed it due to huge numbers of returns and warranty servicing costs. You can only find a few on eBay, and ALL are "open return" or "used return, not tested". NONE sold by happy users.
AFAIAC, digital camera reviews are nearly worthless, no matter who is doing the review.
.
That's not a nice thing to do to your sister you pervert.
We bought our own charts. Oddly enough, We're located in Rochester, NY too. At least those of us not outsourced to china (which was me, too).
Seriously tho, this target is worthless unless you print it on a LARGE piece of paper with an ultra high quality laserjet printer. Ideally you'd want a real laser, not the cheap LED ones, and furthermore you'd need at least 1200 DPI.
Inkjet? Forget it. Dot placement is poor quality- your lines will get fuzzy. So is that aliasing or is that bad printer alignment.
Remember your printer is being sold to make you buy ink. It is NOT a highquality piece of equipment (some high end epsons do don't count;P)
Great for play. Buy the real thing (chrome on glass) if you're serious. I'm surprised CMU doesn't have at least 3 of these- I can walk into any lab and find at least 1.
Just printed the sheet. Printer can't handle 12lpm. Can't handle any slant targets. Can see the vertical swipe target but using a microdensitometer note that the line spacing is of poor quality.
I guess I should junk my 1000$ Lexmark and find something better.
I printed this out on my HP4MV laser printer at 11 x 17 inches. Its 600 dpi are hopelessly inadequate for this test pattern. The best way to print this would be to take it to a print shop that has a good direct-to-film printer (2540dpi or better) and ask them to make a PMT (positive mat transfer) from the file at the highest resolution possible. You'll get a near-perfect print. But you'll pay about $25, and it may not be archival: some PMTs fade over time.
I'm not aware of any printer that prints white. If I'm using orange paper, it will be a black and orange target.
"I'm not impatient. I just hate waiting." - My Dad
I'm not aware of any printer that prints white.
I am. It takes no electrical power; runs manually. But the pixel sizes are inconsistent and probably too large for good quality work (unless tantric). You also don't get any color except white. Also, it generally needs a good snapshot of Natalie Portman in hot grits to prime the pump.
It's probably best to not to take family portraits and act as a Polaroid instamatic. That's just wrong.
modern SLR viewfinders aint what they used to be thanks to AF. Gone are most of the focus aids.
A-men!
I had (well, was borrowing) a Nikon FE-2 that's currently in the shop due to a munged shutter. Manual focus, but the (I don't know the words for any of these things) focusing screen with the split circle in the middle, where if it was in-focus the two halves would line up, and the ring around the central circle would be
absolutely usable, even in low light.
I recently got a Canon EOS Rebel GII. The autofocus is dodgy when not in direct sunlight. The focusing screen contains three little boxes to represent the AF points, which blink when they read as in-focus. That's it. Manually focusing is absolutely unassisted.
A decent viewfinder screen can't be that expensive. Some of us still like to focus manually.
--grendel drago
Laws do not persuade just because they threaten. --Seneca
Come along an hour later and say the same damn thing that the previous post said. Even 30 minutes after it was rebuffed.
It seems like it ought to be possible to print an enlarged version of the test target spread across a number of pages making the resolution of an average consumer printer adequate.
My question is, how big would the output have to be, to cleanly show everything on the target, with something like an Epson C82?
Why is "Error 404: File not found" such a good test image? I don't get it.
Table-ized A.I.
Unless the professional reviews are sponsored by the camera manufacturer or an affiliate. Or they gave the reviewer a free camera in exchange for a favourable review. Or the manufacturer poisoned some web forums with fake comments saying "my (whatever brand) camera is awesome! You all should buy one!"
sustainable living
Which it probably isn't.
This is what that money is paying for - a printout of the target that's guaranteed to be within a certain range of variability from the Right Colors. And to have the Right Aspect Ratio. And the Known Reflective Characteristics.
What you print out on your Epson or HP or whatever, probably using typing paper instead of photo paper, probably with color management off, is not going to work for precise calibration of a camera.
egypt urnash minimal art.
My God, it's full of starrrs....
Assume I was drunk when I posted this.
...but it's not a simple task, so you probably got value for money if you did buy the commercial version. I've documented the project to use the Westin drawing to create a working replica ISO resolution test target here: URL:http://dpnow.com/.dpnow.com/1263.html Did it work? - well, yes - sort of! Ian
Sorry - that should have been http://dpnow.com/1263.html Ian