Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
11-29-2006, 10:04 AM   #1
Inactive Account




Join Date: Sep 2006
Posts: 2
"Banding" and visual perception, does anyone know?

All this talk about banding that some see and some don't (I see something, but I'm not sure what ) made me think of something. Human perception is in general very good ad seeing patterns, in fact it is designed to do that. So if you present it with noise, what is it to do? Well, maybe manufacture patterns.

So maybe these "bands" are just a result of the visual perception process, like the Martian canals. Are there any experts here who can throw more light on the subject?

Gerard Stafleu

11-29-2006, 10:30 AM   #2
Senior Member




Join Date: Nov 2006
Location: Kirkland, WA
Photos: Albums
Posts: 173
Are you trying to say the Martians DIDN'T build those canals?
11-29-2006, 11:03 AM   #3
Veteran Member




Join Date: Nov 2006
Location: Cincinnati, OH
Posts: 419
Okay ...

First, I wish people would stop calling the high ISO noise pattern "banding." I know you're not to blame, many people have been calling it banding, but banding is a very specific problem.

If you want to see great examples of "banding" then look at many of the D200 images that landed on the web right after Nikon released that camera. Banding looks like "gaps" in the photo. It's also one of the reasons I decided to quit using Nikon gear earlier this year.

The K10D (like many DSLRs) suffers from strong vertical noise patterns when using high ISOs for long exposures ... or if you underexpose high ISO images and "push" process them.

My D70 used to suffer from MUCH WORSE high ISO noise patterns.

Long story short ... the vertical pattern noise is there. It's NOT banding, but it's some ugly noise. "Maybe" Pentax can tweak the firmware to cover it up better, but regardless of whether Pentax can fix the noise issue the K10D is still a great camera and the noise isn't as bad as other DSLRs I've used.
11-29-2006, 11:28 AM   #4
Veteran Member
arbutusq's Avatar

Join Date: Sep 2006
Location: Vancouver BC canada
Photos: Gallery
Posts: 380
I've seen the high ISO pattern noise that seems to be in some K10D samples using my D70. IMO the cuprit is the blue channel which is frequently boosted in incancescent and flash images.

In one of the Nikon forums there was speculation about using color balancing filters (like what you used to use with daylight films under incandescent light) to balance out the readouts. I never tried it but apparantly it works quite well at reducing noise.

11-29-2006, 01:16 PM   #5
Veteran Member
jeffkrol's Avatar

Join Date: Sep 2006
Location: Wisconsin USA
Photos: Gallery
Posts: 8,434
And in "red light situations" it is the red channel. Blue or red I've split some of the more offending photos and just see some "mottled" patterns in one or the other channels. Mr. Chong over at Fred Miranda is apparently going to check it out. If you care for some "sample images" see this. Nov 28th date is 2 Nikons and a D PEF and a k10 DNG
Index of /Postings

K10 DNG (note this, will get to it again) images when pushed hard (2 stops under base exposure, then pushed 3 stops in silkypics) exhibit some really bad noise, even in comparison to other 10mp cameras (at least the 2 Nikons). There is some really "bad stuff" that happens to the DNG.. See for yourself, too hard to explain.
Now to be clear, I don't really trust the RAW converters handling of DNG, and this could be a big part of the problem. My copy of Pentax software won't handle the DNG's and nobody's posting PEF's, which again few could use anyways. Actually this is a historic moment in this regard. Bunches of people not needing to download updates ect. to see the files themselves. This is NOT an endorsement of DNG BTW...
And personally it's a fairly non-issue at this point (the patterning"). I've seen other more questionable things that I prefer to leave to myself. If there is an issue, it will surface.
ADDENDUM: If the sensor is not read off the long dimension side (top or bottom) I assume that type of banding as in the Nikon, is not possible. If it is read off right to left (or visa versa) then vertical banding (as defined (?)) could not occur. Is this correct?




QuoteOriginally posted by arbutusq Quote
I've seen the high ISO pattern noise that seems to be in some K10D samples using my D70. IMO the cuprit is the blue channel which is frequently boosted in incancescent and flash images.

In one of the Nikon forums there was speculation about using color balancing filters (like what you used to use with daylight films under incandescent light) to balance out the readouts. I never tried it but apparantly it works quite well at reducing noise.
11-29-2006, 02:25 PM   #6
Veteran Member
arbutusq's Avatar

Join Date: Sep 2006
Location: Vancouver BC canada
Photos: Gallery
Posts: 380
QuoteOriginally posted by jeffkrol Quote
ADDENDUM: If the sensor is not read off the long dimension side (top or bottom) I assume that type of banding as in the Nikon, is not possible. If it is read off right to left (or visa versa) then vertical banding (as defined (?)) could not occur. Is this correct?
Are you referring to the D200 banding? I can't comment on that since I don't own one. (Of course that didn't stop me from commenting on this thread) I don't know how readout direction would affect pattern noise but if you read from left to right wouldn't you then have horizontal banding.....My brain gets mushy when things start getting too technical.
11-29-2006, 04:52 PM   #7
Veteran Member
jeffkrol's Avatar

Join Date: Sep 2006
Location: Wisconsin USA
Photos: Gallery
Posts: 8,434
That's the way I think of it. The Nikon had a 4 channel readout, right to left (or left to right). So the banding caused by errors or whatever in the readout produced horizontal banding. At least this is my understanding of the cause and effect. The Pentax uses a 2 channel readout and if the same, and caused by a same problem as Nikon, would produce horizontal banding. The vertical banding, if having the same cause would only be produced by a read error if the sensor is read up to down. Sensor data transfer is serial, one point after another till a row is read. Then on to the next row. But to complicate this, data is passed from one row to the next. See diagram. But that implies only 1 channel readout. Not 2 as in Pentax or 4 as in Nikon. Kind of like an interlaced tv
http://www.shortcourses.com/how/sensors/ccd_readout.gif
11-29-2006, 05:20 PM   #8
Veteran Member




Join Date: Nov 2006
Location: Sydney, Australia
Photos: Gallery
Posts: 593
QuoteOriginally posted by gstafleu Quote
All this talk about banding that some see and some don't (I see something, but I'm not sure what ) made me think of something. Human perception is in general very good ad seeing patterns, in fact it is designed to do that. So if you present it with noise, what is it to do? Well, maybe manufacture patterns.

So maybe these "bands" are just a result of the visual perception process, like the Martian canals. Are there any experts here who can throw more light on the subject?

Gerard Stafleu
You may be onto something there.

Some see the banding and others don't makes me think it is either a monitor problem or some other problem. I do not see the issue on either of my two CRT monitors. Also, apparently it is more pronounced on LCD monitors, so this also adds fuel to my specualtion. IMO, I think the monitor definitely has a role in this as there maybe a line up of the colour pixels of the screen that just happen to co-incide with the noise of the high ISO shots.

Also, as it is colour, it could very well be a colour blindness problem or partial colour blindness or some other such anomoly of the eyes of the people who can see the "banding"(or "stranding" as a term I coined) or it could have the opposite effect and affect those who cannot see the "Banding/stranding".

The colour blind test involves different coloured dots with a number/letter/shape hidden in these dots in a different colour/colours. People who are colour blind cannot see the number/letter/shape. I do believe that 12% of males are colour blind so I wonder if 12% see the issue? I am definitely NOT colour blind but I cannot see the "banding/stranding". See a link here as it may open your eyes(sorry about the pun

Color blindness - Wikipedia, the free encyclopedia

This COULD combine with some peoples colour blndness or non colour blindnes so that they see a pattern.

Just another thought on the matter.

11-29-2006, 06:11 PM   #9
Site Supporter
slip's Avatar

Join Date: Oct 2006
Location: 2 hours north of toronto ontario canada
Photos: Gallery | Albums
Posts: 3,515
QuoteOriginally posted by Lance B Quote
You may be onto something there.

Some see the banding and others don't makes me think it is either a monitor problem or some other problem. I do not see the issue on either of my two CRT monitors. Also, apparently it is more pronounced on LCD monitors, so this also adds fuel to my specualtion. IMO, I think the monitor definitely has a role in this as there maybe a line up of the colour pixels of the screen that just happen to co-incide with the noise of the high ISO shots.

Also, as it is colour, it could very well be a colour blindness problem or partial colour blindness or some other such anomoly of the eyes of the people who can see the "banding"(or "stranding" as a term I coined) or it could have the opposite effect and affect those who cannot see the "Banding/stranding".

The colour blind test involves different coloured dots with a number/letter/shape hidden in these dots in a different colour/colours. People who are colour blind cannot see the number/letter/shape. I do believe that 12% of males are colour blind so I wonder if 12% see the issue? I am definitely NOT colour blind but I cannot see the "banding/stranding". See a link here as it may open your eyes(sorry about the pun

Color blindness - Wikipedia, the free encyclopedia

This COULD combine with some peoples colour blndness or non colour blindnes so that they see a pattern.

Just another thought on the matter.
here is likely why banding seems worse on a LCD instead of a CRT:

most CRT can show many more colors, hues, tones, and better dithering then a LCD (one color next to the other color will blend better). so, in effect the banding that shows up on the CRT would likely be less pronounced then on a LCD.

unless you pay in the thousands for a LCD, they are only capable or 65,000+ colors. a good CRT can produce millions (LCD has 16 bit color, CRT 24 bit(millions) and 32 bit colors(billions)

hope this makes sense

cheers

randy
11-29-2006, 08:34 PM   #10
Site Supporter
Sailor's Avatar

Join Date: Nov 2006
Location: Coastal Texas
Photos: Gallery
Posts: 18,206
Too Many Variables

QuoteOriginally posted by gstafleu Quote
All this talk about banding that some see and some don't (I see something, but I'm not sure what ) made me think of something. Human perception is in general very good ad seeing patterns, in fact it is designed to do that. So if you present it with noise, what is it to do? Well, maybe manufacture patterns.

So maybe these "bands" are just a result of the visual perception process, like the Martian canals. Are there any experts here who can throw more light on the subject?

Gerard Stafleu
Interesting comment, Gerard. My K10D came Monday evening, and - of course - following all of the hoopla, I shot some pictures around the house at 1600. The images were noisy, but ISO 1600 on my DS2 looks dirty to me as well (for me, 1600 is like a temporary spare tire - it'll get you home, but I'd replace it before I head for the gymkhana), but I see no patterns in the noise. Testing cameras (or anything else for that matter) at the limits of their performance envelope is difficult; the signal-to-noise ratio (literally in this case) is so low that any environmental variable can lead to results that are hard to reproduce.

A couple of years ago, I saw (or really heard) this whole discussion played out around Yamaha's introduction of a new sampled acoustic piano board for its ES 90 and Moftif synthesizers. Shortly after the board began shipping, a few people began complaining about an artificact - a high frequency noise that drove some to rage. Most of us couldn't hear it even when we tried (which ignited further rage), and my board still sounds as good as you could expect for a $250 addition to a general purpose, performance synth. But who knows why most of us were satisfied while others were not; something is probably there in the board's output that can be detected under particular conditions, but most of us have combinations of amps, speakers, ears and brains that filter it out.

Jerry
11-29-2006, 11:57 PM   #11
Veteran Member




Join Date: Sep 2006
Location: Göteborg, Sweden
Posts: 810
Photo_mom posted here series o pictures including the discussed salt package in her kitchen. In the thread over at DPR the question "can you see it in prints?" was asked. Well, I printed it, and so did the guy asking. We could both see the pattern noise on the prints (one Canon and one Epson printer). That should rule out all these theories about monitors.
At least I believe it would.

/Jonas
11-30-2006, 06:36 AM   #12
Veteran Member
jeffkrol's Avatar

Join Date: Sep 2006
Location: Wisconsin USA
Photos: Gallery
Posts: 8,434
Maybe your printing a screen capture :O Just kidding. FWIW I never thought the monitor explaination held much water.
11-30-2006, 10:52 AM   #13
Veteran Member




Join Date: Sep 2006
Location: Göteborg, Sweden
Posts: 810
Aha, that explains the low resolution!!

Tracy's thread: K10D noise/bands/stripes: Pentax SLR Talk Forum: Digital Photography Review
Picture used: Epsom salt, the big version she posted a link to
Printed straight off from the jpeg, Epson R800, glossy, Qimage, interpolation: med, med, pyramid-5 and sharpening 2 (low)

I think her picture is interesting as it shows a real world situation to me. She developed a PEF using Pentax software and to me the package a salt could just as well have been a face instead, a causual portrait. I guess the background are white walls, partly in shadow.

So, there is no pushing and no loudspeaker speaker standing on a back shelf or anything strange, just a picture just as anyone could have taken it.

Sample differences? Firmware fix possible? Let's wait and see for a while what happens.

regards,

/Jonas
11-30-2006, 12:35 PM   #14
Veteran Member
jeffkrol's Avatar

Join Date: Sep 2006
Location: Wisconsin USA
Photos: Gallery
Posts: 8,434
Not sure you saw mskad's images... Thought this was a nice real world comparison. If I imagine really hard (sorry sometimes you just have to add humor) I can almost see the tone changing from right to left in the tablecloth (B to R). Actually this is a bit more disconcerting than a bit of coherent noise....
Anthony...: Pentax SLR Talk Forum: Digital Photography Review
11-30-2006, 06:09 PM   #15
Veteran Member




Join Date: Sep 2006
Location: Göteborg, Sweden
Posts: 810
QuoteOriginally posted by jeffkrol Quote
Not sure you saw mskad's images... Thought this was a nice real world comparison. If I imagine really hard (sorry sometimes you just have to add humor) I can almost see the tone changing from right to left in the tablecloth (B to R). Actually this is a bit more disconcerting than a bit of coherent noise....
I missed that one.
Uh oh.
I can clearly see both the pattern noise and the tone shifting, no need of a lot of fantasy for that. If the comparision with the Nikon camera hadn't been there I would have missed that and just thought it was the light that changed over the picture width. This tonality change or variation, is it also only showing itself in high ISO pictures?
I wonder if these two problems are connected some way technically, or if they are independent? (Maybe I should read the whole long thread over there and check what people think...) (And there is some pattern noise in the Nikon picture as well, but not as prominent.)

I don't know what to think really. I was earlier today looking at a series of pictures taken at ISO1600, bad light, (indoor school situation with many pupils performing a musical piece). I think that both the pattern noise and the color shift as seen in mskads pictures would have ruined the whole series of pictures in that case.

How bad is this? I again guess it depends from if you need to use the camera at high ISO, say from ISO1000 and up? (I have problems seeing it at ISO800 allthough I saw a report somewhere about it.)

Thank you for pointing it out,

/Jonas
Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
camera, dslr, patterns, perception, photography
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
Post your "Park Bench" "or "Picnic Table" images tessfully Mini-Challenges, Games, and Photo Stories 1345 5 Hours Ago 04:22 PM
Don't say Pentax "Q" in French ... "Q" = "cul" = "A--" Jean Poitiers Pentax Q 52 11-10-2013 06:25 AM
Sports "Highside Exit" took 1st Place in DPReview "Missed It by THAT much, Part 1" Challenge MRRiley Post Your Photos! 27 02-21-2010 08:26 PM
K1000 body "ding" - defect, accident, or "feature"? dannywho Pentax Film SLR Discussion 6 10-08-2009 06:11 PM
K2000/K-m "banding" problems? shuttervox Pentax DSLR Discussion 2 02-14-2009 10:05 PM



All times are GMT -7. The time now is 09:28 PM. | See also: NikonForums.com, part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top