K-5 Banding and Tingeing

Started Mar 15, 2011 | Discussions
MaKeR Contributing Member • Posts: 806
K-5 Banding and Tingeing

I have a K200D and earlier Pentax DSLRs, and most of the best prime lenses from 14 to 200 mm. The K-5 looks great for me, but I was holding off buying it, pending solutions of some of its teething problems. I don't do flash photography, so the remaining problems with flash do not concern me.

Now that most of the major issues have been addressed, I bought one yesterday, serial number 3945nnn. PhotoME reports the manufacture date as 2011 January 29, and the shutter count as 1 in the EXIF data of my first picture. I didn't perform any serious testing with my previous cameras, and have been very happy with them, and have never required service of cameras nor lenses. I thought it would only be prudent to test my new K-5 for sensor stain, however.

I mounted my D-FA 100 mm Macro lens, and have so far taken 12 pictures, and haven't dismounted it.

Picture number 8 is a JPEG of the overcast sky at ISO 80, f/32, and closest focus, and here it is, unprocessed but for re-sizing and conversion to sRGB:

That looks fine.

I applied some adjustment in ACDSee Pro 2.5, which produced the following:

The gross variation in light intensity is due to the clouds in the sky, so please ignore that.

There is vertical banding and edge tingeing along the top and left side, not to mention several large spots on the sensor (not the sensor stain of early production). This is at ISO 80 on a well-exposed image. I haven't seen mention of banding and edge effects with the K-5, so I'm wondering why they are so obvious in my quick test, and whether they are cause for serious concern.

I wonder if the edge tingeing and the banding are reproducible by others. I guess some fairly strong manipulation is required in order to demonstrate these effects at ISO 80. I presume that they could manifest more readily at higher ISO.

It's probably true that these effects wouldn't be visible in most normal pictures (such as the following 50 percent crop of image number 3, developed from raw in camera and resized by half), but if they are latent at base ISO, I would feel a bit cautious about, and disappointed by, this wonderful camera.

I am happy with my K200D, so even though I could use the upgrade to K-5, now I wonder whether to return this K-5 camera, and monitor progress as users here get to understand it further.

I would appreciate your comments.

Thanks.

John Cafarella Contributing Member • Posts: 986
Re: K-5 Banding and Tingeing

The banding you're describing is posterisation. and in every case I've seen discussed here, the root cause has been that the software used was manipulating a JPG. In some cases it was the low-res JPEG embedded in the raw that was being displayed and manipulated, with very bad results.

Be sure your image is in fact really displaying the RAW.

I've not seen any edge fringing as you're describing.
--
John Cafarella
Melbourne, Australia

paulkienitz
paulkienitz Veteran Member • Posts: 5,379
Re: K-5 Banding and Tingeing

John Cafarella wrote:

The banding you're describing is posterisation.

It sure doesn't look like posterization...

 paulkienitz's gear list:paulkienitz's gear list
Pentax Q Pentax K-3 Pentax smc DA 55-300mm F4.0-5.8 ED Pentax smc DA* 300mm F4.0 ED (IF) SDM Samyang 8mm F3.5 Aspherical IF MC Fisheye +4 more
MightyMike Forum Pro • Posts: 40,525
Re: K-5 Banding and Tingeing

MaKeR wrote:

Dirt (small dots) and dust (big dots)... but the banding is likely due to either or both the AA filter or a nasty effect of the distortion of light due to diffraction at F64 (Aperture = Aperture + Aperture * Magnification) with the lens set to 1:1 F32 becomes F64. yes it could be due to the sensor too but maybe you should try the same test at F2.8

-- hide signature --

Mike from Canada

'I like to think so far outside the box that it would require a telephoto lens just to see the box!' ~ 'My Quote :)'

http://www.airliners.net/search/photo.search?sort_order=views%20DESC&first_this_page=0&page_limit=180&&emailsearch=mighty_mike88%40hotmail.com&thumbnails=

John Cafarella Contributing Member • Posts: 986
Re: K-5 Banding and Tingeing

paulkienitz wrote:

John Cafarella wrote:

The banding you're describing is posterisation.

It sure doesn't look like posterization...

I'll have a look from MY home PC. there are various issues with images when I view at work.

-- hide signature --

John Cafarella
Melbourne, Australia

OP MaKeR Contributing Member • Posts: 806
Re: K-5 Banding and Tingeing

MightyMike wrote:

the banding is likely due to either or both the AA filter or a nasty effect of the distortion of light due to diffraction at F64 (Aperture = Aperture + Aperture * Magnification) with the lens set to 1:1 F32 becomes F64. yes it could be due to the sensor too but maybe you should try the same test at F2.8

Mike, thank-you for your comment.

I can't see how diffraction around the edges of a circular aperture could cause vertical banding, and regular banding at that. However, the AA filter could be involved with this, although I would expect more of a checkerboard pattern in that case.

I will re-do the test at f/2.8 to rule out diffraction. My prediction is that that will render the dirt and dust invisible, but the banding and tingeing will remain.

Thanks.

OP MaKeR Contributing Member • Posts: 806
Re: K-5 Banding and Tingeing

John Cafarella wrote:

paulkienitz wrote:

John Cafarella wrote:

The banding you're describing is posterisation.

It sure doesn't look like posterization...

I'll have a look from MY home PC. there are various issues with images when I view at work.

Thank-you, John and Paul.

I doubt that posterisation is involved here - the histogram is continuous, not discrete.

The banding is regular, at about 120 pixels. (I'm guessing that it's 1/42 the width of the sensor. I don't know the significance of that fraction, though.)

Thanks.

MightyMike Forum Pro • Posts: 40,525
Re: K-5 Banding and Tingeing

I've seen strange effects due to extreme diffraction on a blank image, and its likely the effect are a combination of diffraction and the AA filter... however possibly also related to defects in the optics, glass that just wasn't designed to be used at some apertures/resolution.
--
Mike from Canada

'I like to think so far outside the box that it would require a telephoto lens just to see the box!' ~ 'My Quote :)'

http://www.airliners.net/search/photo.search?sort_order=views%20DESC&first_this_page=0&page_limit=180&&emailsearch=mighty_mike88%40hotmail.com&thumbnails=

paulkienitz
paulkienitz Veteran Member • Posts: 5,379
Re: K-5 Banding and Tingeing

To put this issue in perspective, I bet the same test with my K10D would look vastly worse.

-- hide signature --

K10D, Sig 17-70, DA 55-300, FA 50/1.4 "Billy Bass", M 400/5.6 "the Great Truncheon"

 paulkienitz's gear list:paulkienitz's gear list
Pentax Q Pentax K-3 Pentax smc DA 55-300mm F4.0-5.8 ED Pentax smc DA* 300mm F4.0 ED (IF) SDM Samyang 8mm F3.5 Aspherical IF MC Fisheye +4 more
OP MaKeR Contributing Member • Posts: 806
Re: K-5 Banding and Tingeing

paulkienitz wrote:

To put this issue in perspective, I bet the same test with my K10D would look vastly worse.

I tested my K200D (same sensor as your K10D, but with image processing improvements).

An overcast sky, D FA 100 Macro lens at f/2.8, ISO 100, focus at near limit; default settings in Silkypix, except Cloudy WB:

Same raw file, re-processed for test:

Essentially no banding, but a strong narrow tinge along the bottom edge, and weaker effects along the other edges. (And my sensor needs a good cleaning!)

My conclusion: the edge effects on the K-5 are acceptable, since they are less pronounced than these (when similar processing is performed on a raw file - see my forthcoming post).

Still, this raises the question as to whether these edge effects are present in cameras of other makes, or are only a Pentax artifact. And what gives rise to them? Sensor or image pipeline?

OP MaKeR Contributing Member • Posts: 806
Re: K-5 Banding and Tingeing

MightyMike wrote:

maybe you should try the same test at F2.8

I did that, and took the photo in raw instead of JPEG.

Unprocessed picture (Silkypix DS Pro, defaults except Cloudy WB):

Processed picture (less aggressive adjustments than whatever ACDSee did to the JPEG in my first test):

The vertical banding is still visible (stronger enhancement would render it more visible). So is the edge tingeing along the top, particularly, but also along the left side.

Based on comparison with my K200D, which also has edge tingeing (see my previous post) I can live with the edge effects.

(And the dirt on the sensor may be par for the course.)

The K-5 banding is still a question for me. Is this camera defective, or is the banding weak enough that it virtually never appears as a result of normal processing?

Someone else with a K-5 could help to answer whether the banding is typical, by performing a similar test. Please take a uniformly gray picture with exposure compensation of +1.3, and make the file available.

I'm also curious as to whether other cameras produce similar effects, and what the cause is.

Thanks.

Younes Senior Member • Posts: 1,597
Re: K-5 Banding and Tingeing

OK it looks better than the test with the K5, but you did it with f2.8 , instead of f22....obviously not the same test parameters.

Regards
--
Younes ( Paris, FR)
K100D,K10D,D40 + lenses

OP MaKeR Contributing Member • Posts: 806
Re: K-5 Banding and Tingeing

Younes wrote:

you did it with f2.8 , instead of f22....obviously not the same test parameters.

f/2.8 was at the request of MightyMike, earlier in this thread. f/2.8 eliminates diffaction (at f/32) as a factor.

You can compare f/32 with f/2.8 on the K-5, or compare the K-5 to the K200D when both are at f/2.8.

Diffraction isn't the cause, since both banding and tingeing are visible at f/2.8 on the K-5.

Different lenses were used on the K-5 and K200D (the WR and non-WR versions of the D FA 100 mm Macro, respectively). This eliminates anything specific to the lens used for testing (but not any factor that is particular to the optical design, which the lenses have in common).

In any case, I don't think that this is an optical phenomenon, but an electronic one. However, I am addressing comments made on the forum that suggested the possibility of an optical cause.

Thanks.

dlacoutu Contributing Member • Posts: 811
Re: K-5 Banding and Tingeing

If this is visible at high iso, maybe we could ask GordonBGood to modify his K20 magenta borders removal tool? I'm pretty sure it could be handled the same way...

 dlacoutu's gear list:dlacoutu's gear list
Samsung NX100 Pentax K-5 Samsung NX200 Pentax smc FA 50mm F1.4 Samsung NX 30mm F2 Pancake +6 more
MightyMike Forum Pro • Posts: 40,525
Re: K-5 Banding and Tingeing

I don't see any banding worth noting, possibly some vertical smears but that could just be NR or the demosaicing routine. I still say what you saw was the interaction of great DOF, great diffraction and the AA filter.

however consider this... in the F32(F64) shot how long was the shutter open for compared to the F2.8 (F5.6 assuming you did it at 1:1).

the longer the sensor is active the more chance of heat related issues becoming visible.
--
Mike from Canada

'I like to think so far outside the box that it would require a telephoto lens just to see the box!' ~ 'My Quote :)'

http://www.airliners.net/search/photo.search?sort_order=views%20DESC&first_this_page=0&page_limit=180&&emailsearch=mighty_mike88%40hotmail.com&thumbnails=

Gerry Winterbourne Forum Pro • Posts: 17,518
Re: K-5 Banding and Tingeing

I don't know about banding shown on a bright exposure but here are some comparisons of shots taken with the lens cap on and then opened up in PP. All the shots are at f/8, 1/10s. The dark ones are ar ISO 100, the light ones at ISO 6499.

Don't bother about how dark or bright the colours are - that's a function of the detail of PP and could easily be changed - what matters is the patterns shown.

The first pair is the K20 (original Samsung sensor), then K-7 (later Samsung sensor) and finally K-5 (Sony sensor).

-- hide signature --

Gerry

First camera 1953, first Pentax 1983, first DSLR 2006
http://www.pbase.com/gerrywinterbourne

paulkienitz
paulkienitz Veteran Member • Posts: 5,379
Re: K-5 Banding and Tingeing

MaKeR wrote:

paulkienitz wrote:

To put this issue in perspective, I bet the same test with my K10D would look vastly worse.

I tested my K200D (same sensor as your K10D, but with image processing improvements).

. . . .

Essentially no banding, but a strong narrow tinge along the bottom edge, and weaker effects along the other edges. (And my sensor needs a good cleaning!)

So the K200D apparently fixed the K10D's problem with vertical pattern noise.

The other question is, how much stretch did you apply to the colors in the K200D image, compared to the K Negative Five image? That would give an indication of the absolute magnitude of the effects.

 paulkienitz's gear list:paulkienitz's gear list
Pentax Q Pentax K-3 Pentax smc DA 55-300mm F4.0-5.8 ED Pentax smc DA* 300mm F4.0 ED (IF) SDM Samyang 8mm F3.5 Aspherical IF MC Fisheye +4 more
GordonBGood Veteran Member • Posts: 6,311
Re: K-5 Banding and Tingeing

dlacoutu wrote:

If this is visible at high iso, maybe we could ask GordonBGood to modify his K20 magenta borders removal tool? I'm pretty sure it could be handled the same way...

Dominique, it doesn't really matter if it is visible at high ISO or not; I could likely correct it; however the problem with the K-5 is that the left to right colour tingeing (which looks to be to worst problem) has no corresponding black border along the top and/or bottom of the image to use as a correction model. Also, this tingeing (and edge "blooming") appear to be bright area problems and not black level problems, so even if there were a black border it likely wouldn't give any clues as to the magnitude of the correction to use.

The best that could be done would be to write software that builds a model of the correction required based on a set of test images and then batch process images using that correction model.

Regards, GordonBGood

OP MaKeR Contributing Member • Posts: 806
Re: K-5 Banding and Tingeing

MightyMike wrote:

however consider this... in the F32(F64) shot how long was the shutter open for compared to the F2.8 (F5.6 assuming you did it at 1:1).

the longer the sensor is active the more chance of heat related issues becoming visible.

f/32 is 7 stops from f/2.8, so the shutter was open 128 times as long at f/2.8. Still, it was only a second or so, for which there shouldn't be heat problems. I do lots of 30-second exposures, so if an exposure of a second causes a problem, it would affect me.

Thanks.

OP MaKeR Contributing Member • Posts: 806
Re: K-5 Banding and Tingeing

Gerry, thank-you for your test exposures, which are interesting. Of course, they measure something different from mine, since they are taken with the lens cap on.

I do see a vertical pattern in your K-5 exposures, which corroborates observations with my K-5.

Rich

Keyboard shortcuts:
FForum MMy threads