jlw
Rangefinder camera pedant
Terao said:
If it's blooming, it seems as if it would bloom symmetrically in all directions, the way a video-camera image does. In all the cases I've seen, the streaking has a definite horizontal direction, and always stays aligned vertically with the height of the bright light source.
What I need now is a rainy afternoon with nothing else to do, so I could piddle around with some simple experiments...
jlw
Rangefinder camera pedant
MarcoS said:
Yes, I can see it too. It's not as evident in the Noctilux image because the bright streak (from the small street lamp in the background) lines up fairly closely with a light area in the background. Against a completely black background I suspect it would be more visible.
It's too late and I've got too busy a weekend coming up to do this myself, but here's a suggested test rig for someone who'd like to explore this phenomenon:
-- Get a large sheet of some black, opaque material such as black foamcore board. Given the typical RF lens minimum focusing distance of 1m, you'll probably need a sheet about 0.5m high; the US standard 20x30" sheet should be about big enough.
-- Cut a thin, vertical slot in the center of the sheet, tall enough to occcupy about 1/4 the frame height.
-- Set up the sheet vertically in front of your digital camera, focus on the slot, and then backlight the slot from behind with a very bright light source such as a quartz lamp. Make sure there's no spill around the sheet.
-- Take test shots at various apertures and with the slot at various positions in the frame. This should enable you to determine such things as:
1) Do streaks appear? If so, what amount of overexposure does it take to provoke them?
2) If you get streaks, do they expand uniformly (blooming) or do they run in one specific direction? (The examples I've seen run in a specific direction; I suspect this is because the data is scanned horizontally off the imager.)
3) Is the streaking independent of shutter speed, or not? (I'm guessing it is, since the electronic scan of the imager is much faster than the mechanical shutter curtain's travel.)
4) Is the streaking constant across the field, or does it fade out toward the far edge? (I'd think it would fade out as the excess signal dissipates, but it's difficult to tell on the examples I've got because the backgrounds aren't uniform.)
5) Is the effect independent of ISO setting, or not? (I don't know, but maybe higher ISOs are more likely to streak since this involves greater amplification of the analog signal off the imager. If using a lower ISO definitely reduces the risk of streaks, this would be useful knowledge!)
6) Do different lenses have any effect? (If it's purely an imager phenomenon, they shouldn't -- but lens flares might be mistaken for this type of streaking.)
7) If you've got several different types of digital camera available, comparing them would be an interesting exercise.
You film diehards, with your symmetrical, streak-free, aesthetically pleasing overexposure effects, have my permission to smirk smugly at this point...
Terao
Kiloran
Alternatively, just stop pixel-peeping and go out and shoot 
However, if any M8 owner is genuinely unhappy with what they see as their fundamentally flawed digital camera I'll happily take it off their hands for a small handling fee
However, if any M8 owner is genuinely unhappy with what they see as their fundamentally flawed digital camera I'll happily take it off their hands for a small handling fee
IGMeanwell
Well-known
Banding can be fixed with firmware ... so I don't think anyone really needs to be too worried
plus I am sure that is a big thing controlled by the 6 bit encoding of the lenses...
but like the D200 many people learn to take it in stride, wait for the company to fix it. Or just take care of it post processing
plus I am sure that is a big thing controlled by the 6 bit encoding of the lenses...
but like the D200 many people learn to take it in stride, wait for the company to fix it. Or just take care of it post processing
petermcwerner
Member
There is a very interesting in-depth discussion in the Leica-Camera-User Digital Forum: M8 vs DMR, very different colors in Capture One.
It is worth reading it, although it is in part highly technical
It is worth reading it, although it is in part highly technical
jlw
Rangefinder camera pedant
Terao said:Alternatively, just stop pixel-peeping and go out and shoot
Yeah, but if your artistic conception would be spoiled by purple horizontal streaks running through the final image, it might be a good idea to be aware of the problem so you can plan ahead to avoid it.
This is no different from what we've always had to do in photography. For example, if you use a lens with poor flare control, and there are bright light sources in your picture area, you learn either to recompose to avoid them, or to take advantage of the aesthetic effect they produce.
jlw
Rangefinder camera pedant
petermcwerner said:
Thanks for that link; lots of good info (although you have to skip past a lot of discussion of color balance to find the posts related to horizontal "bleeding" of bright areas.)
Mike Prevette's post provided what seems like a good concise explanation of what's going on:
"The banding however is very hard to correct via firmware. It is a result of the circuitry of the sensor, and how it passes a charge down the row of pixels to the output terminal. It shows up when the sensor has it's gain turned up (high iso) and part of the sensor is exposed to a very bright light source. In order for the sensor to get that pixel data out it has to pass a signal down a row of active pixels to the output port, this causes interference that the other pixels pick up. Hence a band."
As to why people are observing horizontal bleeding with the M8, while most DNR users report they don't observe it, I'm not so sure. One possibility is simply that the DNR people are mis-observing. The effect only shows up when you've got a small, very bright area next to a large, very dark area (it has to be dark for the bleeding to show up.) This is typical of indoor "photojournalism" shots -- the kind of thing that would attract an M8 user -- but not so typical of the types of outdoor and studio photography that seem to be more common subjects for DNR users.
Again, this isn't an M8-specific problem: I've seen it in R-D 1 shots, and people have reported it with various DSLRs as well.
However, there's an interesting speculation in the thread that the M8's offset microlenses may play a role in increasing the problem. This makes sense: the microlenses correct for vignetting by offsetting the incoming light rays, and the amount of offset increases as you get farther from the center of the chip (I base this on Leica's widely published diagram.) In effect, the microlenses are angled so they "look" toward the center of the sensor, and the outward ones "look inward" more than the inner ones.
Now, suppose that a subject includes an area bright enough that one of these little lenses flares a bit. The flare will have only slight effect on the adjacent lenses above, below, and to the inside of it -- because they're all "looking away" from it.
The lens next to it on the outside, however, will be looking toward it, and might easily pick up some of this flare. It, in turn, flares a bit... and, by the same process, passes the flare image along toward the next adjacent lens on the outside, and so on down the row. Result: a strip of horizontal "bleeding" visible in dark areas of the image.
This is all speculation, but it makes sense, no?
jaapv
RFF Sponsoring Member.
My idea is that the blur induced by the AA screen might make this effect less pronounced.jlw said:
greggebhardt
Well-known
It IS poop!
73 closely looked at images from my M8 and I see no banding. If this is the only problem being reported, I will be a happy camper.
73 closely looked at images from my M8 and I see no banding. If this is the only problem being reported, I will be a happy camper.
jaapv
RFF Sponsoring Member.
Thanks Greg. Common sense is better than pixel-peeping.
jlw
Rangefinder camera pedant
greggebhardt said:
"I've owned my new house for 73 days and it hasn't burned down yet. Why should I waste money on fire insurance?"
S
StuartR
Guest
Well, this is definitely a real problem. Hopefully they will solve it with a firmware upgrade soon. I had a chance to use the M8 today at PhotoPlus and two of the shots at higher ISO show significant banding, particularly bad at ISO 2500. At low ISO's I do not see it. The camera itself is a joy to use and the overall image quality, banding aside is excellent. The 16-18-21 is also quite an amazing lens. Unfortunately they did not have the finder there for me to check out.
Anyway, here are some of the shots I took.
ISO 2500: pronounced banding from the light sources...both in the lower light that is in line with the man's head, and in the higher lights up above. Not good...
ISO 640: Banding in the man's jacket from the reflection of the bright overhead lights in the glass of the table.
ISO 320: Very nice image quality. This was taken with the 35/1.4 ASPH.
Anyway, here are some of the shots I took.
ISO 2500: pronounced banding from the light sources...both in the lower light that is in line with the man's head, and in the higher lights up above. Not good...
ISO 640: Banding in the man's jacket from the reflection of the bright overhead lights in the glass of the table.
ISO 320: Very nice image quality. This was taken with the 35/1.4 ASPH.
S
StuartR
Guest
P.S. The firmware listed in exif is 1.06. I do not know if this is the latest.
John Camp
Well-known
I took mine out with a Noctilux, shot at 1250 and 2500 with a garage light in one corner, the rest of the photo being of the garage door, to see if I could make it band. I could. I also got some other artifacts (for one thing, a vertical band that neither began nor ended in a bright light; I also got an odd square shape in green opposite the light; it's so geometrical I think it must be some kind of internal lens reflection. Damn. I may have to figure out how to load a photo here.
John Camp
Well-known
Additional note on banding: I took my worst banding and pretty much fixed it in Photoshop in a minute or two...which would suggest that a firmware fix may be possible, if somebody should need it. I don't think I'd hardly ever need it, but some might.
After spending an hour or so outside shooting garage lights (mostly because I got the camera late in the day and didn't have much light to shoot with) I've come to these conclusions:
There is banding; you get it when you have large very-bright light sources that are extremely over-exposed in photographs that are otherwise correctly exposed. In my experience, it seems that the light has to be large within the frame -- small ultra-bright lights won't do it. And the light has to be really bright RELATIVE to the correct exposure for the rest of the photograph, so that the light is REALLY over-exposed. It doesn't make any difference whether the light is a 100 watt bulb or a 500-watt; it's the relative difference that seems to count.
You can also reduce banding if the lights themselves are slightly out of focus.
There may be some flare involvement, but I'm not sure. I got a couple of weird-looking artifacts when I was pushing the ISO.
On a practical level, it would seem to me that you'd get banding when you are trying to force night-time shots into daylight colors and illumination, but must include head-on shots of very bright lights within the frame. But if you let the night-time shots go darker than daylight-normal (so they look like they're shot at night) then the over-exposure isn't as great and the banding isn't so likely to happen. I should point out that we're really talking about shots that would be impossible with film; the film would simply be fogged in these conditions.
JC
After spending an hour or so outside shooting garage lights (mostly because I got the camera late in the day and didn't have much light to shoot with) I've come to these conclusions:
There is banding; you get it when you have large very-bright light sources that are extremely over-exposed in photographs that are otherwise correctly exposed. In my experience, it seems that the light has to be large within the frame -- small ultra-bright lights won't do it. And the light has to be really bright RELATIVE to the correct exposure for the rest of the photograph, so that the light is REALLY over-exposed. It doesn't make any difference whether the light is a 100 watt bulb or a 500-watt; it's the relative difference that seems to count.
You can also reduce banding if the lights themselves are slightly out of focus.
There may be some flare involvement, but I'm not sure. I got a couple of weird-looking artifacts when I was pushing the ISO.
On a practical level, it would seem to me that you'd get banding when you are trying to force night-time shots into daylight colors and illumination, but must include head-on shots of very bright lights within the frame. But if you let the night-time shots go darker than daylight-normal (so they look like they're shot at night) then the over-exposure isn't as great and the banding isn't so likely to happen. I should point out that we're really talking about shots that would be impossible with film; the film would simply be fogged in these conditions.
JC
stevenrk
Member
John Camp said:
John, curious if you or anyone else has found that there is banding in day time shots -- for example a landscape or street shot where the sun is directly visible or focused light is (such as sunlight through clouds, reflection from a window, etc).
Would be very interested in what anyone might find -- and here you would normaly be working in lower ISOs, although in the nightime shots the lower ISO seems to reduce but not eliminate the problem.
John Camp
Well-known
Stevenrk,
There are some other shots on this forum (the ones in Central Park) that show bright skies with a range of bright color on the ground, and no banding anywhere that anybody has reported. It seems to take a really peculiar set of circumstances. I took a couple of shots outside my local supermarket, with quite a bit of ambient light around, and no head-on bright lights, and saw no banding there, either. I really do have the feeling that this is showing up at the extremes; I pushed mine so hard that I began getting other kinds of artifacts that I think are lens related, rather than sensor related. Ultimately, I guess, my concern is diminishing...
JC
There are some other shots on this forum (the ones in Central Park) that show bright skies with a range of bright color on the ground, and no banding anywhere that anybody has reported. It seems to take a really peculiar set of circumstances. I took a couple of shots outside my local supermarket, with quite a bit of ambient light around, and no head-on bright lights, and saw no banding there, either. I really do have the feeling that this is showing up at the extremes; I pushed mine so hard that I began getting other kinds of artifacts that I think are lens related, rather than sensor related. Ultimately, I guess, my concern is diminishing...
JC
rvaubel
Well-known
John Camp said:
Not mine. Ironically, almost all the shots I have seen posted from PhotoPlus had banding. It appears that all those overhead lights provided almost perfect conditions for banding to breed. But those overhead light conditions are common for the type of photos the M8 should be best at.
Rex
haagen_dazs
Well-known
check out a ton of banding photos at
http://www.rangefinderforum.com/forums/showthread.php?t=30775
even at iso 320, it bands
http://www.rangefinderforum.com/forums/showthread.php?t=30775
even at iso 320, it bands
rvaubel
Well-known
Haagen_Daaz
I would go out tommorrow and take some regular pictures....after I sobered up.
Although a real concern, I think this banding thing is overblown. I have reviewed all my pictures from way back to Photokina and the problem has always been there but nobody noticed it. Thats not to say that Leica doesn't have a responsibility to correct the problem. Both Canon and Nikon have had similar problems and they addressed them.
Rex
I would go out tommorrow and take some regular pictures....after I sobered up.
Although a real concern, I think this banding thing is overblown. I have reviewed all my pictures from way back to Photokina and the problem has always been there but nobody noticed it. Thats not to say that Leica doesn't have a responsibility to correct the problem. Both Canon and Nikon have had similar problems and they addressed them.
Rex