M8 streaking at Hi Iso

[eronald]

Advertisement (gone after registration)

Here is the typical setting in which I'd like to use a Leica. Don't laugh

 

As you can imagine, strong lights and specular reflections are something I do encounter in the real world, along with these nice people with pretty makeup and expensive clothes ...Well, at least they aren't camera shy - the bigger your camera the more they like you. Hmmm, maybe I shouldn't use a Leuia after all, too small This image courtesy of a 1Ds, 24mm TS/E.

 

 

Edmund

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[mitchell]

Edmund,

 

I'm not laughing. That's a very cool photo.

 

Best,

 

Mitchell

[gogopix]

Edmund

So what's the proble. Whatever banding there is innot obvious.

 

Why is it that we accept flare in many lenses, and the movie people just LOVE shots that flare for dramatic effect, and here we try to fix something natural.

 

Banding and flare tell me something about the image

That the spot lights are VERY bright!!! :-)

[jrgeoffrion]

Unless your M8 delivery was imminent it might have been an idea to at least wait until Leica comment on this 'issue'.

 

I've been through this before with the 1Ds Mark II and I have to say that it took so much energy out of me trying to find a way around losing images randomly that it I didn't enjoy photography for a whole year ( went through four different 1Ds Mark II -- all lost images). The risk to me and my clients are way too big to jeopardize their wedding images with large streaks across my raw files. At the end of the day, it's my reputation on the line (and that costs a lot more than my equipment). It's clear that the sensor is blooming and the CCD information download creates the streak. Maybe it can be solve with a firmware issue, maybe not. When Leica tells us that the issue is fixed, I'll re-place my order -- too much risk for a pro at the moment.

[Guest xxl-user]

Edmund

So what's the proble. Whatever banding there is innot obvious.

 

Why is it that we accept flare in many lenses, and the movie people just LOVE shots that flare for dramatic effect, and here we try to fix something natural.

 

Banding and flare tell me something about the image

That the spot lights are VERY bright!!! :-)

 

 

victor,

 

is a 24mm TS/E available for leica m8???

 

good CANON shot, edmund!

[eronald]

Exactly Victor. There is no banding because this is an image from my 4 years old 1Ds. And yes, the spotlights where I go are VERY BRIGHT which is why the banding on the M worries me.

 

Edmund

 

Edmund

So what's the proble. Whatever banding there is innot obvious.

 

Why is it that we accept flare in many lenses, and the movie people just LOVE shots that flare for dramatic effect, and here we try to fix something natural.

 

Banding and flare tell me something about the image

That the spot lights are VERY bright!!! :-)

[eronald]

Advertisement (gone after registration)

I sent Leica the sample file link and called them. They sounded ... embarassed.

 

Edmund

[aj37]

I think I've seen this reported before (might be a different camera) and I think it happens because the sensor reflects back some of the light onto the rear of the lens?

 

I've seen the "sensor bounce" effect on various other digital cameras, but generally just as a secondary edge around bright objects in the scene. Two things that are unusual about yours (and which I've also seen on a couple of other M8 sample photos that show this "solarized" effect) are that:

 

-- The false image is very sharply defined; sensor-bounce images I've seen tend to be somewhat blurry.

 

-- The false image is darker than the original -- a negative, you might say. The other image I saw had a spot from a bright lamp, with a dead-black "echo" of it a slight distance away.

 

These two things, plus the fact that it's more prominent at higher EI settings, make me suspect that like the "bleeding," this is another CCD charge-depletion effect of some kind, and that its displacement from the original image is a result of the charges being moved along the rows of the CCD.

 

The people at Leica who are working on the "bleeding" problem most likely would be interested in seeing this effect as well.

[wattsy]

plus the fact that it's more prominent at higher EI settings, make me suspect that like the "bleeding," this is another CCD charge-depletion effect of some kind, and that its displacement from the original image is a result of the charges being moved along the rows of the CCD.

 

My own take on it is that the 'ghosting' issue is more an optical problem rather than directly related to the CCD. The position and orientation of the ghost image is a mirror of the original bright source. This, coupled with the weird green colour, makes me think it is the bright light source relecting back off the sensor glass onto the rear lens element where it becomes part of the image. Unlike DSLRs, where the distance netween sensor and rear lens element is much greater, this may be something that the rangefinder M8 is more prone to (though my RD-1 never exhibited this behaviour and I think I have read before that one of the last Kodak DSLRs suffered from this effect on occasions).

[wattsy]

The false image is very sharply defined; sensor-bounce images I've seen tend to be somewhat blurry.

 

The false image is very much aperture dependent. It's not unlike sensor dust which is much more obvious the more the lens is stopped down. This adds to my feeling that it is an 'optical' phenomenon rather than an 'electronic' CCD issue.

[marknorton]

If the ghost images were caused by a reflection off the back of the lens, I would expect their position to change as the light source changed. In my tests, they are always at the same vertical position (with the camera held in landscape) and mirror images about a vertical axis bisecting the long dimension of the sensor.

[wattsy]

If the ghost images were caused by a reflection off the back of the lens, I would expect their position to change as the light source changed. In my tests, they are always at the same vertical position (with the camera held in landscape) and mirror images about a vertical axis bisecting the long dimension of the sensor.

 

I shot this snap earlier this morning as the sun shone through the eyelets in some blinds in my office (I was trying, unsuccessfully, to create some banding/streaking). You can clearly see the mirrored ghost of the line of eyelets in the middle of the picture. There are also ghosts of the other lines of eyelets but these are less visible. I'm not sure how this image fits in with your theory, Mark?

 

(Incidentally, what may look like clumps of image noise are large and rather shameful amounts of dust and cobwebs on the blinds.)

 

[marknorton]

Ian

 

The point where the middle column of eyelets and the reflection of it intersect is exactly half way along the image, it's as if the image had been folded in half. It's a nice example because it proves that the left side reflects into the right side and vice versa.

 

Try this example. It shows banding across exactly half the image as the right hand output channel is saturated and a ghost image which is a mirror image in the left channel. If you move the light around the frame, the ghost always appears as a mirror image the other side of an imaginary axis running (in this case) vertically through the centre of the frame. If the ghost was caused by a stray reflection, it would not behave in this consistent way.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[gogopix]

As I understand it all canon have AA filters that spread the light, yes? A good example of no banding with a sensor without AA filter may help Leica more.

 

But the bottom line is; if the banding bothers you, buy something else, or wait till it fixed.

 

Actually, I don't see why one would push someone to buy the Leica if a lot of their work will involve such extremes in lighting.

 

The tool must match the job. Seems using the M8 for these situations is a stretch. If Leica fixes, great, for many it will remain an acadmic issue, since most situations are less stressing.

 

Mark's example however is a strange one

[sean_reid]

Hi David,

 

Thanks, but I really have played a large part in the development of the M8. You can see the streaks at the same "lattitude" as the light bulb near the right edge of the frame. They certainly don't jump out at one.

 

Good idea about the article update list. Within the articles themselves, the updates has dates listed where they begin.

 

Cheers,

 

Sean

 

Ouch, important typo correction. That should have read: "Thanks, but I really have *not* played a large part in the development of the M8."

 

Oy...

 

Cheers,

 

Sean

[rvaubel]

Ian

 

 

Try this example. It shows banding across exactly half the image as the right hand output channel is saturated and a ghost image which is a mirror image in the left channel. If you move the light around the frame, the ghost always appears as a mirror image the other side of an imaginary axis running (in this case) vertically through the centre of the frame. If the ghost was caused by a stray reflection, it would not behave in this consistent way.

 

Mark

 

This is no reflection. Forget filters, lens flare, etc, etc. This is a sensor/firmware artifact.

 

The interesting thing is the banding never mirrors. Only the bloomed highlights mirror. Feed for thought

 

I really think that this is a problem correctable in firmware. This looks like a timing issue and the timing chip is programmable. I wouod find it difficult to believe that Kodak could have produced a chip with such an obvious defect especially since it is 2nd generation.

 

Hope springith eternal

 

Rex

[mike prevette]

Rex I totally agree. if the green ghosting was actually a reflection, the shape of the lenses rear element would at least distort the reflection slightly. Also it would be elongated as it approached the outer edge of the frame. I'm not seeing that happen, this seems to be a perfect digital flip of the original shape. Someone should do a test with a lens that has a convex rear element.

 

 

Or its possibly (not sure how) that the .5mil glass over the sensor surface is doing this. But that seems less likely to me. I really can't wait for Leica's response to all this.

 

_mike

[reddawn]

As I understand it all canon have AA filters that spread the light, yes? A good example of no banding with a sensor without AA filter may help Leica more.

 

But the bottom line is; if the banding bothers you, buy something else, or wait till it fixed.

 

Actually, I don't see why one would push someone to buy the Leica if a lot of their work will involve such extremes in lighting.

 

The tool must match the job. Seems using the M8 for these situations is a stretch. If Leica fixes, great, for many it will remain an acadmic issue, since most situations are less stressing.

 

huh??? How are these situations stressing? So far many of the pics here are of scenes that are fairly typical and nothing out of the ordinary - Sean's picture of a boy flipping burgers, walks in parks at night with bright halogen lights, not to mention convention halls etc etc.....

 

These are not extremes at all.

 

My film loading Leica M3, M6TTL can handle such scenes.

My CMOS sensor Canon 20D can handle such extremes. Apparently, so can Edmund Ronald's CCD using 1DS.

I love Leica Ms, I love their lenses.

 

Putting the 2 together, I want to shoot digital with my M lenses and still use a body with all the experience of shooting Leica M intact, shooting in the same style as I use my film Ms.

 

Is that too much to ask?

[reddawn]

Ouch, important typo correction. That should have read: "Thanks, but I really have *not* played a large part in the development of the M8."

 

Oy...

 

Cheers,

 

Sean

 

it's funny Sean. I read it as "not played a part" the first time round and didn't even notice it until u pointed out

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[clayh]

Some more things to ponder. It turns out if you can tolerate the blocked highlights, the streaking is less evident if you overexpose in camera and reduce exposure in Capture One than if you underexpose in camera and then compensate in Capture One. The gain accentuates the noise along the pixel path. So here are two images, one overexposed by 3EV (on Auto) and then reduced in Capture One versus an image that was underexposed by 1 EV and then gained back up. The difference is striking. Thinking about it, this is consistent with the problem not being as evident at low ISO's. BTW, these both had the camera set to ISO1250.

 

All were M8 35/1.4 Lux asph at f/4

 

Also interesting is the color shift, which can only be attributed to the profile behavior. These images were both set to identical white points in capture one determined by exposure through an expodisc - in this case 2350K. N.B. I used the capture one profile for the generic R8 camera, as it shows less of the magenta cast under artificial light.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!