green blob summary

[jrc]

Advertisement (gone after registration)

Last night, after I found a green blob on a number of shots when I was trying to force banding, there were several suggestions about what it was. I can say:

 

It's not filter related. I had a filter on last night when I first saw it, but tonight, shooting the same light from the same place, I got identical green blobs with or without the filter, using all the same settings.

 

The green blob is a flipped image of the high light source, symmetrical on the opposite side of the centerline of the frame (as has been suggested by others.) In other words, if you put the high light close to the edge of the frame, the green blob will appear close to the opposite edge; if you put the high light close to the centerline, the blob will be identically close to the center line, but on the other side.

 

At 320, I don't see it; at 1250, I do. I was shooting a 50mm f1 wide open, and at 320 the exposure would be longer that at 1250. If it was a simple reflection off the front of the lens and back to the sensor, to register on the sensor, why wouldn't it be stronger at 320? So I don't think it's that. After all is said and done, it seems to me to be related to the ISO setting.

 

Also, if the high light is sharply focused, the counter-image (blob) is sharply focused; if the high light is thrown out of focus, the blob will be identically out of focus. The blob appears whatever the focus.

 

Also, I just put a 50-year-old 5cm Summicron on it, and still got the blobs, although I felt the blobs had a more film-like character. 8-)

 

JC

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[ho_co]

John--

Glad you posted this. I think it's becoming clear that the problem is sensor related; perhaps it's related to the banding/streaking phenomenon. Should be easier for Leica to find and fix if they forget about IR sensitivity and concentrate on the sensor.

 

--HC

 

for cross-reference, green blob is visible at

http://www.leica-camera-user.com/digital-forum/8642-m8-streaking-hi-iso-2.html#post84733

and

http://www.leica-camera-user.com/digital-forum/8672-banding-some-more-tests-8.html#post85798

[blado]

That blob stuff is way weird. That, plus the "banding being exactly 400 pixels of an 800 pixel wide image" theory makes for some strange stuff. They don't seem to be optical problems, ie not light bouncing around inside; plus the blobs and streaks aren't distorted. It's like channel distortion ro something...

[Bob Ross]

The green blob is a flipped image of the high light source, symmetrical on the opposite side of the centerline of the frame (as has been suggested by others.) In other words, if you put the high light close to the edge of the frame, the green blob will appear close to the opposite edge; if you put the high light close to the centerline, the blob will be identically close to the center line, but on the other side.

 

Also, I just put a 50-year-old 5cm Summicron on it, and still got the blobs, although I felt the blobs had a more film-like character. 8-)

 

JC

I think the operative term here is "flipped image" and that implies "reflection" or "mirrored imaged" and "oposite side of the centerline" implies that the reflection is off of a curved surface (rear element?).

The sensor is a very reflective surface (and flat), so light does reflect back at the lens' rear element and mount. Each sensor is different and Kodak's SLRn series for example would produce a red dot with some lenses. The size and possible shape (in cheaper lenses) will very according to aperture and in SLR lenses, if there are any flat rear surfaces on any element and what coatings, if any, are on the rear of the lens elements.

Your example was from extreme conditions compared to those that I saw from DSLRs, where it could appear in normal shooting conditions. I suspect that low end lenses (SLR) that suffer from flare might have this digital artifact minimized by the flare:rolleyes: The newer digitalized lenses have coatings on the rear side of the lens elements to reduce this affect.

Bob

[marknorton]

As I recall, it was the Luminous Landscape review which first reported the existence of blobs. Mine seemed to be a yellow for a while but have now turned regulation green. As I said elsewhere, I think it might be cross-talk between the left and right channels in the analog to digital conversion - as the data is being read out of both channels, the high light level in one channel "bleeds" across to the other channel.

 

Take a look at this diagram from the Kodak Sensor specification. The 3 stage Source Follower is effectively a current amplifier which takes the minute charge coming off the sensor and amplifies it to allow external circuitry to sense it. Those lines marked Vss and Vdd are the power supply lines to this output amplifier. The second diagram suggests the data is read from the centre of the sensor out towards the edge which explains why the green blobs are mirror images of the opposite side. The sensor pin out (if you can read it) also shows that the Vss and Vdd lines for each side are brought out to separate pins. Vss is effectively "ground" and Vdd is nominally +15v.

 

I believe the green blobs are caused by some off sensor interaction through common circuitry. If these separate Vss and Vdd pins are just connected together, with inadequate supply regulation or supply decoupling at the signal frequencies involved, activity on one channel could feed across to the other and influence the other.

 

Ideally, these Vdd and Vss lines (plus any supply lines feeding follow-on analog circuitry) for each channel should be fed from separate power supply line with high frequency decoupling - connection of ceramic capacitors between the supply lines directly to a low impedance ground plane - which have the effect of "stiffening" the supply rails.

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!

[mike prevette]

(this was written before i saw what Mark posted above, I totally agree with him)

 

THIS IS NOT AN OPTICAL PROBLEM. If you guys would reset your thinking a bit and remember your first few years in a wet darkroom, you probably spent a lot of time learning the ropes and trying to understand the magic that was at play in the developer tray. Well you need to do the same thing here. It might look like the camera you've used for fifty years, but it is not. To understand this problem you really need to do some hardcore research into CCD design, and until you do that you will never understand what people are telling you regarding whats happening.

 

CCD's are analog devices. (uber over simplified) Light charges little capacitor buckets (think batteries) and then that voltage is read and given a value depending on how much was there. Thats a pixel, and how the camera figures out how bright it is. The chip needs a lot of amplification in dark places. Think of it like a stereo, when its too quiet you turn it up, and if you turn it up to much it starts distorting the signal. Same thing is happening in your ccd before the analog signal gets turned into a digital one.

 

depending on the circuit design, and the amplifacation applied intefrerence can "jump" between lines of pixels (hence the banding, or smearing) NOW if the control wires (they aren't really wires) that carry the signal to closely to the signal wires OR the pixels of the active chip, interferience will happen. GREEN DOT.

 

Now knowing all this, and since this is happening WAY before the signal becomes digital. that means the chip has no freaking idea that the highlight is 255,255,255 all it know is its 4.5v vs the surounding 2.5v (these voltages are example only) (The chip can hold voltages much higher than will be converted to 255,255,255). This spike in voltage causes interference. GREEN DOT

 

Why is it green you ask? it's green because there are proportionally more green pixels than the others. Ok so they aren't really green pixels they are of course BW, but the camera or raw processing program EXPECTS more green pixels than the others and so it turns them green to us.

 

There are a million reasons to prove it's not optical, and a million reasons why it could be digital.

 

_mike

[marknorton]

Advertisement (gone after registration)

Mike, why do you feel the need to be condescending towards people on this forum who may not have a technical background and why do you adopt this hectoring UPPER CASE shouting writing style?

 

I tried in my post to explain how this might be happening in relatively non-technical terms, quietly. I don't know whether this is the correct explanation but it matches my observations of how my M8 is operating.

[jrc]

THIS IS NOT AN OPTICAL PROBLEM.

 

 

I think we all agree on that. I was simply eliminating some optically-related suggestions made on the forum that the blobs might be related to specific lenses or filters.

 

JC

[mike prevette]

Your right Mark, I'm often proven to be a complete bewb. I really do appreciate your excellent and very in depth yet not overtly technical explanation. I really wish I could be that eloquent in written form.

 

I'm on the road right now and reading this via my cellphone connection so I end up writing things very quickly and passionately. Add to that my disappointment that Leica is rolling out a product that I have waited so long for, and it's going to be held up by a poor circuit design. It makes me a bit of an a$$. I'm sorry to offend, but I feel strongly that if your passionate about something you should also understand it, and so many people don't.

 

Recently a lot of film photographers have converted to digital, and they spend a lot of time on soapboxes complaining about things they refuse to understand. Yet at the same time they will argue the merits of Rodinol vs Dektol, Erwin Puts lens reviews, and tons of other minutia. It makes no sense to me. In the end this is all very unimportant in the world of photography and very rarely effects how i take pictures. i just think understanding how my digital camera works is just as important as understanding how my film camera works. I spent 20 years learning film stuff, and I'm sure i will be learning things about digital for a very long time.

 

Also JC, My comment wasn't aimed at you, I was actually trying to reinforce your comment with some reasons why it could be happening. I'm sorry I didn't make that clear.

 

 

_mike

[marknorton]

That's OK, I think we're all getting excited trying to understand what's going on!

[lct]

Thanks for those explanations, Mike and Mark, very interesting indeed.

[wparsonsgisnet]

I don't know about you guys, I don't plan to photograph any blobs.

[jlancasterd]

I reckon we're all seeing green blobs because the guy who wrote 'The Quatermass Experiment' and 'Quatermass and the Pit' just died.

 

They're here for the funeral... :D

[frc]

With my experience in audio engineering as a background ( signal overload is a common problem here ), I can only say that Mike's explanation makes sense.

Bluntly put or not ;-)

[woodyspedden]

All well and good that we have users who can understand and precisely communicate what might be going on. May be correct or maybe not!

 

As a (patient) user what i want is for Leica and Kodak to understand the issues and fix them. Only then will we users have a camera that does it all.

 

My concern with the overload theory is that the DMR also has no AA filter and can accumulate over charge just like the M8. However no one to my knowledge has ever talked about banding or magenta casts. I wonder what is different with the M8 that creates this problem. Mark you have the Kodak data sheet for the M8 sensor. Do you also have one for the DMR and does the DMR read out the pixel information different from the M8?

 

Fascinating stuff really

 

Woody

[marknorton]

Woody, it certainly does.

 

The M8 reads out the image in about 320mS, which sets the basic 2 frames a second shooting speed. The key parameters are the 24Mhz horizontal clock and the dual outputs.

 

The DMR sensor takes very nearly 4 times as long to readout an image. 1.25 seconds. The reason is that there is only a single output and the clock only runs at 10 Mhz.

 

Single output. No risk of cross talk between the outputs and (if I'm correct), green blobs. Slower clock. Less critical clocking, less need to try to overlap activities, less chance of photo-electrons ending up where they are not wanted.

 

The banding and green blob problems then are likely a consequence of the higher performing (faster readout) M8 sensor which requires fine tuning of the firmware (and possibly the hardware).