The strangest artifacts i´v seen!

[steinzeug]

Advertisement (gone after registration)

Dear Rich,

 

Very grateful for your answers, however

as you can see the problem is "in" the camera.

 

Mike

Hello guest!
Please register or sign in to view the hidden content.

Edited January 18, 2010 by steinzeug

[Advertisement]

Hallo,

Look here: Leica M

[Nicoleica]

I've just tried looking at these DNGs in aperture. Again we have different results. Aperture seems to be doing a slightly better job than some others here, but it's still not good.

 

I have a question though. What lens was in use here, there's nothing in the EXIF. If the lens is coded, have you tried the same with detection on?

 

Here's how aperture interprets the files.

Hello guest!
Please register or sign in to view the hidden content.

Hello guest!
Please register or sign in to view the hidden content.

[steinzeug]

I've just tried looking at these DNGs in aperture. Again we have different results. Aperture seems to be doing a slightly better job than some others here, but it's still not good.

 

I have a question though. What lens was in use here, there's nothing in the EXIF. If the lens is coded, have you tried the same with detection on?

 

Here's how aperture interprets the files.

 

Many thanks for your kind comments.

 

I am using a non-coded summicron-35 (1973)

I just did a test with "detection on"

but got exactly thesame maze-fringe-effect.

 

As mentioned earlier, the defect is seen on the camera screen before i even export the image. So this must mean the problem is "inside" of the camera(?)

 

Mike

[Nicoleica]

Hi,

 

Yes, it does appear to be camera related. Just out of interest, does the same thing happen with other lenses, or only with this one?

 

However, I agree with the others, and suggest that you send copies of your DNG files to Leica with a covering letter. They may be able to help, or at least tell you what the cause is.

 

Sorry that we were unable to help much.

Hello guest!
Please register or sign in to view the hidden content.

[steinzeug]

Hi,

 

Yes, it does appear to be camera related. Just out of interest, does the same thing happen with other lenses, or only with this one?

 

However, I agree with the others, and suggest that you send copies of your DNG files to Leica with a covering letter. They may be able to help, or at least tell you what the cause is.

 

Sorry that we were unable to help much.

Hello guest!
Please register or sign in to view the hidden content.

 

Hi,

 

unfortunately I only have htis lens at the moment, but i am begining to feel more and more that it is a lens related issue (even thought the defects look "digital")

 

thanks for your input!

 

mike

[hoppyman]

Dear Rich,

 

Very grateful for your answers, however

as you can see the problem is "in" the camera.

 

Mike

Mike what you are seeing on the camera LCD is very likely not the same thing that you are seeing in those bizarre processed examples. I suspect that you have more than one thing going on there, including sensor characteristics and extreme cases as well as the Raw processing artefacts. I do see the fringing to some extent in your DNG opened in ACR and LR current versions. Back on the camera LCD, try turning off the clipping warnings in your camera menu and see if the effect is still visible in the previews. Remember that the camera generated previews are not accurate but previews of jpg processed images (and may include clipping warnings too). I would also add that viewing the image previews on your computer screen at anything other than 100% means the preview is inaccurate at individual pixel level. I don't know Aperture but try zooming or whatever to 100% and look again as well.

Edited January 18, 2010 by hoppyman

[steinzeug]

Advertisement (gone after registration)

Mike what you are seeing on the camera LCD is very likely not the same thing that you are seeing in those bizarre processed examples. I suspect that you have more than one thing going on there, including sensor characteristics and extreme cases as well as the Raw processing artefacts. I do see the fringing to some extent in your DNG opened in ACR and LR current versions. Back on the camera LCD, try turning off the clipping warnings in your camera menu and see if the effect is still visible in the previews. Remember that the camera generated previews are not accurate but previews of jpg processed images (and may include clipping warnings too). I would also add that viewing the image previews on your computer screen at anything other than 100% means the preview is inaccurate at individual pixel level. I don't know Aperture but try zooming or whatever to 100% and look again as well.

 

 

Thanks Geoff,

 

clipping was in fact turned off, and the maze-type-artifacts were seen clealy both on the LCD, and on the computer. Now wondering if it might be a fault in the lens (despite the "digital" feel of the artifacts)???

 

best,

 

Mike

[Advertisement]

Hallo,

Look here: Leica M

[jaapv]

It is quite impossible for a lens to produce this kind of artefact.

[mabelsound]

FWIW, I tried converting it with dcraw, and that showed the same artifacts.

[steinzeug]

FWIW, I tried converting it with dcraw, and that showed the same artifacts.

 

 

Yes, the artifacts are seen in the jpeg on the LCD and the raw DNG conversions.

 

its a mystery!!

[jaapv]

If I develop it optimally in C1-5.0, the colour artefacts disappear completely, what remains is the jagged edge. Strangely the slider it reacts most strongly to is : long exposure noise. That the camera does the same is not a surprise - it has to produce a JPG from the same data.

Hello guest!
Please register or sign in to view the hidden content.

[dNorm]

This is a software issue, the artifacts from which can be brought about by certain lenses. I had this very issue with my Olympus E-P1 when using a Zeiss 21/4.5 C Biogon and converting in Lightroom. The only thing was this maze pattern showed up on one side and in the corners. I sent a copy of the RAW file to Eric Chan (a developer for Lightroom) and the release version corrected this problem. Here is a link to the dialog. Cheers. -Norm

 

Adobe Forums: Odd Maze Pattern with Olympus E-P1 &...

 

 

It is quite impossible for a lens to produce this kind of artefact.

Edited January 19, 2010 by dNorm

[adan]

Thinking about the known facts - occurs in the camera, occurs with a pre-ASPH 35 'cron (which worked fine on my M8s), can't seem to be fixed with any of several raw converters (by forum members), is limited to hardish edges on blown highlights....

 

my sneaking suspicion is that this is a problem with the analog/digital converter - the chip that converts the voltages delivered from the sensor pixels into the numerical 1's and 0's for each pixel in the digital file (and also compresses the data nonlinerally (more compression for highlights via root-2 function) to 8-bit).

 

Might be the silicon itself failing - might be that the firmware for the AD process got wracked up during a firmware upgrade.

[hoppyman]

To expand on my earlier contribution, the camera LCD preview is not accurate, it's a low resolution thumbnail of a jpg conversion from the dng. A very useful guide of course but ought not to be taken as an accurate rendering as far as showing or not showing any artefacts. In any event, very different to artefacts that may be visible in post processing and at different magnifications on a computer screen. If you greatly magnify or downsample the pixels in viewing you have introduced more variables and cannot reliably make any judgements (on this issue).

Pixel peeking is an evil temptress

Hello guest!
Please register or sign in to view the hidden content.

I don't know if particular lens optical characteristics may make some artefacts more likely to appear, but it does not necessarily follow that the lens is the cause. I suspect that it is the least important factor here.

For comparison these are two 100% crops from M9 files (processed in Adobe Camera Raw 5.6.

 

Hello guest!
Please register or sign in to view the hidden content.

 

Hello guest!
Please register or sign in to view the hidden content.

[ho_co]

This is a software issue, the artifacts from which can be brought about by certain lenses. ...

 

Interesting link, Norm. Could you post a copy of what the image in question looked like, just for comparison?

 

You say yours was more evident at edges and corners; but Mike's is evident at the juncture of bright and dark areas.

 

It certainly looks like an error introduced by the RAW processor, as others said above. But Mike's situation appears even when reviewing on the camera's LCD. Was that the case with the files you sent Adobe?

 

Andy's suggestion makes most sense to me at the moment.

Edited January 19, 2010 by ho_co

[Advertisement]

Hallo,

Look here: Leica M

[mabelsound]

But the image on the camera's LCD is, I'd imagine, also a RAW conversion of a sort.

[bsmithy]

FWIW, from everything I've read, the M8 performs demosaicing in-camera. From the metadata for an M8 DNG:

 

<crs:CameraProfile>Adobe Standard</crs:CameraProfile>

<crs:CameraProfileDigest>DC6355595165B110B7CBDA45B1C7D6C9</crs:CameraProfileDigest>

<crs:HasSettings>True</crs:HasSettings>

<crs:HasCrop>False</crs:HasCrop>

<crs:AlreadyApplied>True</crs:AlreadyApplied>

 

The "Camera Profile" is "Adobe Standard", which *I think* indicates that the "raw" image data contained in the DNG has already been converted from the original Bayer image to Adobe RGB. This in contrast to a straight-off-the-sensor raw image (usually single-channel Bayer-pattern color filter array), e.g. Canon's CR2 format. In the case of my Canon 1DMk3, the Camera Profile is specified as "ACR 4.4"; my take on this is that Canon is relying on Adobe Camera Raw to do the de-mosaicing. I would imagine that the ACR 4.4 spec includes information on the color filter array applied to the camera's sensor as well as the desired demosaicing algorithm (or something along those lines).

 

At any rate, the camera must de-mosaic the original image in order to show you the preview on the LCD, and in your case the artifact appears in the preview.

 

If this *is* a de-mosaicing artifact, it's pretty wild. Usually those are limited to 1 or 2 pixels. My guess is that it's a defect in your camera's hardware/electronics.

 

Cheers,

Brad

[ho_co]

But the image on the camera's LCD is, I'd imagine, also a RAW conversion of a sort.

 

Good point.

 

I guess if you get down to it, everything is a RAW conversion since even the DNG is a conversion of the data impinging on the sensor.

Hello guest!
Please register or sign in to view the hidden content.

 

Somewhere, the "smooth" real world data are becoming jaggies. In this case, it looks as if it's in the camera rather than in later processing. That's not proven, as you suggest, since something has to process the data for us to see it; but seeing the defect on the LCD puts the miscalculation pretty early in the chain.

 

That's why Andy's A/D converter suggestion makes sense to me: If the initial conversion to digital is bad, the input to any later converter will also be bad.

[marioalessi]

THe fact that you see it in the camera, the camera is in fact producing it.

 

It could be one of the following combination:

 

Software in the camera

Software in the camera with that specific lens

Software in the camera with that specific lens with Electronics of the Camera

 

Software in the camera with Electronics in the Camera

That specific lens with Electronics in the Camera

 

Software in the camera with anyones lens of that model

Software in the camera with anyones lens of that model with Electronics of the Camera

 

Anyones lens of that model with Electronics in the Camera

 

Etc....

 

You essentially have different combinations of your camera (Which includes firmware and electronic chips inside) and the Lens

 

Without having other lens, or someone else with an M8 with your lens or a the same lens,

 

You will really have to send it in to Leica i would think.

[steinzeug]

I am very grateful to all of you for your thoughtful an generous replies and theories. I have written to Leica and hope they might have a solution. However what is clear to me (due to the issue already appearing in the camera) is that the defects are occuring either in the cameras own software/chip, or with that particular lens.

 

What surprises me is that no one has had this kind of experience themselves (which might strengthen the case that it is the lens over reacting to the contrast in light?)

 

Mike

Edited January 19, 2010 by steinzeug