Merrill blotching - help please

Sorry to ask, but are you sure that this is the well-known Foveon blotching?

The reason I am asking is that the effect is pretty local. There are parts of the picture which are of the same structure and color and equally dark or even darker but don't show any blotching. In addition, there is a part in the foreground which gets some sun and therefore has much higher luminance but still shows the same blotching as if the blotch "doesn't care" about the change of light.
I think it is. I have three Merrill cameras. SD1M, DP2M & DP1M. They all do it at different times but it is only really obvious against surfaces like grey carpet, bitumen etc. And I think although it occurs more at 200iso & more, it does happen at 100iso.

Check out the grey carpet in the attached image. The magenta is not as pronounced as in my first upload - but it is still obvious & if this was a photo I was going to use, I would have to remove it all.

The crop of the photo below was taken with the SD1M at 100iso. (I did not want to show the whole image for other reasons) But the full image had some mild editing in LR & was corrected in pt lens, but it was only a record shot that was not worth spending much time on.

So, I don't think the magenta blotchiness is unique to my DP1M, as all 3 do it. It just shows one of the subject limitations under certain light conditions that foveon sensors struggle with.

However the duds are far outweighed by the successes, & I thanks to the help here, I have a couple of ways of correcting the problem.
I think FDecker's question is rather relevant.

The magenta blotches are local to the lower right (upper right for landscape images). And they look like some streaks across the sensor, with a clear pattern.

This might be a defect on the sensor. Do you have those streaks on other images in tha same corner?

Whether Foveon blotches mainly are defects or caused by truly random patterns is not well known IMHO. To check that, you need to take several images and try to correlate the blotch patterns.
 
Or maybe the software is faulty....
Why should the software treat different parts of the sensor differently?

The calibration of the sensor might be faulty though.
 
Regarding the sensor being faulty or otherwise - no matter which Merrill camera I would have used for the first shot, or the interior shot I posted later - the result would have been the same.

All three of my Merrills were bought at different times, yet they all exhibit the same magenta blotches when the subject matter is a textured neutral grey. It gets a bit worse as the iso creeps up, but it is always there. No matter whether I process the X3F files on SPP5.5, 5.7 or the latest SPP6 version on my macbook.

If all 3 of my sensors were faulty, that would be an astonishing coincidence. I think it is the subject matter - lightly textured even grey(ish) surfaces that highlight the magenta noise/blotches that are so often referred to here & elsewhere.

On most of my shots, the blotches do not appear, or they are lost on the varied detail & colour.

I would be surprised if other Merrill owners did not find similar issues should they point their cameras to similar dull surfaces in partially shaded conditions, (or well lit interior conditions as that other shot I posted shows.)
 
The pattern looks very much like "JPG-blocks".
I'm not sure if the analysis based on a JPG is valid.
Why should there by JPEG block only in this spot of the image? And why should they be magenta only?

Moreover, JPEG blocks are 8x8 or 16x16 pixel squares; they look different.
 
Last edited:
Regarding the sensor being faulty or otherwise - no matter which Merrill camera I would have used for the first shot, or the interior shot I posted later - the result would have been the same.

All three of my Merrills were bought at different times, yet they all exhibit the same magenta blotches when the subject matter is a textured neutral grey. It gets a bit worse as the iso creeps up, but it is always there. No matter whether I process the X3F files on SPP5.5, 5.7 or the latest SPP6 version on my macbook.

If all 3 of my sensors were faulty, that would be an astonishing coincidence. I think it is the subject matter - lightly textured even grey(ish) surfaces that highlight the magenta noise/blotches that are so often referred to here & elsewhere.

On most of my shots, the blotches do not appear, or they are lost on the varied detail & colour.

I would be surprised if other Merrill owners did not find similar issues should they point their cameras to similar dull surfaces in partially shaded conditions, (or well lit interior conditions as that other shot I posted shows.)
Yes, you are right that all Foveon sensors show blotches, under some circumstances. That is a part of their properties.

The reason why I wonder if this particular sensor is faulty (or faulty calibrated) is that the blotches are not spread all over the black tarmac. It is some quite clear streaks near the corner.

If I owned that camera I would take some test images to see if those blotches turn up in the same place. If they do, I would contact Sigma.
 
Did you look at the full original size image? There are similar patterns all over...????

Lin
The pattern looks very much like "JPG-blocks".
I'm not sure if the analysis based on a JPG is valid.
Why should there by JPEG block only in this spot of the image? And why should they be magenta only?

Moreover, JPEG blocks are 8x8 or 16x16 pixel squares; they look different.

--
/Roland
X3F tools:
http://www.proxel.se/x3f.html
https://github.com/rolkar/x3f
 
Did you use the color noise reduction "slider" in SPP? If not, then try setting that all the way to the right (maximum setting). It's not really a slider. I think it has five settings, from minimum to maximum color noise reduction. When I've used that in the past it seemed to get rid of at least some of that blotching.
 
Regarding the sensor being faulty or otherwise - no matter which Merrill camera I would have used for the first shot, or the interior shot I posted later - the result would have been the same.

All three of my Merrills were bought at different times, yet they all exhibit the same magenta blotches when the subject matter is a textured neutral grey. It gets a bit worse as the iso creeps up, but it is always there. No matter whether I process the X3F files on SPP5.5, 5.7 or the latest SPP6 version on my macbook.

If all 3 of my sensors were faulty, that would be an astonishing coincidence. I think it is the subject matter - lightly textured even grey(ish) surfaces that highlight the magenta noise/blotches that are so often referred to here & elsewhere.

On most of my shots, the blotches do not appear, or they are lost on the varied detail & colour.

I would be surprised if other Merrill owners did not find similar issues should they point their cameras to similar dull surfaces in partially shaded conditions, (or well lit interior conditions as that other shot I posted shows.)
You may not be aware of this, but EVERY camera has problems with that blotching to some degree. Check out this thread:

 
Hi Roland,

The original posted is a very highly compressed jpg with literally tons of jpg compression blotching. The problem is applying any type of software which reveals any imperfections to a highly compressed image like this. Look at the orange colored posts on the building in the rear of the original image at high magnification. The orange color is splashed between the posts and smear is very evident. This is the result of super high compression. Any pattern of blotch revealed by other software is highly suspect unless it comes as the result of applying it to a tiff or at least to a very low compression jpg. I would draw zero conclusions from this myself...

Best regards,

Lin
Did you look at the full original size image?
Yes
There are similar patterns all over...????
Then we are not talking about the same thing.

--
/Roland
X3F tools:
http://www.proxel.se/x3f.html
https://github.com/rolkar/x3f
 
The original posted is a very highly compressed jpg with literally tons of jpg compression blotching.
yes
The problem is applying any type of software which reveals any imperfections to a highly compressed image like this. Look at the orange colored posts on the building in the rear of the original image at high magnification. The orange color is splashed between the posts and smear is very evident. This is the result of super high compression. Any pattern of blotch revealed by other software is highly suspect unless it comes as the result of applying it to a tiff or at least to a very low compression jpg. I would draw zero conclusions from this myself...
The magenta blotches in the tarmac are quite strong.

But ... you have a point .... maybe we should ask for the RAW file.

I assume though that the OP has seen the image before doing JPEG compression.
 
Hi Roland,

The original posted is a very highly compressed jpg with literally tons of jpg compression blotching. The problem is applying any type of software which reveals any imperfections to a highly compressed image like this.
What led you to say that, Lin?

The meta-data says "YCbCr4:4:4 (1 1)" and that means no sub-sampling at all. It also says that the software was "SIGMA Photo Pro 5.5.2.0000" and in SPP the sub-sampling method and the amount of compression can not be adjusted separately. The meta data gives the JPEG quality selected as "9".

From my point of view, therefore, "a very highly compressed jpg with literally tons of jpg compression" is both inappropriate and misleading.

Not discussed here yet, oddly enough, is that the original file was saved as a ProPhoto JPEG which might explain some of the now you see 'ums now you don't in this benighted thread.

Stuff can be saved as ProPhoto in JPEG but how well everybody's color-management copes with it could be a different matter.
 
Hi Roland,

The original posted is a very highly compressed jpg with literally tons of jpg compression blotching. The problem is applying any type of software which reveals any imperfections to a highly compressed image like this.
What led you to say that, Lin?

The meta-data says "YCbCr4:4:4 (1 1)" and that means no sub-sampling at all. It also says that the software was "SIGMA Photo Pro 5.5.2.0000" and in SPP the sub-sampling method and the amount of compression can not be adjusted separately. The meta data gives the JPEG quality selected as "9".

From my point of view, therefore, "a very highly compressed jpg with literally tons of jpg compression" is both inappropriate and misleading.

Not discussed here yet, oddly enough, is that the original file was saved as a ProPhoto JPEG which might explain some of the now you see 'ums now you don't in this benighted thread.

Stuff can be saved as ProPhoto in JPEG but how well everybody's color-management copes with it could be a different matter.
You have hawk eyes Ted!
 
Hi Ted,

Sometimes you have to look past what may seem to be obvious. Metadata may or may not actually be correct and it may or may not actually belong to the image posted.

Here is the original downloaded from the OP's post. Both the image itself and the properties scream huge compression:



Look at this at the original 400% screen capture size. Look at the jpg compression smear to the right and left of the orange post and between the posts. Look at the properties below:
Look at this at the original 400% screen capture size. Look at the jpg compression smear to the right and left of the orange post and between the posts. Look at the properties below:



 When I create a jpg from one of my DP2 Merrill captures using SPP it is rarely smaller than 10 mb. This one is only 3.73 which is highly compressed and the image itself agrees.
When I create a jpg from one of my DP2 Merrill captures using SPP it is rarely smaller than 10 mb. This one is only 3.73 which is highly compressed and the image itself agrees.



Hi Roland,

The original posted is a very highly compressed jpg with literally tons of jpg compression blotching. The problem is applying any type of software which reveals any imperfections to a highly compressed image like this.
What led you to say that, Lin?

The meta-data says "YCbCr4:4:4 (1 1)" and that means no sub-sampling at all. It also says that the software was "SIGMA Photo Pro 5.5.2.0000" and in SPP the sub-sampling method and the amount of compression can not be adjusted separately. The meta data gives the JPEG quality selected as "9".

From my point of view, therefore, "a very highly compressed jpg with literally tons of jpg compression" is both inappropriate and misleading.

Not discussed here yet, oddly enough, is that the original file was saved as a ProPhoto JPEG which might explain some of the now you see 'ums now you don't in this benighted thread.

Stuff can be saved as ProPhoto in JPEG but how well everybody's color-management copes with it could be a different matter.

--
Pedantry is not a felony.
Ted
 
Hi Ted,

Sometimes you have to look past what may seem to be obvious. Metadata may or may not actually be correct and it may or may not actually belong to the image posted.

Here is the original downloaded from the OP's post. Both the image itself and the properties scream huge compression:

Look at this at the original 400% screen capture size. Look at the jpg compression smear to the right and left of the orange post and between the posts. Look at the properties below:
Look at this at the original 400% screen capture size. Look at the jpg compression smear to the right and left of the orange post and between the posts. Look at the properties below:

When I create a jpg from one of my DP2 Merrill captures using SPP it is rarely smaller than 10 mb. This one is only 3.73 which is highly compressed and the image itself agrees.
When I create a jpg from one of my DP2 Merrill captures using SPP it is rarely smaller than 10 mb. This one is only 3.73 which is highly compressed and the image itself agrees.
Hi Roland,

The original posted is a very highly compressed jpg with literally tons of jpg compression blotching. The problem is applying any type of software which reveals any imperfections to a highly compressed image like this.
What led you to say that, Lin?

The meta-data says "YCbCr4:4:4 (1 1)" and that means no sub-sampling at all. It also says that the software was "SIGMA Photo Pro 5.5.2.0000" and in SPP the sub-sampling method and the amount of compression can not be adjusted separately. The meta data gives the JPEG quality selected as "9".

From my point of view, therefore, "a very highly compressed jpg with literally tons of jpg compression" is both inappropriate and misleading.

Not discussed here yet, oddly enough, is that the original file was saved as a ProPhoto JPEG which might explain some of the now you see 'ums now you don't in this benighted thread.

Stuff can be saved as ProPhoto in JPEG but how well everybody's color-management copes with it could be a different matter.

--
Pedantry is not a felony.
Ted
Quite disheartening, that. Hopefully the OP will be along shortly to explain how all his meta-data became wrong and why he saved in ProPhoto color space to post on the web.

That post is certainly badly smeared - until now I had never read the term "JPEG smearing" but we live and learn. Looks even worse at 800%

90a877a842dc467da16dd90fb080d231.jpg

Pardon me for stubbornly replying at the bottom ;-)

--
Pedantry is not a felony.
Ted
 
Last edited:
Hi Ted,

Low compression jpgs often "bleed" colors from adjacent borders. Here is a sample which was originally saved as the tiff then purposely saved at a very high jpg compression. Look at the subtle bleeding across borders in the jpg. Sometimes this is very bad on highly compressed Merrill images - seemingly more so than with CFA images for whatever reason.... download and enlarge considerably and you can see the smear on the highly compressed jpg across the borders versus the PNG..

7fc37af768654f5e84c968c4d27414ed.jpg

PNG - dPReview won't accept tiff....
PNG - dPReview won't accept tiff....
 
Last edited:
Hi Ted,

Low compression jpgs often "bleed" colors from adjacent borders. Here is a sample which was originally saved as the tiff then purposely saved at a very high jpg compression. Look at the subtle bleeding across borders in the jpg. Sometimes this is very bad on highly compressed Merrill images - seemingly more so than with CFA images for whatever reason.... download and enlarge considerably and you can see the smear on the highly compressed jpg across the borders versus the PNG..

7fc37af768654f5e84c968c4d27414ed.jpg

PNG - dPReview won't accept tiff....
PNG - dPReview won't accept tiff....
Now it is clear, I was able to see 1 or 2 pixel width bleeds at 800% in FastStone Viewer. I was certainly looking at the wrong post previously - now I realize you were talking about the two on the left.

--
Pedantry is not a felony.
Ted
 

Keyboard shortcuts

Back
Top