Thoughts on the vertical banding in some SD1M images

[maple]

To me, SD1M is very much the same old story all over again - a camera of both exceptional merits and fatal faults, like its elder siblings. While the older SD's are good at resolving details, the latest SD is phenomenal like no Bayer except the most capable FF can. It has even improved on the previous models where they failed most. In spite of the much greater pixel density of the sensor, colour blotches in shadows is much less of a problem, for example, and low light images are cleaner, too, by one stop or so. On the other hand, the old Foveon look that we like so much is largely gone. Colours do not have the same density and appeal, and its images often look “dry”.

But worst of all though is the vertical bandings in the sky.

After many attempts at “restoring” the Foveon look and supressing the vertical bandings in pp, it finally occurred to me that it is actually these vertical “cracks” that make the image look “dry” and ruin the Foveon look. They are the culprits. I reckon that if there are bandings in the sky of a picture, they must also be elsewhere all over that picture. They are just more visible against smooth background like skies. Where there are great amount of details, the bandings are pretty much camouflaged and can even be falsely perceived as details. They are, however, not real, and images with a lot of these false details won’t look right, as they don’t match our memory that well.

There are remedies, the simplest being leaving NR to its default 0 value. As far as you do not apply excessive processing later on, it’ll take care of much of the vertical banding problem, at very little cost to fine details. So my suggestion is not to lower the default NR setting in SPP, unless you really need that last bit of details, and do not mind the trouble of further processing the image discriminatively, in another program, of course.

--
Maple

 

[HBowman]

Banding in the sky ... I try hard to find it with my SD1 But you can search for me, I'm kinda lazy ... If you want to search harder, go in my gallery and clic on original, you will have maybe better chances !





--
http://www.hulyssbowman.com

 

[stevechong]

--

• steve -

http://www.stevechong.com
http://www.pbase.com/sigmadslr/steve_chong

 

[HBowman]

NR ?
--

• steve -

http://www.stevechong.com
http://www.pbase.com/sigmadslr/steve_chong

NR or not, I always process my SD1 files as I process my DP2 and SD15 files. Some people, at one time, came here saying "how to process SD1 files correctly" but, as far as I know, I process SD1 files correctly but not their way

SPP do the job. Same thing, original in my gallery.









--
http://www.hulyssbowman.com

 

[michaeli]

Does SD1 have this problems either?
--
My gallery: http://www.flickr.com/photos/lixun/ ;
http://michaeli.tuhigh.com

 

[Hng]

Then when you face high dynamic range scenes, portraits with mixed lighting, landscape with people, ... I am sure SPP only can not deliver potential capabilities of SD1 camera. Going back and forth with many methods in SPP and PS, try many plugins and exploring some hidden features of PS, suffering so much up and down feelings, now I feel a bit confident with ways to post process by various types of photography with my SD1M.

 

[DaSonyGuy]

To me, SD1M is very much the same old story all over again - a camera of both exceptional merits and fatal faults, like its elder siblings. While the older SD's are good at resolving details, the latest SD is phenomenal like no Bayer except the most capable FF can. It has even improved on the previous models where they failed most. In spite of the much greater pixel density of the sensor, colour blotches in shadows is much less of a problem, for example, and low light images are cleaner, too, by one stop or so. On the other hand...

...Highlight handling is absolutely terrible!

Add that to the seemingly unsolvable banding problem, no video option and paltry battery life and its clear that only someone with more money than sense would buy one!
The D800 is looking better and better!!

 

[Joerg V]

[...] ...Highlight handling is absolutely terrible!

[...] seemingly unsolvable banding problem, no video option and paltry battery life and its clear that only someone with more money than sense would buy one!
The D800 is looking better and better!!

Well, you have a point. Unfortunately a D800 + an adequate lens costs 3.5k€ - and then you need a tripod, filter(s), backback (maybe even a new computer)... for an overall cost of > 4k€. A DP2m which seems to give a low-iso image quality that is at least not that far behind will cost a grand (plus 100€ for filters).

That is causing me a major headache since months: I am unsure if I want what I can afford (DP2m or DP1m) but I am pretty sure that I cannot afford what I want (D800). :|

Joerg

 

[maple]

Obviously you do things right.
--
Maple

 

[maple]

--
Maple

 

[Mostly Lurking]

but from your recent posts I am of the opinion that you do not have an SD-1 or SD1M. That means that your post is based upon the on-line shots of others; always far less than perfect representations of photos. Still, your post comes across as that of an expert rather than opinions based on the work of others seen in an 'imperfect light'.

You may be completely correct in your assessment for all I know. However, not everyone who owns an SD-1 or SD1M is making the same complaints.

To be sure, my skills are probably not up to either camera and the cameras are probably not up to my lack of skills . No probably about it, my bank account is not up to either of them!

Regards!
--
William Wilgus

 

[mike earussi]

but from your recent posts I am of the opinion that you do not have an SD-1 or SD1M. That means that your post is based upon the on-line shots of others; always far less than perfect representations of photos. Still, your post comes across as that of an expert rather than opinions based on the work of others seen in an 'imperfect light'.

You may be completely correct in your assessment for all I know. However, not everyone who owns an SD-1 or SD1M is making the same complaints.

To be sure, my skills are probably not up to either camera and the cameras are probably not up to my lack of skills . No probably about it, my bank account is not up to either of them!

Regards!
--
William Wilgus

There are plenty of raw files available online that anyone can process for themselves to see exactly how the SD1 looks. I've processed over 50 raw files myself even though I don't own one (and don't ever plan on owning one given it's buffer limitation) and have come to the conclusion that if you shoot at iso 100 only banding is minimal. And looking at the DP2M jpegs from Sigma's site what banding that did exist now seems to be gone.

Sigma does seem to be slowly improving the output of their 15mp chip and so I am looking foward to the next incarnation of the SD line, hopefully with a much larger buffer, faster processing time, and live view. With these three improvments coupled with the continued improvement in IQ from the chip the SD2/SD16(?) will finally be a camera worth buying.

 

[JohnLindroth]

--
http://www.johnlindroth.com/
[email protected]

My future starts when I wake up every morning ...
Every day I find something creative to do with my life.
--Miles Davis

 

[JohnLindroth]

Maple,

The vertical banding is one of the things that bothered me the most with the SD1 images. Since I'm testing out an original SD1, I've been looking for it, and trying to decide how to handle it. By setting NR to it's minimum (0 or -2, depending on your reference point) in SPP, it's more visible. The internal sharpening in SPP (why you want it set to -2) makes it more visible. The default setting in SPP (both chroma and lumin in the middle) seems to eliminate most of it, even moving chroma to the lowest is reasonable. I do see a slight softening of the image with this NR setting, but following up with a sharpening in photoshop of 0.1 or 0.2 pixels at 200-300% brings back the sharpness quite well.

I have only one photo decoded from RAW in RawDeveloper. It was the portrait, shot at 100ISO and I don't see the banding in it at all, in either program, and I've got all NR set to minimum in SPP. I would like to know if RD could provide a better decoding related to the banding.

I don't have time to put together some of the tests I've done, but I'm planning to do that when I'm done with the testing.

To be honest, I see some vertical banding noise in Canon 5DMkII images under some conditions, but it's better controlled internally in the camera. I'm guessing that with making smaller pixel sensors, this might be part of the mechanical issues faced, and this would be Sigma's first encounter with it in processing the images.

-John

--
http://www.johnlindroth.com/
[email protected]

My future starts when I wake up every morning ...
Every day I find something creative to do with my life.
--Miles Davis

 

[rainbyrd]

take rawdeveloper and your banding is history.spp is a pain to use.rawdeveloper takes 5 seconds to develop a picture, while spp takes minimum 25 (!)

 

[Kendall Helmstetter Gelner]

Obviously you do things right.

So since you can take pictures without banding, how is some banding in some images a "fatal flaw"?

The fact is that almost all of the banding examples I have seen are in images where people pushed settings a huge amount to make the banding show, if there's any present. If you are shooting for a neutral image without much processing, it's not really an issue in real images...

And as others have noted it's not even an issue with the camera, it's an artifact of SPP.

--
---> Kendall
http://www.flickr.com/photos/kigiphoto/
http://www.pbase.com/kgelner
http://www.pbase.com/sigmadslr/user_home

 

[Roland Karlsson]

Its all quite simple

If you cannot, via any manipulation of contrast, can get a RAW image to contain the banding - then its SPP that makes the banding.

If you can get the banding by e.g. increasing the contrast in a RAW image, then RAW therapy is "hiding" the banding.

BTW - no mater what the result of the analysing of the RAW image gives regarding banding - I am quite sure that this is not the cause for the loss of Foveon effect. I rather think it is lack of resolution - either by lack of good enough lens or some smoothing noise reduction or both.

--
Roland

X3F tools:
http://www.proxel.se/x3f.html
https://github.com/rolkar/x3f

 

[maple]

Thank you, John,

By NR “0”, I meant the default setting of NR in the neutral position, both luminous and chromatic, while “-2” means all the way to the left. Basically we come to the same finding: leaving the luminous and chromatic NR at its default setting, i.e. unchanged, seems to be a good balance between keeping the vertical banding away and retaining fine details.

BTW, a few more words that I did not start this post for no reason, as some seem to have found it. Quite a few have suggested both luminous and chromatic NR should be set to minimum, i.e. to the left most position, for ISO 200 images. I had a different opinion and thought it may be a good thing to share it.

--
Maple

 

[Basalite]

To me, SD1M is very much the same old story all over again - a camera of both exceptional merits and fatal faults, like its elder siblings. While the older SD's are good at resolving details, the latest SD is phenomenal like no Bayer except the most capable FF can. It has even improved on the previous models where they failed most. In spite of the much greater pixel density of the sensor, colour blotches in shadows is much less of a problem, for example, and low light images are cleaner, too, by one stop or so. On the other hand, the old Foveon look that we like so much is largely gone. Colours do not have the same density and appeal, and its images often look “dry”.

But worst of all though is the vertical bandings in the sky.
Maple

I don't know why you are not seeing the same colors as I sure do see the same exact colors as I do with my DP1.

 

[maple]

So since you can take pictures without banding, how is some banding in some images a "fatal flaw"?

My use of “Fatal” here does not refer to the banding issue in particular, but to all the flaws together in general. The frustratingly slow write time compounded by small buffer, the lack of live view and video, and the poor high ISO images quality, are all jointly and severally enough to push many potential buyers away to other brands. Not that it will cause the camera to die, but it’s rather fatal to its success as a product in a competitive market. I did not elaborate on that as I was only discussing about the banding.

Personally, the vertical banding is a very serious, i.e. fatal, flaw – I’m very fussy about IQ, and that’s why I picked SD1M instead of, say, D800. Now this banding thing in some images. Note qualifying word “some”. How infrequent is “some” supposed to be so that it’s not fair to describe the flaw fatal?

The fact is that almost all of the banding examples I have seen are in images where people pushed settings a huge amount to make the banding show, if there's any present. If you are shooting for a neutral image without much processing, it's not really an issue in real images...

Now you are saying banding is not even a flaw of the camera, (let alone being a fatal one) but user's misuse of it. That's an interesting perspective. One thing I like Foveon is that its files are tolerant to “pushing” (though not with shadow lifting). Now with this latest model, we are told to be gentle, and stick to the “Neutral” colour mode only! Putting aside the question why all those other colour modes, is that how we keep the “some” to an negligible percentage? Seriously, are you happy with how “Neutral” mode renders the colours every time? Personally, I find that rather limiting, if not fatally so.

And as others have noted it's not even an issue with the camera, it's an artifact of SPP.

I haven’t done any serious test, but I have not seen any vertical bandings in SD14 images processed in SPP 5 with my regular and usually quite generous dose of processing (pushing). Accusing SPP doesn't stack up well.

--
Maple