Banding and compression

[edwardkaraa]

Advertisement (gone after registration)

Just a quick question: Since I got my M240, I have been shooting uncompressed dng and have very rarely encountered high iso banding but since I switched to compressed mode, I have been noticing stronger banding, even at lower iso settings. In my understanding, the banding intensity varies with the card used, and if several shots are fired in succession, but has anyone investigated also this aspect? Is it possible that the extra cpu work to compress the files may increase the electronic interferences causing the banding?

[mjh]

Banding should be independent of either the card you use or whether lossless compression is applied.

[Advertisement]

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

[CheshireCat]

Interesting.

The lossless compression algorithm applied to DNG should not affect banding.

Electrical and thermal issues may affect banding.

[jip]

I always use compressor and never have banding only sometimes at 3200-6400

[edwardkaraa]

Thank you guys for your replies. I am aware that banding is caused by electrical interference. Shooting in continuous mode increases banding because of the high activity of the CPU. Also some cards seem to cause electrical interference, and also slower cards seem to cause less banding, probably because of the lower CPU activity. It has been shown also that some copies have more banding than others due to sealing issues. Now since compression requires more CPU power and hence a higher activity, it dies make sense in this context that it may increase the electrical interference and consequently the banding. This is at least what I noticed with my camera.

[k-hawinkler]

Thank you guys for your replies. I am aware that banding is caused by electrical interference. Shooting in continuous mode increases banding because of the high activity of the CPU. Also some cards seem to cause electrical interference, and also slower cards seem to cause less banding, probably because of the lower CPU activity. It has been shown also that some copies have more banding than others due to sealing issues. Now since compression requires more CPU power and hence a higher activity, it dies make sense in this context that it may increase the electrical interference and consequently the banding. This is at least what I noticed with my camera.

 

http://www.l-camera-forum.com/leica-forum/leica-m9-forum/256428-m9-banding-noise-issues-gone.html

[justj]

Advertisement (gone after registration)

I had these high ISO banding in the shadow as well but I always thought that's how Leica handle high ISO and didn't give it another thought.

 

Now after reading this and other links, I had another look at some of the pictures I took for the last two evenings. I find that the bending start showing up at ISO 800 and is at it's worst at ISO 1000, and then gradually getting better once past ISO 1000 then become normal again at ISO 3200.

 

I am using uncompressed DNG, I will try doing some more tests this evening with JPG and using different SD card to see if there is any improvements.

 

1000

 

2000

 

2500

 

3200

Edited July 20, 2014 by justj
photos

[CheshireCat]

Now since compression requires more CPU power and hence a higher activity, it dies make sense in this context that it may increase the electrical interference and consequently the banding.

 

That is possible in theory, but:

- The JPG lossless algorithm used is not very CPU intensive (if properly implemented).

- The compression happens after a shot has been taken, so the increased envelope temperature would affect only the next shot, and only if taken shortly thereafter.

- The compression usually reduces the amount of data written to flash by a factor of 2, therefore reducing the alleged flash controller related issues.

 

At the same time, increased environment temperature (summer in the northern hemisphere) will affect the average sensor temperature.

 

Curioser and curioser...

[FlashGordonPhotography]

Banding should be independent of either the card you use or whether lossless compression is applied.

 

Not my experience with either the M9 or type 240. I have had banding increase with different cards and different compression on both bodies.

 

Gordon

[edwardkaraa]

That is possible in theory, but:

- The JPG lossless algorithm used is not very CPU intensive (if properly implemented).

- The compression happens after a shot has been taken, so the increased envelope temperature would affect only the next shot, and only if taken shortly thereafter.

- The compression usually reduces the amount of data written to flash by a factor of 2, therefore reducing the alleged flash controller related issues.

 

At the same time, increased environment temperature (summer in the northern hemisphere) will affect the average sensor temperature.

 

Curioser and curioser...

 

Indeed all of the above are facts, so the reason of the banding increase might be somewhere else...

[rscheffler]

I had these high ISO banding in the shadow as well but I always thought that's how Leica handle high ISO and didn't give it another thought.

 

Now after reading this and other links, I had another look at some of the pictures I took for the last two evenings. I find that the bending start showing up at ISO 800 and is at it's worst at ISO 1000, and then gradually getting better once past ISO 1000 then become normal again at ISO 3200.

 

I am using uncompressed DNG, I will try doing some more tests this evening with JPG and using different SD card to see if there is any improvements.

 

This is interesting (BTW, I could only see your ISO 1000 image inline with your post) because ISO 1000 and 1250 seems to be the transition point where the M240 changes conversion gain.

 

Jim Kasson looked into this and check his links in this post:

 

http://www.l-camera-forum.com/leica-forum/leica-m-type-240/337638-correcting-green-shadows.html#post2739925

 

Edward: not sure what to say. At least my own experience with the M240 has been that I don't see banding in regular use, including some heavier treatment in post. This is in contrast to my experience with the M9 where I definitely would see banding in later images of a sequence and also lurking in the shadows of random base ISO images. With the M240 I generally try to stick to ISO 200, and if not, then will go to ISO 1000, and rarely 1600 or 3200. But I try to stick to 1000 and lower to keep the 3fps continuous advance rate (drops to 1fps at ISO 1250 and higher). I kind of assumed the banding problems of the M9 were not shared by the M240...

 

I received this week a PNY SDXC Class 10 UHS1 card and just shot a wedding on it, with images mostly at ISO 200 and 1000, but also with some at 1600 and 3200. I'll be sure to look at these for signs of banding problems. I also shot them as compressed DNG and with the camera set to continuous advance....

[edwardkaraa]

Hey Ron! Good to see you here

 

Yes I know that your M9 was very prone to banding while mine was absolutely free of banding. There definitely are body to body differences, or maybe it's due to the shooting style (I shoot very sparely) or the cards.

 

Even with the M240 I'm very lucky that my camera shows very minimal banding, much better than what I've seen with others. To be honest I didn't investigate this issue further, being too lazy and busy with work. I was kind of hoping someone has already looked into it. I will do some tests the coming weekend, but it could very well be the card at fault.