Anyone come across any banding issues yet?

[David Hull]

I looked at that one the other day and that streak appears to be in the actual scene that was photographed - not sure it is from the camera or the processing. It is not parallel to the edges of the frame which I would expect if it were a sensor issue.

The banding in the example above is in the vertical (Up/down) and looks like purple and green alternating bands of different widths. All seem parallel to the edges. Is that what you are seeing?

He used the camera verticaly, so the banding is horizontal as common.

I'm aware of that but was describing the banding in the example above to help understand if David was seeing what I was seeing as he was referring to "That Streak" which seemed to not be related to the banding in the example

Yes, I am seeing it, now that you pointed it out which is a bit telling in and of itself. Not only that but the photo is literally hammered to get it and looks fine when exposed rather conventionally. So, yes, we are seeing the same thing.

I am still trying to figure out whether this is a real issue or just the usual DPR "the sky is falling" stuff that seems to show up with each new camera. I started with a 20D, went to a 5DII then a 5DIII then added an M5 and got good photos with all of them.

--
Flickr: http://www.flickr.com/photos/drhull
SmugMug: http://davidhull.smugmug.com/

 

[mmeerdam]

It looks like this is happening in grey areas with almost no color in it?

The camera is jumping back and forth between two distinct mean black levels, and the converters are ignoring it and treating it like signal. This is independent of subject matter. It is just more obvious when it drives a neutral color towards one or two different colors. If the subject had a strong orange color, it would just be two subtly different oranges; not two very different colors.

I'd like to learn some more: Is this something other cams also do?

Only in the new Canons, AFAIK. Most banding on other cameras has more random black offsets for lines, rather than two distinct groups.

meaning: Does the 5d4 do this also and do adobe and capture one have a way to suppress the results when native support for the cr3 's is provided?

This is the same as the 5D4; it just seems more common, so far. Did these converters correct it for the 5D4? If not, then perhaps they never will for the R.

ACR was just updated. Banding is the same as with the DNG's

just shot the same scene with my 5d4 and I can push it 5 stops with any problems.

EOS R starts showing banding from 2 - 2.5 stops exposure push...

Dual pixel raw is a bit better, but the banding is still there

 

[Mako2011]

I am still trying to figure out whether this is a real issue or just the usual DPR "the sky is falling" stuff that seems to show up with each new camera. I started with a 20D, went to a 5DII then a 5DIII then added an M5 and got good photos with all of them.

It is a real issue for the few folks who push the equipment to the limits. For most it's not a practical concern. Canon has a tech that has a tendency to show the issue more than most but it can be well controlled in post...most the time.

As you say...not a problem for you at all.....but how often did you go to that dark cathedral and expose the stain glass windows for maximum detail then try to also bring the dark areas back in post (from the RAW capture) to show what the human brain was seeing vs only what the OOC unprocessed jpeg would show? For those pushing the limits in that way....banding is real and worth discussing. That's a valid concern. For those folks...they may indeed be served better with a different body.

I suspect if you go back and look for it...you'll see some banding in your 5DII images. Only an issue if folks notice

 

[mmeerdam]

I am still trying to figure out whether this is a real issue or just the usual DPR "the sky is falling" stuff that seems to show up with each new camera. I started with a 20D, went to a 5DII then a 5DIII then added an M5 and got good photos with all of them.

It is a real issue for the few folks who push the equipment to the limits. For most it's not a practical concern. Canon has a tech that has a tendency to show the issue more than most but it can be well controlled in post...most the time.

As you say...not a problem for you at all.....but how often did you go to that dark cathedral and expose the stain glass windows for maximum detail then try to also bring the dark areas back in post (from the RAW capture) to show what the human brain was seeing vs only what the OOC unprocessed jpeg would show? For those pushing the limits in that way....banding is real and worth discussing. That's a valid concern. For those folks...they may indeed be served better with a different body.

I suspect if you go back and look for it...you'll see some banding in your 5DII images. Only an issue if folks notice

There is a real difference.




EOS R NORMAL RAW




5D4 NORMAL RAW




EOS R Dual pixel raw




EOS R 5 stop push




5D4 5 stop push



They said it was an improved version of the sensor. I'm not seeing a 2500 euro improvement. It's going back

 

[KrisAK]

I am still trying to figure out whether this is a real issue or just the usual DPR "the sky is falling" stuff that seems to show up with each new camera. I started with a 20D, went to a 5DII then a 5DIII then added an M5 and got good photos with all of them.

My last Canon was a 40D, which was prone to FPN or banding. As a result, I tossed quite a few images from that camera. But that was entirely on me, as I tend to get sloppy, didn't control the light and didn't work within the limits of the camera. So for me it's frankly a bit of irrational 'once bitten, twice shy.'

I never do anything approaching one-shot HDR, and I don't recall ever pushing anything as far as I did to get these issues to show up. But clearly some people do, and I respect their needs.

Anyway, if I don't buy the R, it won't be because of this.

 

[John Sheehy]

I am still trying to figure out whether this is a real issue

It's real, but whether or not it is an issue depends on the user. If you only use ISO 160 for "correct" ISO 160 exposure, or ETTR at ISO 100, use f-numbers greater than 2.2, and shoot in light that is not very weak in one color channel, it may never be an issue. If you shoot in green shade with ISO 250 and HTP at f/1.2, you're probably going to have issues.

or just the usual DPR "the sky is falling" stuff that seems to show up with each new camera.

The sky is going up and down, in regard to the level of banding noise vs random noise. We see excellent noise character in the 7D2, 5Ds(r) and 1DxII, and it is sad to see it revert with later cameras, back to visible banding despite the random noise.

I started with a 20D, went to a 5DII then a 5DIII then added an M5 and got good photos with all of them.

If you treat them like slide film, true, perhaps, of any camera, but there is no reason that photography should have to emulate slide film, when we know the flexibility that a virtually ISO-invariant camera has. How about emulating negative film? That requires gross "under-exposure" of base ISO.

 

[Yake]

Oh no, the 5-stop push again. It's a thing now. In other words: let's make a trash photo and blame the camera.

 

[David Hull]

I am still trying to figure out whether this is a real issue or just the usual DPR "the sky is falling" stuff that seems to show up with each new camera. I started with a 20D, went to a 5DII then a 5DIII then added an M5 and got good photos with all of them.

It is a real issue for the few folks who push the equipment to the limits. For most it's not a practical concern. Canon has a tech that has a tendency to show the issue more than most but it can be well controlled in post...most the time.

As you say...not a problem for you at all.....but how often did you go to that dark cathedral and expose the stain glass windows for maximum detail then try to also bring the dark areas back in post (from the RAW capture) to show what the human brain was seeing vs only what the OOC unprocessed jpeg would show? For those pushing the limits in that way....banding is real and worth discussing. That's a valid concern. For those folks...they may indeed be served better with a different body.

I suspect if you go back and look for it...you'll see some banding in your 5DII images. Only an issue if folks notice

Oh yea, I have spent a lot of time looking for this and can find it if I look hard enough (and sometimes it doesn't take that much looking). My test scenario has always been to expose for detail in something in bright sunlight with shadows near by and then pull them up. Often times a white house with the garage door open -- expose the whites for detail and then pull up the shadows and snoop in the garage.

I have done the cathedral shot but I generally have the ISO up around 1600 or so that the amplified sensor noise is masking any post gain stuff (which is what seems to be here).

I see we have EOS "R" capability in LR now so I can snoop some of those IR files (there are a boat load of them to look at).

 

[John Sheehy]

I suspect if you go back and look for it...you'll see some banding in your 5DII images. Only an issue if folks notice

Put an f/1.2 lens on the 5D2, set it to f/1.2, ISO 250 w/HTP, and walk into an area covered by trees with dense undergrowth, too, in the shade. The banding will have a party in the shadows of the images that you take there.

We should never forget that the cameras and converters can be pushing under the hood by many stops, even if we do not explicitly push in a converter.

The above scenario has an implicit push in the red channel of about 4.5 stops!

We should all look forward to the day when the only reason that we use ETTR at base ISO is to get a lower ISO exposure index, AKA, a greater exposure of middle grays in the image. Having to keep the exposures near the RAW "right" to avoid spatially-correlated read noise is not a nice place to be for everyone.

 

[John Sheehy]

They said it was an improved version of the sensor. I'm not seeing a 2500 euro improvement. It's going back

From the samples, it looks like the 5D4 with more of a tendency for the pixel row black points to toggle. Maybe the power supply fluctuates more.

I think that perhaps Canon only gave us banding-free 7D2, 5Ds(r), and 1DxII cameras by accident, and it is not part of the road map. I am going to be very p****d if the 7D3 is like this, as I like to use HTP and I shoot in very red-deprived light very often.

 

[John Sheehy]

Oh no, the 5-stop push again. It's a thing now. In other words: let's make a trash photo and blame the camera.

There is no reason that it should be trash. "Under-exposure" is not a negative quality, per se. It is simply using an ISO setting for a higher exposure index. There are no qualities to underexposure, other than greater noise, which would be greater, too, had you used the higher ISO setting for the same absolute sensor exposure.

The point is not what ISO 100 and -5 EC looks like as an ISO 100 image; it is what it looks like as the ISO 3200 image that it really is, compared to the ISO 3200 setting, which typically has much larger RAW files and 5 stops less headroom.

 

[Lobbamobba]

The camera is now supported by Adobe so the first banding tests can now be done.

https://helpx.adobe.com/camera-raw/kb/camera-raw-plug-supported-cameras.html

 

[mmeerdam]

Oh no, the 5-stop push again. It's a thing now. In other words: let's make a trash photo and blame the camera.

Thanks Yake, really helpful.

Hope your a great photographer taking his time to comment. it's well appreciated.

 

[Yake]

Just because the software has a 5-stop push setting doesn't mean it's a sensible thing to do, or a valid way to evaluate a camera. I'm sure that if the software offered 6-, 7-, 8-, 9-, 10-, or even a 100-stop push, someone would do that too, and they would find an issue there.

 

[gimp_dad]

I never shoot dual pixel.

For make sure, I just take a look on my R, and Dual Pixel RAW is set: disable

thanks, I'm gonna go to the store on Wednesday to try another copy and hopefully it is like yours.

Before you return it to the store, please make sure that you make the adjustment in Lightroom using the TIFF file that converted from CR3 (not the C-RAW) using DPP.

So, from camera CR3 - convert to the TIFF using DPP - then you may edit the TIF files using Lightroom.

This tips is shared on another thread and works very well on me

thanks for the tip! Just one thing: I'm not buying a 2500 euro camera next to my 5d4 (which is fine) to have to go this route to not get banding.

Addition: Also 16 bit tif is way different color and highlight response than a normal raw or dog file in car or c1. SO it's not really a solution for me.

This is just the temporary solution until other tools fully support the new camera and is normal for any new camera release. Going this route just helps you see what the limits of the camera really are instead of seeing issues with converters or editors. It certainly doesn't imply adopting this path permanently in your processing flow.

Not sure what differences you are seeing in color and highlight responses in a 16 bit tif file. It is the one way not to throw away any of the original data and if you are careful in how you use DPP to do the conversion you should get everything you need from the original file. The one thing you can't get rid of (at least to my knowledge) when you go to 16-bit tif is the demosaicing filter but even that has several choices.

 

[mariuss]

Just because the software has a 5-stop push setting doesn't mean it's a sensible thing to do, or a valid way to evaluate a camera. I'm sure that if the software offered 6-, 7-, 8-, 9-, 10-, or even a 100-stop push, someone would do that too, and they would find an issue there.

Lets put another way.

Just because the software has a 5-stop push setting doesn't mean you may not use it!

And btw, who are you to judge what is "sensible" for others?! We will appreciate if you allow us to be different. Thank you.

 

[mmeerdam]

Just because the software has a 5-stop push setting doesn't mean it's a sensible thing to do, or a valid way to evaluate a camera. I'm sure that if the software offered 6-, 7-, 8-, 9-, 10-, or even a 100-stop push, someone would do that too, and they would find an issue there.

I just used the 5 stop to make it clear what we are talking about, because of online compression of files. I never push more than 3 stops. But I want clean results. Not that big of an ask imho.

But Fine:
let's take some real world photo's: The new 35mm preview sample gallery.

All photo in that sample gallery look really great. I love thm, I love the lens I love the colours. It's making me doubting to return the cam.

But then this: Here is a 2,5 stop and some light shadow push of a great photo from the 35mm sample library, This photo needs it to be what it was intended to be.







This is basic streethptography stuff and any other latest generation cam, even the one with the same(ish) sensor (my 5d4) can pull this off with no sweat.

I'm not willing to spend money on the same but worse and I want to make sure others buying know up front what they are buying. They are buying a great camera, a really great camera, but it's not performing like a 5d4 when in post processing. Just so they know.

 

[zbh]

Can be any of those, but overall i am pretty happy with the IQ, will do more tests, but i've been busier doing photos, than testing bands. Yes, there are corner cases, but the more i use it, the more i like it.