5D2 RAW "Crosshatch" Banding

[dryanparker]

I'm wondering if anyone's seen this subtle "crosshatch" style banding on 5D2 RAW files...

EXIF:

5D2
TS-E 24/3.5L II
f/5
1/40
ISO 100
B+W 010 UV Haze Filter

By no means do I feel like I was "pushing" the camera. It was a pretty standard sunset position. This is the first I've seen this on my files, though I haven't shot much at dusk since purchasing the 5D2 a few months ago.

I could actually see the bands even more clearly on the camera LCD. Pushing the exposure up a bit in ACR reveals the pattern.

Full, Crop and Crop+3EV from different part of the image attached.

 

[Drizzt321]

Yes I've seen that on mine, although usually at higher ISOs and/or where there is a seriously underexposed section of the image or it's pushed +EV too far in post.

I've found the 5d3 has significantly less ISO banding, and generally doesn't have a grid/crosshatch patter when there is some.

 

[Marsu42]

Full, Crop and Crop+3EV from different part of the image attached.

+3EV simply isn't what these kind of sensors are made for, full stop. And in comparison to crop bodies like the 7d, the banding on your shot is practically non-existent and only occurs if you're trying to recover shadow data that simply isn't there in the raw file. For more dynamic range esp. at base iso, get a d800

I've found the 5d3 has significantly less ISO banding, and generally doesn't have a grid/crosshatch patter when there is some.

"Less" in comparison to what? 18mp aps-c (obviously)? 5d2? Any reviews/proof or just your (valid) feeling?

 

[dryanparker]

+3EV simply isn't what these kind of sensors are made for, full stop.

To be clear, I wasn't processing the image +3EV. I only did that to better show the banding I was seeing in the untouched RAW file and on the LCD after capture.

No plans to jump ship over this. The 5D2 rocks. Just wondering if it's commonly seen, or if there's something wrong.

 

[dryanparker]

Perhaps this better illustrates the issue...different exposure in the same set.

f/5.6
1/20
ISO 100

This is simply opened in ACR, cropped and saved as a JPEG. Visible vertical banding, no?

 

[Marsu42]

To be clear, I wasn't processing the image +3EV. I only did that to better show the banding I was seeing in the untouched RAW file and on the LCD after capture.

Yes, I understood that, sorry for conveying the wrong message. But it was really necessary to +3EV it, because on the original shot imho the banding is next to none and you really have to look for it. I could show you some 60d shots which were a little underexposed to begin with and then after +1.5ev'ing them were absolutely unusable due to shadow banding. My advice: expose to the right, this makes shadow recovery easier.

All things considered: Rejoice, there's plenty of room for future sensor improvements during the next decades and countless occasions to liberate you of your money

 

[dryanparker]

...there's plenty of room for future sensor improvements during the next decades and countless occasions to liberate you of your money

Quite true! Perhaps I should start saving up for the jump to Phase One...ha!

 

[Marsu42]

I recently got my 5DII and I noticed the same thing this weekend. I think it is the result of using the anti-clipping feature in the Menu

Highlight Tone Priority? This essentially underexposes by one stop.

Although the 60D is supposed to be an inferior camera, my 60D never demonstrated this pattern.

I also seldom see it, only when I screw up the exposure in the first place or there's a very dark background (like dark green trees) I'd like to raise in pp behind a very bright front subject.

 

[sweetcancer]

This is nothing uncommon, I see it sometimes with my 5d2.

 

[Drizzt321]

I've found the 5d3 has significantly less ISO banding, and generally doesn't have a grid/crosshatch patter when there is some.

"Less" in comparison to what? 18mp aps-c (obviously)? 5d2? Any reviews/proof or just your (valid) feeling?

Less in comparison to similar shooting conditions (same event), shooting with both my 5d2 & 5d3. At ISO 3200 my 5d2 exhibited similar grid pattern of ISO in the darkest areas, even without any exposure changes in post. My 5d3 at ISO 6400/8000 does of course exhibit noise, but even in those darkest areas it looked to me like it wasn't all that banded, and what there was did not have a grid pattern, it was just horizontal or vertical (I forget which right now). I can dig out a couple of photos from that evening and post some full res JPEGs if you'd like.

 

[Marsu42]

I can dig out a couple of photos from that evening and post some full res JPEGs if you'd like.

Yes, some samples would be interesting, esp. if the 5d2 shows banding @3200 or even @1600 and the 5d3 doesn't. Atm I am going to buy a 5d2 because I don't see significant improvements in the 5d3 sensor at least at these iso ratings I'm likely to use a lot.

 

[sandymandy]

ISO Performance is probably the biggest improvement of the 5D MK3

 

[Marsu42]

ISO Performance is probably the biggest improvement of the 5D MK3

Again, this is a mostly a marketing myth. The 5d3 has some nice features, but the sensor is not among them and basically the same as on the 5d2.

Using *raw* I cannot see any difference up to iso1600 and next to none at iso3200 - there is some at iso6400+ but then dynamic range decreases. That's by looking at the dpreview samples, I even imported them myself in Lightroom for comparison. The 5d3 only has the "nicer" noise pattern.

 

[Drizzt321]

I can dig out a couple of photos from that evening and post some full res JPEGs if you'd like.

Yes, some samples would be interesting, esp. if the 5d2 shows banding @3200 or even @1600 and the 5d3 doesn't. Atm I am going to buy a 5d2 because I don't see significant improvements in the 5d3 sensor at least at these iso ratings I'm likely to use a lot.

Alright, let me dig them out tonight after I get home. I think there are some improvements to the 5d3 sensor, such as gapless microlenses, but as you say, probably not as much as marketing is making out. The larger number of read lines and improved initial processing probably make a difference.

I think it's just a bunch of small tiny improvements that make it look (to me at least) like there's a decent improvement in image quality (leaving aside DR) at super high ISOs. Just being able to get a pretty usable image at ISO 6400+ with decent shutter speeds (f/4, 1/80+) is pretty awesome, don't you agree? All depends on your use case of course.

 

[Mt Spokane Photography]

If you have to use plus 3 EV to see it, its not a issue. It does depend a lot on the exposure, I usually expose raw images to the right, this allows you to pull down the exposure rather than boost it.
My 5D MK III was a little better in this regard, but you can still pull it up 5 EV and see the sensor.

 

[Kernuak]

I see less banding on my 7D than on my 5D MkII (and much less than my 40D had), although overall, it is noisier. From the short test I did with the 5d MkIII, there is a significant improvement in high ISO noise, perhaps a stop or so. However, I tend to slightly overexpose and rarely have problems with banding on my 5D MkII at low ISO. Any shadow noise is usually minimal and irrelevant in print, the exceptions (apart form when I've underexposed) is in long exposures (i.e. 20+ minutes), where most digital sensors start to get problems. It's probably one area where film is still better in terms of overall image quality. Some of the noise is due to heating up of the electronics, so there is an argument for boosting the ISO and reducing the exposure times.

 

[dryanparker]

If you have to use plus 3 EV to see it, its not a issue. It does depend a lot on the exposure, I usually expose raw images to the right, this allows you to pull down the exposure rather than boost it.
My 5D MK III was a little better in this regard, but you can still pull it up 5 EV and see the sensor.

Well, it's present and visible in the untouched RAW file at ISO 100. I pushed to +3EV simply to show the pattern.