Yesterday, I posted an article comparing the banding noise on today’s mirrorless cameras. The results showed the Canon EOS R to have a bit more such noise than Nikon or Sony. However, one Photography Life reader, Todd (thanks!), noted Canon released a firmware update last week to fix horizontal linear noise. I downloaded it and re-ran the test today, and the results turned out much *worse* – our analysis is below. For now, don’t update your EOS R firmware.
To start, here is the image I posted yesterday from the EOS R with the original firmware, 1.0.0. I underexposed this shot by six stops for the test, which, of course, is a laughable thing to do in real-world shooting. Although there is some horizontal noise, it is likely nothing you will see in the realm of actual photography (which was the whole point of yesterday’s article in the first place):
However, under these extreme conditions, the Canon’s banding was still a bit worse than that of the Sony A7 III, Nikon Z6, and Nikon Z7. So, I assumed there was a real possibility that Canon’s “horizontal linear noise” fix in their most recent firmware update would put it on par with the others, or at least offer some improvements. I downloaded version 1.1.0, retested the camera today, and got this result:
Not good at all. The new version has some gigantic lines of banding that didn’t exist at all in the original, especially the copper-colored streak of banding that goes across all four color swatches at the top of this crop. A wider crop of these two photos shows the problem even more (ignore the shifting figures in the background; I didn’t expect to do crops this wide):
My first thought was that the studio light might be doing something strange. So, I shot a different test outside to eliminate any confounding variables from artificial light. Same result:
(And here’s the properly exposed version just so you can see which lines are due to banding and which are due to the subject):
Then I wanted to double check that the process of converting to DNG and opening in RawDigger wasn’t causing any problems. It shouldn’t, but to make sure, I opened the original CR3 file in Capture One. Naturally, some of the colors changed, but the noise pattern did not:
I ran more tests with silent shutter on and off, and then anti-flicker shooting on and off, but the results were always the same. When recovering shadow detail from the EOS R’s new firmware, banding appeared no matter what settings I picked.
To me, that settled it. From the relatively clean file yesterday to the extreme banding today, the only thing that changed is that I updated the firmware. I really wish I had the original firmware back on the EOS R to run more side by side tests. Unfortunately, I only have one other comparable file, which is the same ColorChecker shot but with five stops of underexposure. The difference is not quite as drastic here, but the new version (“After” in the slider below) is still clearly worse. It has essentially a rainbow pattern in the shadows outside the area of the ColorChecker:
My takeaway is that you definitely shouldn’t upgrade the Canon firmware to version 1.1.0 just yet. Who knows? It might only be my specific EOS R that is experiencing this problem. I only have one copy at the moment to test. Then again, better safe than sorry; at least on my version, it seems highly likely that the cause of the new banding is the firmware update. So, if you are a Canon EOS R user who has not yet switched to 1.1.0, don’t. Keep 1.0.0 at least until other photographers have looked into this issue in more detail. After searching for a bit, I, unfortunately, did not find a way to revert to 1.0.0; I couldn’t find that version floating around anywhere online.
Yesterday, I was the one saying how silly it is to look at files that are six stops underexposed, and how unrealistic that is in practice. That remains true. If this noise is mainly visible in photos that are five or six shots underexposed, it’s essentially a non-issue. In my brief real-world tests today with the EOS R’s new firmware version, I still see some of this noise in photos that are four stops underexposed, and there remains a small bit in photos that are three stops underexposed. I don’t yet know how visible it is in properly exposed photos with significant shadow recovery and other post-processing (say, clarity adjustments), or in photos at higher ISOs with any sort of recovery, but I will update this article if I find it to be visible there too.
Have any Photography Life readers with an EOS R updated their firmware and noticed something similar? I’m just one person with a single copy of the EOS R, and I can’t generalize my results beyond that. I’d love to post more before/after comparisons, or even just images after the update (which would help determine if it’s something weird with my single EOS R copy) – but I don’t want anyone else to update their firmware just for the sake of this test. It’s not possible to say when Canon will take note of this issue and fix it in a future firmware update, but until then, keep version 1.0.0 if you haven’t changed it already.
UPDATE: Some readers below suggested that the reason may be that I haven’t yet updated the lens firmware, only the camera firmware. So, I downloaded the RF 24-105mm f/4 firmware and tried another sample, but the issue, unfortunately, remained just as strong as before. The following image is completely uncropped, the same process as above:
Below, it seems that most readers are not experiencing this problem, which is very good to see. Hopefully, this issue only affects a small number of EOS R cameras.
ive seen on other sites the banding issue too, but would like to know before updating to 1.2 if it helped any?
Can you test again with the new 1.2.0 firmware?
Fixed bending noise problem in firmware version 1.2.0 released today.
Interesting,I have one camera with updated firmware and another not yet.I can do the test to verify this.Shall update here once I made the test.Cheers!
Ronnie
www.rcstudio.com.my
All interesting as I am considering buying the Eos R. Will admit though I probably won’t be pushing 6 stops to find a fault.
One other consideration is the software that you are using on your computer, has this been updated for the camera and lens updates?
Chris
www.cjhphotography.co.uk
When you update the body firmware, make sure that you update the 24-105mm RF lens. This banding only occurs with this specific lens, and the two were meant to be updated at the same time. There are separate firmware updates for BOTH the body and the lens.
Bradley, take a look at the end of this article. I added an image captured with the new camera firmware and new lens firmware. The problem still appears.
The lens firmware for the 24-105 actually caused the banding. There was a firmware update for the lens that came out at the same time as the body update.
Spencer, thank you for taking the time to run these tests and put this article together. Frankly, the review that I watched that made a big deal about the banding made no sense to me.
I am very thankful for this test. The Z 6 is in my near future and this reality check was very helpful.
Glad you found that test so useful! The Z6 is an awesome camera, which you’ll see pretty clearly in our upcoming review.
Hi Spencer, thanks for the warning!
The Canon update site says: ‘Corrects a phenomenon in which horizontal linear noise may occur when using specific lenses together with certain recording image quality settings.’ So it doesn’t mention anywhere that the fix is to correct a problem with 5-6 stop pushed pictures, which we all agree is meaningless anyway. Perhaps the fix improves the noise performance in the more typical situation, like a correctly exposed high-ISO picture in the 1,600-12,800 range, or with certain resolutions of JPEG? Or did you see somewhere they are considering the 5-6 stop push to be ‘relevant’?
Good question. I do have a series of high ISO shots taken on the EOS R both before and after the firmware update. However, even at the maximum ISO value, neither version – 1.0.0 or 1.1.0 – has banding noise in a properly exposed photo. The ISO performance before and after the update is identical. At least as far as I see, it’s only dynamic range performance that suffered after 1.1.0.
You’re right that Canon doesn’t mention 5-6 stops of recovery in their firmware description. I’m not sure exactly what “image quality settings” they were referring to, but there is also some talk online of EOS R JPEGs having linear banding, so perhaps that’s what the firmware update was targeting instead. If Canon was focusing on RAW images, though, perhaps it was even successful on most EOS R copies; no commenters above seem to have replicated the issue I see so clearly. At a minimum, though, if Canon was trying to improve RAW photos, something is clearly wrong with at least my EOS R copy that did the opposite of Canon’s intent.
Seem to be a lot of haters about on all these forums slating one brand or another it’s no wonder people get suspicious of sponserd negative reviews . I have updated firmware on lens and camera and not noticed anything irreguler but then again I don’t push 6 stops in pp. ? My bad I must try harder.
I’m glad that you haven’t noticed anything irregular on your own sample; hopefully that is also the case for most users.
I have to mention, though, that your point about sponsored negative reviews is inaccurate. I’m not sure about the laws behind user reviews left on websites like Amazon, but at least for larger websites, you will always see when a post is sponsored (and I’ve never seen any such articles that are negative in tone, usually over-the-top cheesy positive). To my knowledge, Dpreview and Fstoppers are the only two big photography sites that do such articles, and they always display the “sponsored” notification prominently. To do otherwise wouldn’t even be legal, either for the website or the camera company, and I don’t think anyone (website or company) is very willing to risk paying the fines associated with that sort of deception.