R5 LR noise

[EdSafari]

I recently obtained a R5 but seem to be struggling with noise shown in the RAW file in LR even for low ISO (e.g. ISO 800). The same image on DPP shows hardly any noise. I also tried converting to DNG and importing that to LR but it still shows too much noise. Below is the .CR3 image on LR at 100% showing noise followed by the same image in DPP showing less noise. Is this a LR or R5 issue? Do I need to run all images through DPP (export to JPG) before editing in LR? (I use Manual Auto ISO). Please help.

 

[AlanF]

I recently obtained a R5 but seem to be struggling with noise shown in the RAW file in LR even for low ISO (e.g. ISO 800). The same image on DPP shows hardly any noise. I also tried converting to DNG and importing that to LR but it still shows too much noise. Below is the .CR3 image on LR at 100% showing noise followed by the same image in DPP showing less noise. Is this a LR or R5 issue? Do I need to run all images through DPP (export to JPG) before editing in LR? (I use Manual Auto ISO). Please help.

View attachment 201290

View attachment 201291

This has come up a few times in other posts from people who use LR and some other software where it takes real skill to remove noise. Use decent software for removal of noise like DxO PL for RAW, Topaz Denoise etc and you will have an easy answer to your problems. Here is a thread where I went as high as iso 40k https://www.canonrumors.com/forum/threads/extenders-and-high-iso-with-the-r5.40575/

 

[EdSafari]

Thank Alan. I do use Topaz Denoise but other than losing sharpness doing that it seems a bit cumbersome to have to include Topaz DeNoize in my workflow for every image ISO 600 and greater for such a highly rated camera. But I suppose it is what it is.

 

[HenryL]

I recently obtained a R5 but seem to be struggling with noise shown in the RAW file in LR even for low ISO (e.g. ISO 800). The same image on DPP shows hardly any noise. I also tried converting to DNG and importing that to LR but it still shows too much noise. Below is the .CR3 image on LR at 100% showing noise followed by the same image in DPP showing less noise. Is this a LR or R5 issue? Do I need to run all images through DPP (export to JPG) before editing in LR? (I use Manual Auto ISO). Please help.

View attachment 201290

View attachment 201291

I don't use DPP so can't make recommendations regarding that, but in the DPP sample you provided there is a lot of noise reduction being applied. Significant feather detail has been lost in the process, it looks pretty bad imo. You might get a good result using DPP as a first step, but I personally find it slow and cumbersome to use so I couldn't tell you how.

Large areas of solid color like the background in your photo will show noise much more than a "busy" section of an image. It also appears that the shadows have been lifted to a greater degree in the LR example, which also adds to the appearance of noise. Still, depending on the rest of the image, a background like what is shown is super simple to clean up.

The LR version is not bad and you could probably improve it quickly with judicious use of the noise reduction sliders. Personally, for that example, I would run it through DxO PL5 first, then import the DNG into LR. That is not by any means required for all photos, just certain cases. Depending on whether you use Lightroom to catalog/manage your photos, you may not even need LR if you had PL5. Topaz Denoise would also fix that right up, works as a LR plugin, but I more of a one trick pony.

Now would be a good time to download free trials of the software options - they usually have Black Friday deals of 30% or more off. Worth a look imo.

 

[HenryL]

I should add - I regularly shoot up to ISO 6400 without hesitation with the R5, occasionally higher, and find that 800 is no problem at all. If you are struggling with noise consistently at ISO 800, there's probably some aggravating factors in play.

 

[AlanF]

Thank Alan. I do use Topaz Denoise but other than losing sharpness doing that it seems a bit cumbersome to have to include Topaz DeNoize in my workflow for every image ISO 600 and greater for such a highly rated camera. But I suppose it is what it is.

As HenryL wrote, there are some good deals now for software. I highly recommend DxO PL as it takes care of noise automatically in the RAW conversion remarkably well without disrupting work flow. I use a couple of Topaz products, but not DeNoise, when necessary as PL is my choice for RAW conversion.
The files from R5 are no noisier than from other high resolution sensors I have used, like the acclaimed one from the Nikon D850 or the older 5DSR.

 

[EdSafari]

Thanks Henry. I've never used DxO PL5 but see that have a free trial version, so I'll give it a go. I assume "..there's probably some aggravating factors in play" means that the camera could be faulty?

 

[HenryL]

Thanks Henry. I've never used DxO PL5 but see that have a free trial version, so I'll give it a go. I assume "..there's probably some aggravating factors in play" means that the camera could be faulty?

Yes, do try it if you can. it's not the least expensive option, but I find it one of the most capable tools.

By aggravating factors, I'm not suggesting faulty camera at all - far from it. Personally, when I run into situations like this consistently, I tend to look inward first. In short, I struggled with noise and an unwarranted fear of high ISO for example for a long time. This led me to many underexposed images, which in turn required me to lift shadows in when processing, which led to - wait for it - more noise.

Look at your software first - clearly the exposure in DPP and LR that you offered are using different default settings and both end up with unacceptable results but for different reasons. Is that a crop, if so how much of the subject actually fills the frame? Too much noise reduction in DPP version causes loss of detail. Perhaps too much lifting of the shadows in the LR version increases appearance of noise. These are the kind of things I look at first, because I fully understand that my gear is far more capable than I am, so if something isn't working, I'm the weakest link.

What does the full image look like before any adjustments? Can you reset in LR and show the full frame including histogram? That would perhaps allow for a better assessment here.

 

[EdSafari]

Thanks Henry. What you say makes sense. I am at that stage of fearing noise resulting in underexposed images. Here's the uncropped and histogram taken back to original settings. Definitely under exposed. I had to increase exposure and shadows on LR to make the image more acceptable. I'm learning a few lessons here. Thank you.

 

[Michael Clark]

Thanks Henry. What you say makes sense. I am at that stage of fearing noise resulting in underexposed images. Here's the uncropped and histogram taken back to original settings. Definitely under exposed. I had to increase exposure and shadows on LR to make the image more acceptable. I'm learning a few lessons here. Thank you.

View attachment 201293

You'll get less noise by just going ahead and raising the ISO when shooting so that you don't have to raise exposure in post. That's the best way to shoot unless you're keeping the ISO lower to protect from blown highlights that might be in the scene.

(For example, when I shoot sports I tend to underexpose the mids by about half a stop or so to prevent blowing out the brightest spots on the white jerseys that are almost universally worn by one of the two teams in most sporting contests here in the U.S. I then raise exposure in post while pulling the highlights to keep them from blowing. If I shot it at higher ISO they'd be too blown to recover in post.)

 

[Michael Clark]

When you open a raw file in DPP the in-camera settings that were active at the time the image was captured are applied by default. It looks like you've got pretty heavy noise reduction enabled in your camera settings. That, or you've got "Auto" NR enabled and the camera is setting the NR fairly high because of the amount of noise in the underexposed image. While this does not affect the actual contents of the sensor data stored in the raw file, it does affect what NR settings DPP will apply when you open the image file using DPP. LR (and most any other Adobe product) tends to ignore a lot of that and use it's own default settings when you open a raw image file in LR, unless you've changed the default settings for LR. So you're applying two totally different sets of settings to the same sensor data when you open the image file in two different raw conversion applications.

There's no such thing as 'THE raw image". There are only a near countless number of possible interpretations of the data contained in a raw image file. A raw file is much like a latent image on an undeveloped negative - how it's developed will affect things like contrast and overall brightness, etc. The biggest difference is that you can only develop a negative once and then have to live with the results. There's much less flexibility with what we can do at the enlarger than there is with what we can do when developing a negative. With raw data we can go back and redevelop as many times as we want. But once we export to a different format, such as TIFF or JPEG (and many forms of DNG, depending on the implementation of the DNG convertor used), we "lock in" a lot of the development decisions we made when converting from the raw data to a raster image. Much of the information in the raw data is irretrievably lost in the conversion. It's still there in the original raw file, but it's not all included in the exported conversion.

 

[unfocused]

My thoughts.

Your cropped image vs. your uncropped image shows a pretty major crop, so I'm not surprised by the noise, especially since the image was underexposed.

You can get similar noise reduction from Lightroom (Actually I use Camera Raw, but it's the same thing) as the DPP file, depending on how far to the right you move the noise reduction slider. How much noise reduction did you apply in Lightroom? In either case, you will sacrifice detail for noise reduction. It's a balancing act and you have to move the slider until you reach the balance of noise vs. loss of detail that you find acceptable. It looks like the DPP image just has a WHOLE lot more noise reduction applied.

Generally speaking, I've found that Canon's newest sensors (I use an R5 and a 1DX III) perform pretty spectacularly at higher ISOs. Indeed, when I'm in a hurry to process pictures from a sporting event I often forget to apply noise reduction in Camera Raw to files at 6,400 or below and don't get much noise even with default settings. But, then I am generally not making massive crops and most of my images get downsized to about 1140 px for the web and social media. I also take care not to underexpose.

There are a few things you can do to reduce apparent noise. Noise is often most visible in out of focus backgrounds and skies. With Lightroom or Camera Raw, use the new masking tool to select the subject and then invert that selection so that everything but the subject is selected. With that, you can apply some pretty radical noise reduction to the background without impacting your subject. In the example you are showing, all that noise in the sky can easily disappear with no harm done, since there is no detail there anyway (and you really wouldn't want detail there even if there were some.) Oftentimes, simply eliminating the noise in the background will give you acceptable results, because eyes will often tolerate more noise in the subject than in a blurred background.

Of course the best things to do are to not underexpose in the first place, avoid extreme crops and don't get caught up in pixel peeping at 100% when no one but you is going to see your image at 100%

 

[EdSafari]

Super sound advice. Thank you all.

 

[AlanF]

My thoughts.

Your cropped image vs. your uncropped image shows a pretty major crop, so I'm not surprised by the noise, especially since the image was underexposed.

You can get similar noise reduction from Lightroom (Actually I use Camera Raw, but it's the same thing) as the DPP file, depending on how far to the right you move the noise reduction slider. How much noise reduction did you apply in Lightroom? In either case, you will sacrifice detail for noise reduction. It's a balancing act and you have to move the slider until you reach the balance of noise vs. loss of detail that you find acceptable. It looks like the DPP image just has a WHOLE lot more noise reduction applied.

Generally speaking, I've found that Canon's newest sensors (I use an R5 and a 1DX III) perform pretty spectacularly at higher ISOs. Indeed, when I'm in a hurry to process pictures from a sporting event I often forget to apply noise reduction in Camera Raw to files at 6,400 or below and don't get much noise even with default settings. But, then I am generally not making massive crops and most of my images get downsized to about 1140 px for the web and social media. I also take care not to underexpose.

There are a few things you can do to reduce apparent noise. Noise is often most visible in out of focus backgrounds and skies. With Lightroom or Camera Raw, use the new masking tool to select the subject and then invert that selection so that everything but the subject is selected. With that, you can apply some pretty radical noise reduction to the background without impacting your subject. In the example you are showing, all that noise in the sky can easily disappear with no harm done, since there is no detail there anyway (and you really wouldn't want detail there even if there were some.) Oftentimes, simply eliminating the noise in the background will give you acceptable results, because eyes will often tolerate more noise in the subject than in a blurred background.

Of course the best things to do are to not underexpose in the first place, avoid extreme crops and don't get caught up in pixel peeping at 100% when no one but you is going to see your image at 100%

Above iso 800, the R5 sensor is true iso-invariant, even with ES, so there is no noise disadvantage on increasing iso in post in RAW than using a higher iso to start with (between iso 800 and about 40k). It's a real improvement over previous Canon 5-series sensors. (The horizontal parts of the shadow improvement charts in photonstophotos show this https://www.photonstophotos.net/Cha... EOS 5D Mark IV,Canon EOS R5,Canon EOS R5(ES) The 1D later series are similar but not the 5DIV.

 

[kaihp]

Above iso 800, the R5 sensor is true iso-invariant, even with ES, so there is no noise disadvantage on increasing iso in post in RAW than using a higher iso to start with (between iso 800 and about 40k). It's a real improvement over previous Canon 5-series sensors. (The horizontal parts of the shadow improvement charts in photonstophotos show this https://www.photonstophotos.net/Charts/PDR_Shadow.htm#Canon EOS 5D Mark IV,Canon EOS R5,Canon EOS R5(ES) The 1D later series are similar but not the 5DIV.

Side-question here: so since the R6 is at around 1 EV and the R5 is at 0.4EV, the R5 shows less (extra) noise when boosted in post vs exposing right?
Seems like that the measured SNR isn't significantly higher between the 5D3 and the R6/R5, but if I understand the chart above correctly, the R6/R5 produce cleaner images if they are underexposed and then lifted in post, than the 5D3 would.

(Edited for typo and elaborated my question)

 

[Czardoom]

.... avoid extreme crops and don't get caught up in pixel peeping at 100% when no one but you is going to see your image at 100%

If you go looking for noise, you are likely to find it. The amount of noise in your 100% crop is minimal. Unless you see a lot of noise when viewing your photo full size, don't even worry about it. That's my advice.

 

[AlanF]

Side-question here: so since the R6 is at around 1 EV and the R5 is at 0.4EV, the R5 shows less (extra) noise when boosted in post vs exposing right?
Seems like that the measured SNR isn't significantly higher between the 5D3 and the R6/R5, but if I understand the chart above correctly, the R6/R5 produce cleaner images if they are underexposed and then lifted in post, than the 5D3 would.

(Edited for typo and elaborated my question)

With the 5DIV, you get noisier images if you enderexpose and increase exposure to the correct level post exposure. With the R5 in ES or mechanical shutter and the R6 in mechanical, you can underexpose when set at iso 800 or more and push exposure without any increase in noise up to 40k or so. With the R6 in ES, pushing in post increases noise. The R5 suits my style as I am usually at iso 800 or more and, for example, when switching between perched and flying birds I can increase the shutter speed 2 or 3 stops in a fraction of a second without changing aperture or iso, knowing I can compensate post exposure.

 

[Sporgon]

Not sure if anyone has mentioned this, but the default sharpening in LR can emphasis shot noise on the higher mp cameras. I find it’s best to mask slightly with the 5DS if I’m processing in LR. On a Mac if you hold the alt key down whilst moving the mask slider you get a black and white image of what’s being masked. I find this useful.