You are here

This ATEM (mini) Tips” series of short video tutorials for your Blackmagic ATEM hardware includes tips for all users, from beginner to advanced. Scroll down to see more in the series!

This is a collection of all PhotoJoseph LIVE shows, from the PhotoJoseph YouTube channel. Additional episodes are listed below!

Does Noise Reduction Matter When Shooting RAW?

Photo Moment - May 04, 2018

Your digital camera has a Long Exposure Noise Reduction feature. But does it make a difference if you're shooting RAW?

 

Products Mentioned In Today's Photo Moment

Share
Hopefully that cleared up the picture for you.

See something on the show that you want to buy?
Head on over to kit.com/PhotoJoseph

•• Panasonic LUMIX GH5 ••
B&H - https://bhpho.to/2kQ9LT3
Amazon - http://amzn.to/2lE4X1U
Adorama - http://jal.bz/2lNYpQv

•• Panasonic LUMIX G9 ••
B&H - https://bhpho.to/2Al5ZGr
Amazon - http://amzn.to/2zGxOww
Adorama - http://jal.bz/2zsiLW7
Thank you I had to what is 3 times before I got it. I alwayes save in RAW and JPG
Not in all cases, in nightcape photography it does matter, becuse it effects camera metering at least in Nikon system, and you'll get some weird noise colors, you can fix in post, but if im taking single images not tracked, or stacked i would rather keep long exposure NR On, it definitely helps.
Yes you should enable NR in Raw if you are doing night sky photos. If you don't with a G9 you will get all kinds of red, green and blue dots all over the screen that are extremely hard to remove. Its much better to turn on NR and makes a huge differnece.
Exactly my question. I'll have to interpret to my own uses of course, thanks.
The reason the JPG with noise reduction ON get strange colour noise in the image is because NR is mainly mend to reduce DC (dark current) noise (a slow and steady build-up of charge in the pixels due to leakage current, which varies with heat) and maybe some readout noise (a type of noise that is added to an image while the signal is being read and converted from analog to digital). But the signal noise (each signal has its own noise) is not because signal noise is random and the other noises are patterns. And because the darkframe (the image with the mirror closed that is shot as second to reduce the noise) also contains it's own signal noise and this noise can not be filtered without exposurestacking, this signal noise is actually INJECTED (!!!) in your lightframe (actual photo).
Good video 👍
can you please create a video how to edit noise in raw images
thanks a ton.. important info.
Most welcome!
You used only 1 camera... I came here because I heard somewhere that more recent cameras do started allowing NR to affect the RAW...
Unless I have completely the wrong idea and perspective, what you've done is to completely confuse the issue. First of all, LENR is not about high ISO noise and improving that. Using ISO12800 is irrelevant, as is using an exposure of 1 second. LENR is about mapping out hot pixel artefacts caused by 'rogue' pixels that go crazy when they heat up. A sensor heats up during long exposures and hot pixels show up particularly in shadow areas because they shine brightly in comparison to those around them, so you will see hot pixel 'noise' (coloured pixel dots) in shadow areas in particular. Please try your tests again using ISO100 and proper long exposures - how about a minute or two minutes perhaps or maybe even longer. Looking at your images I saw no hot pixel noise at all because the exposure you used was way too short - LENR will not have done a great deal to improve things if no hot pixels were present - maybe that's why the JPEG got worse, who knows. Hope that helps - unless I'm wrong of course!! ;-)
Totally agree. This video unfortunately completely misses the point, but to be fare there is not a good explanation in the sony literature. LENR is only about removing thermal noise and it dose a good job, This will show up at long exposures at lowish ISO in particular.
Great informative video. Thanks! I want to experiment with NR myself and see if it is worth it.
Hi Joseph! Just a comment here that you may want to check since I am not 100% sure about it. I think that in some cameras, the long exposure noise reduction is a feature that has more to do with punctual pixels artifacts rather than general noise reduction. It makes sense since by definition noise does not present a pattern, even if you repeat exactly the same conditions and settings. But, when it comes to punctual pixel artifacts, patterns do appear and you can recognize them by repeating the shot on a known image (completely black in this case). By doing this the camera software can identify those wild (hot) pixels which go to the maximum level even though no light is reaching them. Moreover, the hot pixel behavior is observed almost particularly on long exposure shots, that's why it makes sense to have this option devoted for long exposures pictures. Anyway, I hope you can confirm if this story that I've read somewhere is true. Thank you! Franco
Noise reduction in the G9 is a MUST if you shot raw for something like 50sec and more. Without it, the image is full of hot pixels in the shadows, completely trash after zooming in. The noise reduction gets rid of that, it's a life saver!
Great Video and super important Topic!!!!!!!
Long exposure NR only kicks in for exposures OVER 1 second.. And it most definitely makes a big difference. It employs DFS, for starters.🤦🏽‍♂️
*QUESTION to PhotoJoseph or anyone that can HELP!*
I've switched from a Canon 6D to GH5. I've noticed the RAW Canon .CR2 files to be much cleaner (in terms of noise/grain) compared to the RAW Panasonic .RW2. 
After watching this video, apparently it's because you mentioned most DSLRs auto apply NR in-camera, correct?
How the heck do I apply clean NR to my GH5 without the photo becoming muddy like the example you've shown?! Especially since you also mentioned I doesn't make a difference in-camera or in post!? Definitely don't want the GH5 sensor noise showing in my pics. 

THANKS and love your channel mate!
You really shouldn’t have much if any visible noise at that ISO for a normal exposure. Maybe the NR that Lr is applying is more aggressive to the Canon? Yes a full frame sensor will likely have less noise than an MFT one; that’s just physics. But it should still be very clean. Maybe turn up the NR a little and see what happens and yes a default preset on import is a good solution if that gives you what you’re looking for. Personally I very rarely add any additional NR and certainly not at low ISO like that. Maybe it’s just a difference of what I’m used to though? Instead of comparing side by side I’d say just start shooting and see if you feel like your files are noisy.
PhotoJoseph Thanks for the speedy reply!
Test images: ISO 100 & 400 (I believe native ISO) in an outdoor overcast environment.

Using Adobe LR Classic CC (Adobe standard profile) to pixel peep 1:1 in comparison mode. Got my .CR2 file right up next to the .RW2 file and the .CR2 file is noticeably smoother/less noisy. 🤷🏻‍♂️

I may have to save NR as a preset and apply upon import? 🤔

Appreciate the help you legend!
How high of ISO are you shooting? What software are you using to process the RAW files?
on the gx80 the raw file does have a great difference between on and off
got a lot of hot pixel at longer exposures like 128 seconds and if i turn the feature on it removes them all
no problem :)
one question since we are talking about long exposure.
Do you know a way to get a timer on the screen while using the T-Mode?
On my Canon the Bulb mode shows a timer displaying how much seconds my shutter is open.
On the Lumix I always have to guess because the screen is just black with a red dot blinking. For example I would like to take a picture with 22 seconds shutter speed but my only solution for now is to have a clock and look at the time.
Couldn't find anything that could answer this.
Very interesting. OK I need to re-test, clearly. At longer exposures. Thank you!
here is the difference left without longshutterNR and right with NR both raw files
i.imgur.com/rAUsIiS.jpg
128 seconds max exposure of the gx80
it makes sense
thats why the noise reduction takes as long as the shot itself because it takes a second picture to see all the hotpixel and eliminates them
and it combines the pictures even in raw
Really! How long was your exposure test?
Although I find you're test interesting, I think you're not quite right with your opinion. You only tested shutter speeds of 1 second. Make the same test with 60 seconds shutter speed. You'll find many hotpixels in the picture which has no NR in opposite to the one with NR. :)

Greetings vom Germany. :)
I know I am late to the party, but I encountered the same yesterday. I did some tests with my GX9 vs my Nikon d780 in low light outside in the garden.
On The GX9, if you turn off the LENR, above 2 or 3 seconds exposure time, the images are just not usable, because hot pixels are creeping in everywhere. This btw is further more strongly correlated to the ISO value.

Greetings from Germany!
Sounds like I need to revisit this test
I too agree that Josephs test is wrong. Long exposure noise reduction in RAW may not touch ISO grain, but indeed removes hot pixels! Take 2 photos in Raw (one with and without NR), use ISO 200 to mitigate grain @ 8 seconds each. You will see a huge difference.
Yeah, that's the point. But is very simple to test it as I did. Just put the lens cap on your lens, then take a shot of 30-40 seconds with LENR on and then OFF. I've done it with my G9 and results are clear: with the LENR off your image is frankly unusable!
Great point. I also wonder if it would make sense to shoot a black raw afterwards to sustract it in post in order to get a NR-Raw.
I’d not heard Mirrorless systems referred to as “DSLM” before. Makes sense. 👍🏼
Thinking about the camera as a little computer, it would also stand to reason, that it probably uses more battery power to have the processor make adjustments to every shot. We can probably make batteries last a bit longer by turning off all these little adjustments. Especially since they only affect the JPG files.
I am almost sure that one long exposure of one minute takes way more energy that just taking regular pictures in that minute, because the mechanical shutter needs to be kept agitated. I remember when I took some long exposures, the battery was empty in quite a short time.
Yeah, that’s likely true. Of course this only applies to long exposure photos, which you may not have many of, but yes it’s still a valid point!
Great video. I learned something today. Thanks!
Awesome!
I think the difference you see between no long shutter NR and long shutter NR comes from different algorithms that are used for the two cases. For long shutter NR subtracts "black" frame (the noise pattern) from the photo, while the other mode tries to perform low pass filtering. Hence different results.
I'm wondering whether the results would be the same using a different camera to the G9?
Almost certainly. The question is how big the difference would be. Each brand needs to code how and how much of an adjustment is applied. They might also be scaling it based on the power of the processor in the camera. The ones with a powerful CPU could potentially run far more complex operations or make the same adjustments far quicker.
You may login with either your assigned username or your e-mail address.
Passwords are case-sensitive - Forgot your password?