Interresting video. I appreciate your introduction because many beginners photographers think that with high Iso, they will get a very clear and bright picture, although the subject is dark. As I always say : "in photo, dark is still dark. If your subject is not really visible with your eyes, don't take the picture"
Very informative video, thanks. I use Auto ISO, but cap it at the level where noise starts to appear, usually 3200 for M43, maybe 1600 for older cameras. If the shot is underexposed, then I adjust in post. There are certainly exceptions as you note. If there is high contrast and I am specifically trying to keep detail at one end or the other, I'll adjust for that. Otherwise, Auto ISO with a cap works pretty well for me.
Great comparison! I own the GH series cameras plus I've had the S5 for a month now. One thing I learned early with the MFT lineup is NEVER OVEREXPOSE! I generally try and stay about a stop under as my rule. The FF S5 on the other hand from many of the reviews I've watched seemed to handle overexposure better in the reviews I saw on TH-cam. I will say that I love the S5 low-light capabilities, but when pushed even a stop over, the same thing happened, I lose important details in the highlights I cannot bring it back as you showed herein. So I agree completely with your findings. Doesn't matter what format, shooting images or video, -1 stop is where it's at!
Matti; It's only a personnel preference but I prefer using the camera settings to get the best picture. I feel more like a photographer when I do this as oppose to using post production software which seems less like photography to me. I'm old fashion I suppose and I am missing some improvements that post processing can provide but I'm still smiling when I get a great photo!
I enjoyed the video! It's good to learn new things every day. In similar studio situations, I was a bit reluctant to apply ETTR to the actual light source but after this video, I am going to expose to the light source and get more details from it. Regarding ISO settings, mine vary depending on my activity. If it's landscape, studio, astro and cityscape, I get the best ISO I can in-camera. For street with a lot of moving subjects, I use auto-ISO. At night, I adjust my Auto-ISO depending on the light sources. If the streets, are well-lit, I will adjust the maximum to 6400 ISO but if the lights sources start to become weaker and weaker, I boost the max to 25,600. I still want to get my moving subjects as sharp as possible. For bird and wildlife, I manually select the lowest ISO possible but if the light from the sky varies greatly, I use Auto-ISO, as well, with a lower max. Cheers.
Interesting tests Matti, thanks. I usually shoot my G9 about one third to two thirds over exposed as I find that I can still pull back the highlight details very we’ll which benefits the shadow too. I will have to give this under exposure method a try and see how it goes. Thanks Matti!
I've noticed when I push lower ISO in post that my GX85 has almost too much saturation, but other than a simple fix on the saturation slider, this confirms my thoughts for getting the best dynamic range tonality from these micro 4/3 sensors. I'll have to try it as well on my OM-D E-M1 MkII!
Thanks! On G9, there really is such a feature, when pulling from the shadow +2 and more stops, dark colors are painted in magenta. You need to expose correctly and not be very afraid to raise ISO to 1600 :)
Thank you for this video. I have been experimenting with changing the ISO for pictures and video. In my opinion keeping the ISO as low as reasonably possible is best for preserving your best photo. However, getting the detail may cause overexposure/underexposure in one area or another, which can be addressed in post. Sometimes you have to work with what you have or abandon the shot. I found on the G9 that ISO 3200 is very good in low light, and 6400 is useable. Above that it depends on where I need to use the photo, example Facebook. As we get better at sensor quality and ability to record more light waves as it is being shot.
I have only encountered the G9 colorcast issue twice. Both were indoor shots where I wanted to keep details in the landscape through the windows while bringing up the indoor shadows. I was able to correct in DXO PL4, but it took some trial and error. I guess that's the price we pay for better high ISO noise performance on the G9 relative to the GX9 and G90. I use the G9 like you would use a Canon 5DIII, letting the auto ISO do it's thing gets you close, then push or pull highlights and shadows to taste.
I've not experienced the magenta cast to which you refer with my G9. I generally use the zebras to expose for the highlights as I know I can more easily recover shadow detail in post (although there's a bit of latitude at the high end as well). Since I got DxO PhotoLab 4 Elite with its amazing DeepPRIME noise reduction, high ISO shots are no longer a concern (there goes another supposed FF advantage). In low light situations, I just set aperture and SS to suit the scene and use auto-ISO, knowing I can get the best possible image in post. It all seems just too easy now compared to film. Of course, there are cases where HDR bracketing may be the best option, but that's not often the case for me. Also, a bit of grain never hurt anyone and in fact I add some now and then as it can look more "natural" and pleasing than a too-tightly-rendered image (like when you see those 4K TVs in the store--the images are often so "harsh" to the point of being annoying). If I want sharp, I can get razor-sharp--it's just not always called for.
Yes, I think it was Tony Northrup that stated that pushing underexposed images was exactly the same as using a higher ISO. It seems your observation of improved highlights is what he omitted to notice.
The magenta cast on the G9 occurs randomly. In a 300 image timelapse about 30 have this magenta shift. Has nothing to do with dynamic range. It's just a glitch in the G9.
This makes sense as the GX9 has the exact same sensor but seems to not suffer from the problem as much. This glitch seems software related rather then hard backed in the camera.
Tony northrup has an interesting video showing that it is lack of light, not iso per se that causes noise. Reducing the aperature or fast shutter speed to limit light will cause noise too.
Thank you for your videos! they are very detailed and very well put together! I am new to the photography and your videos has helped me a lot. I bought the G95 (with the 12-60 mm F3.5-5.6 kit), also bought the 25mm/F1.7 and lastly the Leica DG Nocticron (brand new for $850 USD!!! (I couldn't pass on this opportunity)) Please keep up the work. Thank you once again!
My experience on all cameras I have shot is that you need to expose correctly in camera. The ISO control on the Panasonic camera in standard situations (no extended ISO or super high ISO) controls analog gain. Increasing gain improves SNR although it also reduces dynamic range. Digital gain increase in your software does not have the same effect and at best gives you 1-2 stops on small sensors before it creates a lot of noise. I think shooting at artificially low ISO is one of the worst things you can do for your photography
Sensors like that in the G9 deliberately trade off DR for reduced read noise at higher ISO. It can depend on how it is implemented. On the G9 there is a step around ISO 12800 where the noise drops, but so does the DR. This is not NR, I suspect it is a lower level discriminator i.e. a point on the histogram to the left of which is mainly noise (ideally this is set by looking at signal to noise ratio). As the gain (ISO) is turned up so the noise is dragged to the right. The LLD needs shifting but in doing so cuts off some signal, the DR is reduced. The hardware and firmware in the camera is tailored and tuned to it and can often be better than the usual suspect computer software at doing the job. If the sooc jpg is bad, the raw will be too and both want binning.
@@jeffslade1892 no at 12800 the analog gain stops and you have digital bit multiplication. It doesn’t really matter much as that is security camera domain
@@jeffslade1892 it is what is happening it is written right into the image exif file. The analog gain stops at 12800 and then it goes with a multiplier. Most panasonic models have the same behaviour. It is not a mystery use exiftool and it says it there. This is the reason why dynamic range at 25600 is the same as 12800 but IQ is crap
I never considered keeping the ISO at one value, underexposing and then fixing the exposure in post. I do street photography and several times I have missed a critical moment because I keep changing the ISO to avoid motion blurrr. I will try this out this weekend!
Good episode. Now we need a Peter Forsgard test on Olympus. 😉😀 The magenta colour cast may be specific to the raw processing software. I experienced pink recovery highlights from more than one raw software with files from cameras. There must be unexpected changes to linearity behaviour to the relevant colour channels. Did you try adjusting the sliders in the Calibration Panel?
Thanks, my friend. I used Lightroom and did my best to neutralize the magenta, but it was not easy. You are right, that the results may vary in other applications.
@@mattisulanto I use Affinity Photo, and the magenta cast from my G9 RAWs is very strong, esp. over dark colours. So it sounds like the problem is not software specific.
Interresting. It seems Olympus' raw preserves more highlights (but less shadows) than Panasonic. So with Olympus I tend to ETTR (overexpose) to preserves shadows (without noise) and recover later the highlights (in-camera raw processing).
Okay, experiments never lie. Given this subject, however, we should also discuss the so-called "exposing to the right" (ETTR) technique which postulates exactly the contrary of what has been shown here. What I have learned in this lecture anyway, is how to solve the highlight extinction problem using a lower ISO. Thank you very much for demonstrating this!
Thanks. I have a video about ETTR also and it's not contradictory to this video. When you expose for the high lights, you of course expose those high lights as bright as possible without any loss of detail.
@@mattisulanto I was irritated by the fact that "ISO + 3 steps" should show a histogram in the camera where everything seems shifted to the very left side --- that appeared to me like a pseudo-contradiction to ETTR. I am surprised that RAW processing can accomplish such a decent amplification shown by your samples. Very interesting!
@@christophbodner6518 If you have bright light(s) in a dark scene most of the histogram is on the left, but you should still see the bright light part on right side of the histogram. If you look at the dark photo of my desk in the video, you can see the the histogram goes all the way to the right. Anyway, these experiments are interesting indeed. Thanks for watching😀
I would think the magenta cast is dependent upon the software used to process the raw images. Have you tried pushing the same exposures with different software?
@@mattisulanto I have photos that are happy accidents: non-intended camera shutter was activated. and the interest value of a lot of those non-intended are amazing, probably because it's something I would not normally take a image of, and the quality is sometimes ...MeH... but the overall image overwhelms the "poor" quality of the photos; and providing the image with a "feeling" rather than a super-sharp, well composed image. I have been discovering lately the world of "wet-plate" and "pin-hole" photography, and medium-format camera loaded with INSTAX Instamatic film, talk about image quality reduction, but with mass "feeling" appeal. Is it really the sharp images, the deep rich colors, the super fine paper; or can an image be grainy, out of focus, sepia tone, or B&W and still be an image captured that pushed the physical image to reproduce a human arousal response?
Great comparisons. I have both the G9 and Gx9. The G9 is better at 6400 than my 5Dmkii! Also, could just be me, but I've noticed that in general, lightroom does not handle either camera well with regards to noise. DXO Photolab does a much better job at processing the images at high / higher ISO. I think it is due to more than just their prime and Deep prime noise reduction technology. Lightroom has also been shown to not handle Fuji images as well as Capture 1 does either. I try to get the exposure right in camera (I think most photographers that started life in film do this). I have never experienced the magenta cast issue. I have seen it with poor quality filters on FF cameras and wonder if with the G9 this is possibly due to a particular lens or copy of a particular lens? Or even due to something with Lightroom. Can the magenta issue be recreated with other processing software?
i am interesting in lumix g9 and lumix 12-35 f2.8 or olympus zuiko 12-40 f2, both lens are pro, can i use this camara (lumix g9) in wedding photography, and night wedding photography?
I think they know about it, and this is how thay fix the yellow color skin tones from gh5 and older models. I shot the g9 white balance towards magent plus and love this magenta - pinkish tones. And this is way better than ugly yellow thing, that there is no way to get rid of it on the gx80.
Hi Matti, How does this correspond to the concept of ETTR, you've promoted once - and what I apply (low ISO/longer exposure) whenever possible? Thanks.
This is not contradictory to ETTR. Look at my dark exposure of my desk in the video. The bright part (the lamp) is exposed to the right, but the rest of the scene is so much darker, that it becomes almost black.
Hi Matti. Very interesting topic. I have a G85 which has HDR. In a high contrast situation using a G85, would you recommend shooting below 2 stops and fixing in post, or shoot on stop using HDR? Thank you!!
How you expose depends on your situation, but like I said in the video, you should try to nail the exposure in camera. In my opinion, that is. The HDR is for jpeg only and may work, if you like jpeg.
@@mattisulanto Well , Sir. I read the part of extending iso sensitivity , but can't click on it becuase it is disabled (grey), The manual says the only way this option is not available is for 4k recording, which I'm not doing; I am in the manual picture mode.
I try and get everything sorted in camera... However, there are the occasional shots that demand that exposure to the right... expose for the highlights... But I haven’t experienced any magenta cast from my G9.. Maybe I haven’t pushed post process to the limit.. interesting thought anyway Matti...👋🏻👍🏻
Great video by the way!! I recently did some testing with my G9 using flash and 60mm. Some images were underexposed but never saw the magenta when pulled back the raw file using Affinity Photo. I'll have to test for the magenta issue exclusively I guess. While we are on the subject of ISO, why does the G9 have the ISO 200 as a default where you have to change the settings to allow it to use ISO 100? Is there a particular reason? Is 200 better for the G9? Thank you.
Thanks. Panasonic has decided that ISO 200 is the best base setting for that sensor. There is not much reason to use lower settings, unless you have to for some reason.
On my S1R, I didn't do scientific tests because I don't really care, but I use auto ISO with 6400 being the upper limit for color, and 12800 for black and white. The noise at 12800 isn't distracting for me as it still retains a lot of details and looks pretty nice to my eyes (you can see an example here on my website (with the Sigma 45 2.8 lens): yoannlee.com/wp-content/uploads/2021/01/PYLE0151.jpg). In fact, even in color, if I really need more than 6400, I won't hesitate too much to push to 12800. Most of the time, it works good enough. All my thoughts are when using Capture One with the default settings for noise reduction and sharpening. Didn't play too much with those settings as I found that they work well for me by default. Also, I didn't try more than 12800 yet...
@@mattisulanto Exactly. I was surprised and impressed that even with 47MP, high ISO is quite usable with a lot of details and a noise pattern not too "digital"...
Iso or post cannot reveal information that isn't there. So I expose based on the content of the photo. If necessary I can compress or stretch the exposure in post.
Thanks, very interesting test ! What do you mean by "Highlight roll-off" ? I have a G9, and i am deceived about that magenta tint problem on a body i was considering near perfect. I often use high ISO like 3200 to 12800, and since i shot RAW and denoise with Photolab 4 Deep prime, i get good result.
@@mattisulanto Strange, i redo some of your test with my G9, and i found that when i push up the exposition in post 3 stops, i get that magenta colour cast, but when i try to push it 4 stops, the colour cast disappear completely, maybe even a little the opposite colour drift. I will do more test to see what will happen.
@@mattisulanto like on the s1h and the s1 and the s5 have dual iso in video so how does it work with photos or does it? what has been your best results for cleanest image in low light not doing long exposures . and are they iso invoice?
@@ericsamuels6250 My understanding is that the dual ISO is not the same for photo and video, but I have not investigated it. I find that the S1and S1R are so good I don't have think about the ISO too much.
Idea for the video: is it possible to take a good picture using GX9 at 6400, 12800 or even 25600 ISO? P.S. I'm using the gx9 and want to say that it is almost impossible to take a good clear picture at night at ISO more than 3200. Personally I try to never use ISO value more than 1600 with my GX9. I have an instagram account and I am doing a lot of night photography, the ISO 1600 is the best choice I think in terms of editing possibilities.
I wonder about night sky (stars, meteors, also aurora borealis) lowest possible noise topic for lumix cameras...Mine is GX80..please consider to elaborate om this subject, thank you
Thanks for the video but I am not sure about your conclusion for high contrast scenes. You showed that using the same aperture and speed and altering only the iso, may make a difference in these scenes, but you said clearly that the total light hitting the sensor will be the same for both iso values, I guess the only difference is that something changes in the raw file for higher isos. But then in your conclusion you say "expose for the highlights" which may mean reduce the overall light hitting your sensor, which can be achieved with many methods, e.g. by using a higher speed. I think what you really mean is only to underexpose using the iso value only, not to do it through aperture or speed. Is my understanding correct? Because reducing the amount of light hitting the sensor in general is not supported by your tests. Thanks again.
To protect the highlights you can't over expose them. Maybe I'll make another video in the future where I'll explain this further. Thanks for watching.
@@mattisulanto I understand that. My point is that according to your tests we should underexpose using only iso, but in any case I look forward to the future video. Thanks for all the interesting videos you post, I always enjoy watching them.
@@Ad-eq3cu Underexposure is underexposure no matter how you do it, but in post you can't increase the amount of light. You can only brighten the picture, which is the same as increasing ISO.
I think you are missing my point. You explained that iso underexposure is not a reduction of the light hitting the sensor, it is just a reduction of the gain. On the other hand closing the aperture or increasing the shutter speed leads to actual reduction of the light hitting the sensor. This is what I try to get at, that one should reduce iso but not reduce overall light hitting the sensor which is what you showed.
@@Ad-eq3cu That is true, but if you brighten the pictures in post you can only use ISO, kind of. And because of that, in post it doesn't matter how you made the underexposure. Of course if you used low ISO you'll get cleaner results.
I will take a guess(!) where the magenta shift is caused by the special anti-reflective coating of the sensor which sounds to be a unique feature of the G9 - as mentioned by Lumix Ambassador Shiv Verma in this presentation of the G9: th-cam.com/video/jweovXkNslo/w-d-xo.htmlm25s The sensors of the G9, G90 and GX9 are supposed to be the same (it is mentioned over again in tech videos). Of course the G9 might have different processing of the RAW data, but here you have a distinction made by a Panasonic associate.
I took an underexposed picture of my G9 and pushed it 4 stops. I could not see for the life of me any colour shift, no magenta. I used Darktable for that. Maybe it doesn't happen all the times. I had a green moon once only as well.
@@jeangoupil8129 I didn't push it to 4 right away, but basically in several steps including 3 and observed the changes, which were none regarding colour.
@@mattisulanto I just bought the G9 at Christmas, so I played a lot with it and have a lot of different testshots. So I tried a couple more in darktable and also in RAWTherapee. Luckily, I cannot get that colour shift. Isn't that sensor used for all MFT cameras now? Would be wierd, if it happened only with the G9.
Hello friend, As an ambassador please help me to solve this problem? The image is like this, with digital scratches, with your experience, what could be the problem? Thank you very much a hug
Hi Matti , I tried this with my GX880 1 image @ 6400 ISO and 1 @ 800 it seems to me that the problem with the magenta in the image @ 800 ISO is more to do with the white balance , how ever I still think the 6400 ISO image looks better and better still after using DeNoise AI . Thanks for the video interesting as always .
You're absolutely right that underexposing RAW is superior to exposing correctly. So if you shoot only RAW, then you should do this. The bonus, as you have shown, is that you preserve more DR. But, I like JPEGs because it means almost zero time processing. So, alas, I have to adjust ISO. This will not work in cameras with ISO variance, though. And, it pays to know if your camera has dual ISO. I believe the S1 has a second base ISO of 4000.
Interresting video. I appreciate your introduction because many beginners photographers think that with high Iso, they will get a very clear and bright picture, although the subject is dark. As I always say : "in photo, dark is still dark. If your subject is not really visible with your eyes, don't take the picture"
Thanks.
Excellent experiment and analysis. The thought process you outlined here applies to all cameras, not just Lumix. This is very useful stuff.
Thanks. Yes absolutely the same principle applies to all cameras, but in the video I tested only Lumix.
Very informative video, thanks. I use Auto ISO, but cap it at the level where noise starts to appear, usually 3200 for M43, maybe 1600 for older cameras. If the shot is underexposed, then I adjust in post. There are certainly exceptions as you note. If there is high contrast and I am specifically trying to keep detail at one end or the other, I'll adjust for that. Otherwise, Auto ISO with a cap works pretty well for me.
Thanks for sharing your workflow.
Great comparison! I own the GH series cameras plus I've had the S5 for a month now. One thing I learned early with the MFT lineup is NEVER OVEREXPOSE! I generally try and stay about a stop under as my rule. The FF S5 on the other hand from many of the reviews I've watched seemed to handle overexposure better in the reviews I saw on TH-cam. I will say that I love the S5 low-light capabilities, but when pushed even a stop over, the same thing happened, I lose important details in the highlights I cannot bring it back as you showed herein. So I agree completely with your findings. Doesn't matter what format, shooting images or video, -1 stop is where it's at!
Thanks for sharing.
I didn’t know about the magenta shift on the G9 hmm 🤔 good to know now
Matti; It's only a personnel preference but I prefer using the camera settings to get the best picture. I feel more like a photographer when I do this as oppose to using post production software which seems less like photography to me. I'm old fashion I suppose and I am missing some improvements that post processing can provide but I'm still smiling when I get a great photo!
Thanks. Yes, I'm like you. I like to see in camera what I do most of the time.
I enjoyed the video! It's good to learn new things every day. In similar studio situations, I was a bit reluctant to apply ETTR to the actual light source but after this video, I am going to expose to the light source and get more details from it.
Regarding ISO settings, mine vary depending on my activity. If it's landscape, studio, astro and cityscape, I get the best ISO I can in-camera. For street with a lot of moving subjects, I use auto-ISO. At night, I adjust my Auto-ISO depending on the light sources. If the streets, are well-lit, I will adjust the maximum to 6400 ISO but if the lights sources start to become weaker and weaker, I boost the max to 25,600. I still want to get my moving subjects as sharp as possible. For bird and wildlife, I manually select the lowest ISO possible but if the light from the sky varies greatly, I use Auto-ISO, as well, with a lower max. Cheers.
Thanks for sharing your workflow.
Interesting tests Matti, thanks. I usually shoot my G9 about one third to two thirds over exposed as I find that I can still pull back the highlight details very we’ll which benefits the shadow too. I will have to give this under exposure method a try and see how it goes.
Thanks Matti!
Thanks. Try it, but it really makes sense only in high contrast scene.
I've noticed when I push lower ISO in post that my GX85 has almost too much saturation, but other than a simple fix on the saturation slider, this confirms my thoughts for getting the best dynamic range tonality from these micro 4/3 sensors. I'll have to try it as well on my OM-D E-M1 MkII!
Thanks!
On G9, there really is such a feature, when pulling from the shadow +2 and more stops, dark colors are painted in magenta. You need to expose correctly and not be very afraid to raise ISO to 1600 :)
Thanks.
Thank you for this video. I have been experimenting with changing the ISO for pictures and video. In my opinion keeping the ISO as low as reasonably possible is best for preserving your best photo. However, getting the detail may cause overexposure/underexposure in one area or another, which can be addressed in post. Sometimes you have to work with what you have or abandon the shot. I found on the G9 that ISO 3200 is very good in low light, and 6400 is useable. Above that it depends on where I need to use the photo, example Facebook. As we get better at sensor quality and ability to record more light waves as it is being shot.
Thanks for sharing.
Really interesting video - thanks. 😀
I have only encountered the G9 colorcast issue twice. Both were indoor shots where I wanted to keep details in the landscape through the windows while bringing up the indoor shadows. I was able to correct in DXO PL4, but it took some trial and error. I guess that's the price we pay for better high ISO noise performance on the G9 relative to the GX9 and G90. I use the G9 like you would use a Canon 5DIII, letting the auto ISO do it's thing gets you close, then push or pull highlights and shadows to taste.
Thanks. That magenta effect is not an issue in real life, in my opinion.
I've not experienced the magenta cast to which you refer with my G9. I generally use the zebras to expose for the highlights as I know I can more easily recover shadow detail in post (although there's a bit of latitude at the high end as well). Since I got DxO PhotoLab 4 Elite with its amazing DeepPRIME noise reduction, high ISO shots are no longer a concern (there goes another supposed FF advantage). In low light situations, I just set aperture and SS to suit the scene and use auto-ISO, knowing I can get the best possible image in post. It all seems just too easy now compared to film. Of course, there are cases where HDR bracketing may be the best option, but that's not often the case for me. Also, a bit of grain never hurt anyone and in fact I add some now and then as it can look more "natural" and pleasing than a too-tightly-rendered image (like when you see those 4K TVs in the store--the images are often so "harsh" to the point of being annoying). If I want sharp, I can get razor-sharp--it's just not always called for.
Thanks for sharing your thoughts on this.
Yes, I think it was Tony Northrup that stated that pushing underexposed images was exactly the same as using a higher ISO. It seems your observation of improved highlights is what he omitted to notice.
Thanks. It depends on the camera how much you can push the exposure in post.
Great content. Thank you!
Thank you.
The magenta cast on the G9 occurs randomly. In a 300 image timelapse about 30 have this magenta shift. Has nothing to do with dynamic range. It's just a glitch in the G9.
This makes sense as the GX9 has the exact same sensor but seems to not suffer from the problem as much. This glitch seems software related rather then hard backed in the camera.
Interesting idea. Will have to test this on my GX7. Thanks
Thanks. Yes, try it and see what happens.
Tony northrup has an interesting video showing that it is lack of light, not iso per se that causes noise. Reducing the aperature or fast shutter speed to limit light will cause noise too.
Yes, underexposure is underexposure no matter how you do it.
Thank you for your videos! they are very detailed and very well put together!
I am new to the photography and your videos has helped me a lot.
I bought the G95 (with the 12-60 mm F3.5-5.6 kit), also bought the 25mm/F1.7 and lastly the Leica DG Nocticron (brand new for $850 USD!!! (I couldn't pass on this opportunity))
Please keep up the work. Thank you once again!
Awesome, thank you!
My experience on all cameras I have shot is that you need to expose correctly in camera. The ISO control on the Panasonic camera in standard situations (no extended ISO or super high ISO) controls analog gain. Increasing gain improves SNR although it also reduces dynamic range. Digital gain increase in your software does not have the same effect and at best gives you 1-2 stops on small sensors before it creates a lot of noise. I think shooting at artificially low ISO is one of the worst things you can do for your photography
Thanks for sharing your experience.
Sensors like that in the G9 deliberately trade off DR for reduced read noise at higher ISO. It can depend on how it is implemented. On the G9 there is a step around ISO 12800 where the noise drops, but so does the DR. This is not NR, I suspect it is a lower level discriminator i.e. a point on the histogram to the left of which is mainly noise (ideally this is set by looking at signal to noise ratio). As the gain (ISO) is turned up so the noise is dragged to the right. The LLD needs shifting but in doing so cuts off some signal, the DR is reduced.
The hardware and firmware in the camera is tailored and tuned to it and can often be better than the usual suspect computer software at doing the job. If the sooc jpg is bad, the raw will be too and both want binning.
@@jeffslade1892 no at 12800 the analog gain stops and you have digital bit multiplication. It doesn’t really matter much as that is security camera domain
@@Interceptor121 no, not according to the charts I've examined, nor in my experience as radiometrologist. That is not what is happening at all.
@@jeffslade1892 it is what is happening it is written right into the image exif file. The analog gain stops at 12800 and then it goes with a multiplier. Most panasonic models have the same behaviour. It is not a mystery use exiftool and it says it there. This is the reason why dynamic range at 25600 is the same as 12800 but IQ is crap
I never considered keeping the ISO at one value, underexposing and then fixing the exposure in post. I do street photography and several times I have missed a critical moment because I keep changing the ISO to avoid motion blurrr. I will try this out this weekend!
Try it, if you shoot raw😀
Good episode. Now we need a Peter Forsgard test on Olympus. 😉😀
The magenta colour cast may be specific to the raw processing software. I experienced pink recovery highlights from more than one raw software with files from cameras. There must be unexpected changes to linearity behaviour to the relevant colour channels. Did you try adjusting the sliders in the Calibration Panel?
Thanks, my friend. I used Lightroom and did my best to neutralize the magenta, but it was not easy. You are right, that the results may vary in other applications.
@@mattisulanto I use Affinity Photo, and the magenta cast from my G9 RAWs is very strong, esp. over dark colours. So it sounds like the problem is not software specific.
Interresting. It seems Olympus' raw preserves more highlights (but less shadows) than Panasonic. So with Olympus I tend to ETTR (overexpose) to preserves shadows (without noise) and recover later the highlights (in-camera raw processing).
I like this channel :) have a gh5. Excellent JPEG control even with Iso 6400
Thanks.
Okay, experiments never lie. Given this subject, however, we should also discuss the so-called "exposing to the right" (ETTR) technique which postulates exactly the contrary of what has been shown here. What I have learned in this lecture anyway, is how to solve the highlight extinction problem using a lower ISO. Thank you very much for demonstrating this!
Thanks. I have a video about ETTR also and it's not contradictory to this video. When you expose for the high lights, you of course expose those high lights as bright as possible without any loss of detail.
@@mattisulanto I was irritated by the fact that "ISO + 3 steps" should show a histogram in the camera where everything seems shifted to the very left side --- that appeared to me like a pseudo-contradiction to ETTR. I am surprised that RAW processing can accomplish such a decent amplification shown by your samples. Very interesting!
@@christophbodner6518 If you have bright light(s) in a dark scene most of the histogram is on the left, but you should still see the bright light part on right side of the histogram. If you look at the dark photo of my desk in the video, you can see the the histogram goes all the way to the right. Anyway, these experiments are interesting indeed. Thanks for watching😀
I would think the magenta cast is dependent upon the software used to process the raw images. Have you tried pushing the same exposures with different software?
Good point. It didn't occur to me to try another application, which I have to do and see what happens.
@@mattisulanto What happened?
thanks! I appreciate comparisons that are matched images; it's difficult for me to see results from comparing different images to each other.
Thanks. You need to watch in 4k on a big screen. However, this also proves, that the differences are not that big.
@@mattisulanto I have photos that are happy accidents: non-intended camera shutter was activated. and the interest value of a lot of those non-intended are amazing, probably because it's something I would not normally take a image of, and the quality is sometimes ...MeH... but the overall image overwhelms the "poor" quality of the photos; and providing the image with a "feeling" rather than a super-sharp, well composed image. I have been discovering lately the world of "wet-plate" and "pin-hole" photography, and medium-format camera loaded with INSTAX Instamatic film, talk about image quality reduction, but with mass "feeling" appeal. Is it really the sharp images, the deep rich colors, the super fine paper; or can an image be grainy, out of focus, sepia tone, or B&W and still be an image captured that pushed the physical image to reproduce a human arousal response?
@@mindseyeproductions8798 True, good content or good feeling in the picture is much more important than the technical quality.
Will try It out with my good old Lumix G6!
Have fun and let me know what happened.
Great comparisons. I have both the G9 and Gx9. The G9 is better at 6400 than my 5Dmkii! Also, could just be me, but I've noticed that in general, lightroom does not handle either camera well with regards to noise. DXO Photolab does a much better job at processing the images at high / higher ISO. I think it is due to more than just their prime and Deep prime noise reduction technology. Lightroom has also been shown to not handle Fuji images as well as Capture 1 does either.
I try to get the exposure right in camera (I think most photographers that started life in film do this). I have never experienced the magenta cast issue. I have seen it with poor quality filters on FF cameras and wonder if with the G9 this is possibly due to a particular lens or copy of a particular lens? Or even due to something with Lightroom. Can the magenta issue be recreated with other processing software?
Thanks for sharing your experience.
i am interesting in lumix g9 and lumix 12-35 f2.8 or olympus zuiko 12-40 f2, both lens are pro, can i use this camara (lumix g9) in wedding photography, and night wedding photography?
If you intend to shoot a lot of very low light events on MFT camera, I recommend high speed prime lenses with a max aperture around F1.4-1.2.
How does Panasonic respond to your finding of "magenta colour cast"?
I have no idea how they will respond. However, I think they know about the magenta effect already.
I think they know about it, and this is how thay fix the yellow color skin tones from gh5 and older models. I shot the g9 white balance towards magent plus and love this magenta - pinkish tones. And this is way better than ugly yellow thing, that there is no way to get rid of it on the gx80.
Hi Matti,
How does this correspond to the concept of ETTR, you've promoted once - and what I apply (low ISO/longer exposure) whenever possible? Thanks.
This is not contradictory to ETTR. Look at my dark exposure of my desk in the video. The bright part (the lamp) is exposed to the right, but the rest of the scene is so much darker, that it becomes almost black.
Hi Matti. Very interesting topic. I have a G85 which has HDR. In a high contrast situation using a G85, would you recommend shooting below 2 stops and fixing in post, or shoot on stop using HDR? Thank you!!
How you expose depends on your situation, but like I said in the video, you should try to nail the exposure in camera. In my opinion, that is. The HDR is for jpeg only and may work, if you like jpeg.
Hello, how can I select iso 100 with a lumix G7 in video mode?
I'm afraid I don't know, because I don't have a G7. Maybe the manual knows?
@@mattisulanto Well , Sir. I read the part of extending iso sensitivity , but can't click on it becuase it is disabled (grey), The manual says the only way this option is not available is for 4k recording, which I'm not doing; I am in the manual picture mode.
@@mattisulanto Yeah, that's the problem I read the manual but can't figure it out still. But thanks.
I try and get everything sorted in camera... However, there are the occasional shots that demand that exposure to the right... expose for the highlights... But I haven’t experienced any magenta cast from my G9.. Maybe I haven’t pushed post process to the limit.. interesting thought anyway Matti...👋🏻👍🏻
Thanks. I think you can't see the magenta effect in any normal shooting.
Great video by the way!! I recently did some testing with my G9 using flash and 60mm. Some images were underexposed but never saw the magenta when pulled back the raw file using Affinity Photo. I'll have to test for the magenta issue exclusively I guess.
While we are on the subject of ISO, why does the G9 have the ISO 200 as a default where you have to change the settings to allow it to use ISO 100? Is there a particular reason? Is 200 better for the G9? Thank you.
Thanks. Panasonic has decided that ISO 200 is the best base setting for that sensor. There is not much reason to use lower settings, unless you have to for some reason.
@@mattisulanto thanks for the explanation, much appreciated. I always wondered about that.
On my S1R, I didn't do scientific tests because I don't really care, but I use auto ISO with 6400 being the upper limit for color, and 12800 for black and white.
The noise at 12800 isn't distracting for me as it still retains a lot of details and looks pretty nice to my eyes (you can see an example here on my website (with the Sigma 45 2.8 lens): yoannlee.com/wp-content/uploads/2021/01/PYLE0151.jpg).
In fact, even in color, if I really need more than 6400, I won't hesitate too much to push to 12800. Most of the time, it works good enough.
All my thoughts are when using Capture One with the default settings for noise reduction and sharpening. Didn't play too much with those settings as I found that they work well for me by default. Also, I didn't try more than 12800 yet...
Thanks. With the S1/R you really don't have worry about ISO too much, because the image quality and noise are so good.
@@mattisulanto Exactly. I was surprised and impressed that even with 47MP, high ISO is quite usable with a lot of details and a noise pattern not too "digital"...
i want to buy g85.What is the best buying with kit lens or g85 body witjh 25 f1.7 combo?
If you want a zoom, buy the zoom kit and if you want a prime, buy the prime kit.
Iso or post cannot reveal information that isn't there. So I expose based on the content of the photo. If necessary I can compress or stretch the exposure in post.
Wise words😀
What picture profile you shoot this video?
I'm pretty sure it was the 10-bit Log profile, which I graded in post.
You haven't seen my desk
Then again you cannot see my desk
Thanks, very interesting test ! What do you mean by "Highlight roll-off" ? I have a G9, and i am deceived about that magenta tint problem on a body i was considering near perfect. I often use high ISO like 3200 to 12800, and since i shot RAW and denoise with Photolab 4 Deep prime, i get good result.
Thanks. I mean how softly the white graduates to darker tones.
@@mattisulanto Strange, i redo some of your test with my G9, and i found that when i push up the exposition in post 3 stops, i get that magenta colour cast, but when i try to push it 4 stops, the colour cast disappear completely, maybe even a little the opposite colour drift. I will do more test to see what will happen.
I wonder if using the high res mode would make a difference for the results.
High Res mode will suppress noise. If you have a compatible Lumix, please test it and tell me what happened.
does the duel iso have any effect with it with the full frame cameras?
It does, depending how the dual ISO is implemented.
@@mattisulanto like on the s1h and the s1 and the s5 have dual iso in video so how does it work with photos or does it? what has been your best results for cleanest image in low light not doing long exposures . and are they iso invoice?
@@ericsamuels6250 My understanding is that the dual ISO is not the same for photo and video, but I have not investigated it. I find that the S1and S1R are so good I don't have think about the ISO too much.
Idea for the video: is it possible to take a good picture using GX9 at 6400, 12800 or even 25600 ISO?
P.S. I'm using the gx9 and want to say that it is almost impossible to take a good clear picture at night at ISO more than 3200. Personally I try to never use ISO value more than 1600 with my GX9. I have an instagram account and I am doing a lot of night photography, the ISO 1600 is the best choice I think in terms of editing possibilities.
Thanks for the idea. Let me think about it for a while😀
I wonder about night sky (stars, meteors, also aurora borealis) lowest possible noise topic for lumix cameras...Mine is GX80..please consider to elaborate om this subject, thank you
I may consider that in the future, but I'm sure there are already many tutorials on that.
Does anyone know what LUMIX FZ2500 native ISO?
Check out DPReview, they usually have a lot of details in their reviews.
VERY GOOD VİDEO THANKS................
Thank you too.
Thanks for the video but I am not sure about your conclusion for high contrast scenes. You showed that using the same aperture and speed and altering only the iso, may make a difference in these scenes, but you said clearly that the total light hitting the sensor will be the same for both iso values, I guess the only difference is that something changes in the raw file for higher isos. But then in your conclusion you say "expose for the highlights" which may mean reduce the overall light hitting your sensor, which can be achieved with many methods, e.g. by using a higher speed. I think what you really mean is only to underexpose using the iso value only, not to do it through aperture or speed. Is my understanding correct? Because reducing the amount of light hitting the sensor in general is not supported by your tests. Thanks again.
To protect the highlights you can't over expose them. Maybe I'll make another video in the future where I'll explain this further. Thanks for watching.
@@mattisulanto I understand that. My point is that according to your tests we should underexpose using only iso, but in any case I look forward to the future video. Thanks for all the interesting videos you post, I always enjoy watching them.
@@Ad-eq3cu Underexposure is underexposure no matter how you do it, but in post you can't increase the amount of light. You can only brighten the picture, which is the same as increasing ISO.
I think you are missing my point. You explained that iso underexposure is not a reduction of the light hitting the sensor, it is just a reduction of the gain. On the other hand closing the aperture or increasing the shutter speed leads to actual reduction of the light hitting the sensor. This is what I try to get at, that one should reduce iso but not reduce overall light hitting the sensor which is what you showed.
@@Ad-eq3cu That is true, but if you brighten the pictures in post you can only use ISO, kind of. And because of that, in post it doesn't matter how you made the underexposure. Of course if you used low ISO you'll get cleaner results.
I will take a guess(!) where the magenta shift is caused by the special anti-reflective coating of the sensor which sounds to be a unique feature of the G9 - as mentioned by Lumix Ambassador Shiv Verma in this presentation of the G9: th-cam.com/video/jweovXkNslo/w-d-xo.htmlm25s
The sensors of the G9, G90 and GX9 are supposed to be the same (it is mentioned over again in tech videos). Of course the G9 might have different processing of the RAW data, but here you have a distinction made by a Panasonic associate.
Thanks for you guess. The same sensor doesn't always mean identical sensor assembly and, like you said, the processing can be different too.
I took an underexposed picture of my G9 and pushed it 4 stops. I could not see for the life of me any colour shift, no magenta. I used Darktable for that.
Maybe it doesn't happen all the times. I had a green moon once only as well.
It's possible some apps do a better job than Lightroom.
Did you try pushing it 3 stop ? In my experiment, i have the magenta colour shift when i pushed 3 stops, but none when pushing 4 stops
@@jeangoupil8129 I didn't push it to 4 right away, but basically in several steps including 3 and observed the changes, which were none regarding colour.
@@jeangoupil8129 It seems to depend on the shooting situation and how much you actually under exposed.
@@mattisulanto I just bought the G9 at Christmas, so I played a lot with it and have a lot of different testshots. So I tried a couple more in darktable and also in RAWTherapee. Luckily, I cannot get that colour shift.
Isn't that sensor used for all MFT cameras now? Would be wierd, if it happened only with the G9.
And by "pushed" you mean you underexposed the original and adjusted exposure in post?
Yes, that is an old expression we used back in the film days😀 Push processing means underexposing and overdeveloping film.
Hello friend,
As an ambassador please help me to solve this problem?
The image is like this, with digital scratches, with your experience, what could be the problem?
Thank you very much
a hug
I'm sorry, but I don't understand your question.
By the way, I'm not a brand ambassador any more😀
Hi Matti , I tried this with my GX880 1 image @ 6400 ISO and 1 @ 800 it seems to me that the problem with the magenta in the image @ 800 ISO is more to do with the white balance , how ever I still think the 6400 ISO image looks better and better still after using DeNoise AI . Thanks for the video interesting as always .
Thanks. The GX880 might behave differently, because it has the old 16 Mp sensor. Thanks for sharing this.
@@mattisulanto I agree Matti but maybe be the same for the G9 have a try if you have the time .
@@adrianwhareham8921 I tried to correct it, but it's very hard to get perfect.
@@mattisulanto Thanks for trying Matti 👍
You're absolutely right that underexposing RAW is superior to exposing correctly. So if you shoot only RAW, then you should do this. The bonus, as you have shown, is that you preserve more DR.
But, I like JPEGs because it means almost zero time processing. So, alas, I have to adjust ISO.
This will not work in cameras with ISO variance, though. And, it pays to know if your camera has dual ISO. I believe the S1 has a second base ISO of 4000.
Just shoot in native ISO. Anything else is electronically charged.
best for me is is0 400 or 200
Great information, thank you!
Glad it was helpful!