In this video I compare the sensor read noise of the A7IV, A1 and A7III and explain Sony's "Dual Gain ISO" and the variations between the three cameras.
You’re welcome! In this video I mistakenly said the A7RIV has the same second gain stage of ISO640 but I meant to say the A9II. The A7RIV has a lower gain conversion point, I think it’s ISO320.
Thank you for this info. I had no idea this happens. Will now use iso 80-160 and avoid 200-320 by going straight to 400 on my A74. Can you let’s know if there is a ‘reset’ higher up the range? If so where do we avoid for best results.
Great video. It would be nice if you put your findings in the description of your video. Use ISO 100-125 or a7iv 400, a7iii 640 a1 500, for the most dynamic range and the least amount of noise in your image.
Thank you, this has really made this subject so much clearer. I have A99ii and A7iv and am a bit disappointed with the amount of noise, but this has helped enormously, thank you.
This was an excellent video, thank you! I am still shooting the A99ii, which uses the same sensor as the A7rii and A7riii, but I never knew about this. I have just tested it myself, with no lens attached (just body cap), and the difference between ISO400 and ISO500 is immense! It also means that the difference between ISO800 and ISO320 is almost nothing - such awesome info to have after so long using this system. Thanks a lot for this, I really do appreciate it!
Iso 640 is super clean on the Riii sensor as it is the second gain stage start point. I try to be at 100 or 640 always if it can work for the shot. The sensor gets noisy quick. I personally won't go above 1600 and even then it rubs me wrong sometimes.
Thank you, as an owner of A7iv, A1 and A7C, this is very useful information. I may start setting my ISO manually to second gain, rather than leaving on automatic. I photograph wildlife, and often have to heavily crop.
If I ever use Auto ISO, I do set my lower limit to the start of the second gain stage to avoid those noisy, low dynamic range points just before the high conversion gain stage starts.
WOW WOW WOW.. Thank you for clarifying and showing examples. This is going to help me immensely If i am over ISO 160 on my A73. I can just shoot at ISO 640 at that point for the next gain reset.
Higher resolution doesn’t collect less light, an individual pixel collects less light but there’s more of them, collecting as much light (signal) across the picture. You shouldn’t get much of a difference in noise when viewing the same image a the same size with higher res over similar sensor tech unless you’re cropping etc.
Yes ! The sensor on the a7sIII its a real 48mp in a quad pixel Bayer binding so the output it’s 12mp! But each sub pixel can be set to different parameters, im this way u archive fantastic dynamic range.
If you look at graph from precise low light test, you can clearly see an a7c 24mp perform better than a7iv with 33mp, across all the range but clearly over 640 iso. Clearly test indicates than you shouldn’t go over 3200 iso in a 61mp when a7c can go easily over 6400 iso and even more. So bigger pixel have an importance, not the number of pixels. I prefer my 24mp a7c over a7iv, a7cii and a7rv or a7cr. I think it is best hybrid at reasonable price.
Thanks for this trick! im shooting with apsc and i found how noisy that sensor is. This has been helpful to understand the capabilities of my camera more than ever.
Thanks for doing this. I just had a photoshoot with a large group yesterday and had my A74 at ISO 320 😩and was wondering why it was so grainy! Good to know to just skip to ISO 400. Denoise is helping and a bunch of lightroom work.
Technically, noise does not come from a lack of signal, and you cannot test SNR without signal calibration. When you plug in two different microphones into an audio interface to test their noise floors, you cannot just turn the gain knob to the max and compare that way, because they might have different sensitivities.
Absolutely brilliant analysis! I have been a Nikon shooter for over 30 years and have noticed that images, dynamic range and noise, shift as I move through the ISO range. It would result in what to me seemed counter intuitive results . . . . you have now explained it. Wonderful work! I am now a fan boy subscriber. If you can now explain to me why I have to wait till December 2022 to get my FX6 . . . sigh
@@erikdurnall9585 I have a lot of Nikon glass and have played devils advocate with myself as to whether I should buy the Z9 for my video projects (it still won't show up on my doorstep till end if year). I lose a little on the run and gun benefits like the slick implementation of internal ND. I gain 8K benefits for post. I've shot projects on rented Sony F5 and Canon C100-300ii. Do you have an opinion on Z9?
Erik, I’d like to hear your thoughts on “iso invariance”, meaning (on the A7iv for example) that all ISOs 400 and above are supposedly the same and only differ by exposure compensation. Meaning a shot that would require ISO 3200 via the camera’s metering would have exactly the same noise if taken at ISO400 and bumped 3 stops in Lightroom. If true, I’m not sure how to think about the practical implications except perhaps that one should always err towards under exposure in dark scenes…
Dude this is awesome. Thanks for sharing, this is exactly what I was looking for. I wish I had known the sweet spots for all the years I had the A7iii. I just jumped to the A1 and I will absolutely put this knowledge to use.
I compared the A7IV and A7III read noise data on the famous site photonstophotos. Above ISO 500, the A7IV has 50% more read noise that the A7IV in DN units (power). This is not a big effect, as the noise is small. But it is easily observable. This represents a 0,5 shift in log2(DN) that are used on this graph. I have no explanation for this. The read noise should not depend on the pixel sizes. But the reading electronics might be slightly different. On the other hand, the size of the pixels for A7IV is roughly 28% smaller that for A7III . (33 Mpixels versus 24 Mpixels) Therefore, the "shot noise" for the A7IV should be 14% higher than for A7III. This is because the shot noise is proportional to the square root of the number of photons detected. This is important because in fair illumination, shot noise always dominates over read noise. I think I will try to measure this shot noise. I just realized this should not be too difficult. But I would need a reasonnably uniform illumination of the sensor. This uniformity would allow an easy estimation of the standard deviation (stdev) of the signal over all the pixels. This is a way to measure the noise. I already measured the read noise for my A7III for 37 different ISO numbers. Retrieving these 37 "full dark" pictures and calculating the stdev, I could reproduce exactly the same results as on the site photonstophotos. By myself. Boy! The comparison is really almost perfect. But my first attempt had shown a slight difference at low ISO. I just discovered that evaluatiog the stdev over all the 24Mpixels was not a good idea. Because there are slight variation of the average dark signal. To get this perfect agreement, I simply had to calculate the stdev over a smaller area of 200x200 pixels. And this is what I will also do with a fair uniform illumination to measure the shot noise.
I find the white dots in the dark areas very distracting on iso 6400 and 300% crop. Try shooting that with lens cap on. Lots of white dots everywhere. Your's the same?
A question, is the second gain stage where the "star eater" problem began? Is this still a problem, or are stars in the night sky exempt from the clean-up process? My old Sony a6000 drove me nuts until someone figured out that at a certain point, Sony was doing some sort of image cleanup--even in raw files, that removed many stars from night photos.
7:22 - this is not really accurate I think, not if you mean "restoring all of the dynamic range". The point of the low gain stage is to have the maximum dynamic range, but that means that low-light areas collect less accurate information. When you enter the high gain stage, you get less noise, but in turn, give up some dynamic range. You can think of this as having buckets for collecting the light. A big bucket can collect more light (high dynamic range), but measuring small amounts of light off of them is not as accurate (more noise). At the high gain stage threshold, you switch over to use small buckets, where you have less noise but also lower dynamic range. But increasing the amplification also reduces the dynamic range, and switching to the high gain circuit reduces the amplification, so you will restore *some* of the dynamic range, but not to the level the lowest ISO on the lower base ISO has. Dan Fox did a great job in explaining how the dual base ISO works in his video "ISO on Sony cameras is often INCREDIBLY misunderstood". If you meant that switching to the high gain circuitry restores *some* of the dynamic range, then I was going on rambling about something you already knew - but anyway, perhaps this was not clear to all viewers...
Thank you so much. Isn’t there another sweet spot in terms of ISO versus noise level up around 6400 in the A7IV? I thought I saw some information on that.
You’re welcome! There isn’t another sweet spot for RAW images, you may be thinking of using picture profiles in video. In the Slog3 Picture Profile the base ISO is ISO800 and the second gain stage is ISO3200 but this is the same as ISO100 and ISO400 in RAW still images, it’s just relative to the selected gamma of the picture profile. I wanted to keep this example as basic as possible so I didn’t go into all of that in this particular video.
Really useful video many thanks. I have just recenty purchased a Sony A7iv and a Sony GM 70-200mm f2.8 OSS II and am starting to shoot astrophotography, trying out deep sky objects, so I'm dealing with very low signals and as a result I need long exposures and maximum magnification and maximum cropping of my images. I'm using ISO 400 on the A7iv as this gives me the maximum dynamic range and minimal noise. However, I didn't know that this is where the second stage amplication kicks in. Good to know. Once again many thanks for your very informative video. I really like your stuff as you delve into the details while making it very understandable.
You are very welcome. ISO400 on the A7IV is amazing. The noise is nearly the same as ISO100 and you get a significant amount of dynamic range back as well. Enjoy!
You have just blown my mind! I have never heard of Sony's Dual Gain ISO. I have the 3 and 4 versions of these cameras. Honestly, I couldn't see any different on my phone screen. So, I switched over to my monitor and I still couldn't see much difference on my screen. But, I trust your voice and if you are seeing it, I believe you. I'm shocked! I'm better off with 400 than anything after 160!?! That's kind of important. I need to watch this video again. These details will definitely improve my images and I feel like they should be written down somewhere as the Gospel. Can you make a list of where all the range jumps are through the spectrum? Please don't make me do this experiment myself. Thanks for this eye-opening comparison.
You’re very welcome. Make sure you watch it on a better screen, set the playback resolution to 4K and turn the brightness all the way up on your screen. There’s only one gain conversion point for each camera and the A7IV is different from the A7III. The A7IV has much better dynamic range and the noise is relatively the same as the A7III but at a much more useable range in my opinion.
@@erikdurnall9585 Thank you so much. I just watched this again and I'm still blown away. I have researched every aspect of these cameras, but you are the only person I've ever seen talk about this topic. This explains so much about what I'm seeing in my own images. And, you described it so well. Can't wait to see more from you. :D
@@LHazle you’re very welcome! I’m so glad you found it helpful. My purpose for making this video was to try to give a visual example of what is going on, as I find that most people (including myself) find looking at various multi-colored overlayed line graphs to be of very little value. Thanks again!
Knowing the base iso of a camera can be important sometimes, but it's more noticeable when filming. The Fuji xt-4 sits at iso 640, arri Alexa at 800 and so on, and high end models can sometimes have a 'dual' base iso. It's not intuitive either to think that you would get a cleaner image when filming in dim light, if you went up to iso 2000 from say 1600 for some specific cameras.
@@tomastuoma this isn’t about Fujifilm cameras, this is specifically about Sony Full Frame cameras that do have dual base ISO, and it also applies to the relative ISO ranges of the selected Picture Profile gammas in video.
Great video, im looking to upgrade A7III to A7RIII or A7IV and this comparison was pretty interesting, one thing that i would wish to see is comparison on higher ISO's, but nevertheless I still think that Sony A7III holds the crown as the best Sony’s low light camera
AWESOME VIDEO! I've wondered about this. I am a Nikon shooter; however, I ventured into mirrorless with Sony. I have a 6300, A7 II and an A7 IV. I also have a 1.8 85mm Sony lens. I could see all of the noise. I have a really good noise suppression plug in; however, I lose a lot of detail. I want to use my Sonys more for professional shoots, but there is just too much noise. I'm going to take your advice and try and use ISO 100 and/or 400. Thanks for this video!
Great and informative video! Does this apply to video on the a7iik as well? Meaning, what are the best ISOs to shoot (video) on the a7iii ? Thanks in advance
Yes. It is for video as well as all modes. It’s always the same range depending on your base ISO of your selected PP. so if you’re shooting video with no PP (PP Off) then the sensitivity ranges are exactly as demonstrated here for the A7III (ISO100 and ISO640). That’s a 2 and 2/3 stops range. So if you’re shooting video and using a PP that has a base ISO of ISO800, the second sensitivity would be ISO5000, the same 2 and 2/3 stops range.
How about In the older A7 II that are still In stores, I have It, I'm guessing that It doesn't have this sweet spot In the higher ISO at all. I will have to try It out?
Thx great video. I think I will jump 100 to 400 on my a7.4 in future. BUT what is the situation with large compressed raw which I shoot and probably many others please
Thanks for another clear and easy to follow video. I have a question. In a previous video on using auto iso, it was recommended to set the minimum iso at 100 and maximum iso at 12800. Since there is less noise and more dynamic range at iso 400 than at iso 160 on the A7 IV, where do you recommend setting the minimum and maximum auto iso on the A7 IV for wildlife photography? My long telephoto is 6.3 at 500mm. I'll be on a safari in Tanzania soon. We'll be out on the road in the early morning hours as well as the later afternoon. So, it'll be a battle between noise and using the right shutter speed to capture motion. Would you keep the auto iso, minimum iso at 100 for wildlife photography to be able to use a faster shutter speed?
It would be nice if the cameras offered an “indexed” ISO option where you could just switch between either ISO100 or AUTO ISO with a preset range of your choice, but since that isn’t an option it’s probably best just to start your AUTO ISO range at ISO100. The camera produces such great image quality it’s really not that detrimental to the overall image unless you’re really cropping in heavily and pushing the exposure.
Funny was shoot at iso 200. Worried about noise ! Ha. Stuck in crop sensor mindset. Also. But good to here about this on Full frame a7iv. Transitioning. Thank you !
I do wildlife photography with the A7R IV, due to rapidly changing conditions and fast speeds I run auto ISO. In good sunlight I just rock 1/1250 shutter and ISO does its own thing, but on the weekend I took some shots in good light and they were noisy as heck, ISO was sticking around 320. But now I think im going to have to change my shooting style, maybe in strong light I just need to set to ISO 640 and then just adjust shutter if need be. Shutter will be quick enough no matter what so its worth a shot
Unfortunately the A7RIV has a lower gain conversion point then what I state in this video. When I say the A7RIV I actually meant to say A9II. The gain conversion point for the A7RIV is actually ISO320 in RAW stills, so if ISO320 is too noisy then a higher ISO value isn’t going to improve your results.
@@erikdurnall9585 Ah ok thanks for letting me know. Strange because I thought ive also heard from another video that ISO 640 is the cleanest for A7R IV, but like I said ISO was all over the place, looking at the photos from the weekend (of a black bird) they were ISO 320, 400, 500. I guess its just the fact of the high dynamic range of the black bird and bright background
@@erikdurnall9585 I have the A1, A7iii, A7Riii and A7Riv. The A7Riv is the noisiest one by a mile. I mean, it starts off noisy at it's base ISO and just goes up from there. Having said that, I still have no problem shooting interiors professionally with the A7Riv to 1000 ISO or so. It's not that the noise isn't there, because it most certainly is, but rather the practical implications of the noise are a lot less problematic than a lot of folks make it out to be, even for paid work.
Hi Erik, for A1, do we multiply by 4 or 2 1/3 stops increment to go beyond ISO 500? e.g. the next gain is 500X4= iso2000, i.e. other channels propose this formula for videos.
Thank you for sharing your knowledge on this! Would you say the second gain ISO number is just as good as ISO 100 even for shooting outdoors in direct sun? So for example shooting sports action with the A1, bumping up to ISO500 to have more flexibility with higher shutter speeds?
Yes. ISO400 is almost identical to ISO100 on the A7IV. For RAW still images you have a lot of range for very clean images well beyond ISO1600. For video with picture profile Off, it is the same ISO ranges. If you are using a Picture Profile like Slog3, which on the A7IV has a base ISO of ISO800, your second gain stage would start at ISO3200 and you would still be seeing very useable footage at ISO12800
Why would thay do that? Also no information about it.. 1..thabk you. 2. I feel frasutraed needed to close my aperture and use lower iso instead knowing i can use higher iso with same results while keeping more thing i needed in focus
Great info, Erik! Thanks for doing the analysis. However, I’m a little confused since you mentioned A7C also has a second native ISO at 640..is that accurate for shooting RAW images? Also, when shooting video (say, Slog-2) - Would you happen to know what’s the second native ISO stage?
The info I give here is specifically for shooting RAW still images. When using a PP like Slog2 in video, the range will be the same but the numbers will be different so if Slog2 has a base ISO of ISO800 then the high conversion gain stage would start at ISO5000.
@@erikdurnall9585 I see what you’re saying. It seems like it matches up with what Gerald Undone had also alluded to, basically 1/3 to 2/3 of a stop better than A73, around ISO 3200 👍🏼
Unfortunately I do not own an A6400 so I cannot do the comparison myself but my understanding is that the A6400 has a similar high conversion gain stage that starts at ISO400
Yes. It is the same in video except if you are using a Picture Profile the ISO ranges will be relative to the base ISO of the gamma curve that is selected.
@@buffaloray yes, the Slog gammas have a base ISO of ISO800 and the second base ISO is ISO3200. It’s always two full stops. The S-Cinetone gamma has a base ISO of ISO125 so the second base ISO would be ISO500, the same two full stop range.
Wow! This is very helpful. Glad someone shared this link on FB. I'm a 7iii user, so shooting with ISO 640 is just the same with ISO 100 regarding on clean image. If my screen will see it as overexposed, I can still shot with ISO 640 and just adjusting the shutter speed right?
@@andyholdsworthmusicphotogr162 there is not another gain stage after the high conversion gain stage but at ISO64000 and beyond there is noise reduction that is applied.
I think you got it backwards. ISO400 is NOT when the higher gain stage kicks in; it is when the camera switches back to LOW GAIN while tapping the smaller charge capacitor on each pixel. At 100 ISO it uses the big cap optimized for ample light also at low gain. At 320 is when it uses the big cap at high gain which is why the noise is high.
It is very interesting point and push me to explore with my camera. From your experience, which camera is your best choice for photography in low light, A7iii, A7iv, or A1? Thanks.
Honestly they’re all basically the same. If I had to pick one it would be the A7IV because it has the best balance of performance, features and resolution. The A7III is the cleanest but it’s also the lowest resolution and the autofocus isn’t nearly as good as the A7IV or A1 so it really depends what is most important to you.
Thanks for the interesting and unique video. I tested my A7iii and also see a big improvement at ISO 640. Interestingly though, ISO 160 actually looks cleaner than ISO 100, and my ISO 200 looks pretty noisy (unlike on yours). Any idea what would explain better results at 160 than 100?
Somebody correct me if I'm mistaken but: The Dual Iso of Alpha 7 IV goes up to 3200, all the dividers and multiples seem to have very less noise. So 3200/160, precisely, equals 20, so 3200 is a multiple of 160 what means it will have less noise problem
@@theluciano354 I think this dual ISO behavior is different for video and stills. And for videos, it differs from one picture profile to another. Example For S-Cinetone: 125 and 500 For Slog3: 800 and 3200
Erik, I'd be interested to learn of video high noise gain when you need to bump your ISO very high. I've used the Sony 7SIII and found 12,800 is awesome at night in lower light scenarios. That said what is the high range of the 7IV for working with video in low light, similar to the 7SIII? Thanks!
The A7SIII (and FX3) has a much higher gain conversion point than all of the other cameras. The A7SIII gives you 4 1/3 stops of normal gain before switching to the high conversion gain. In RAW stills, or with PP off the A7SIII has a base ISO of ISO80 which means the high conversion gain stage starts at ISO1600. This also applies to video, and carries over when using picture profiles. For example; when using the Slog3 picture profile, the A7SIII has a base ISO of ISO640 which means the high conversion gain stage starts at ISO12800. It’s easy to figure out the second gain stage for the A7SIII as you can just apply the base ISO by 20. So if you have a PP that has a base ISO of ISO100 100x20=2000 so your second gain stage would start at ISO2000. The A7IV is similar in this way as you can just multiply your base ISO by 4 to figure out where your second gain stage starts. In Slog3 your base ISO is ISO800 so 800x4=3200 which means your second gain stage starts at ISO3200 in Slog3. If you’re using the S-Cinetone profile, the base ISO is ISO125, 125x4=500 so your second gain stage starts at ISO500. The A7III and A1 are not as simple as just doing simple multiplication, but the A7III gives you 2 2/3 stops of ISO range and the A1 gives you 2 1/3 stops of range before switching to the high conversion gain stage. And this apples to all of the picture profiles as well. Hopefully that helps!
@@erikdurnall9585 Shouldn't your multiple by be a constant number rather than having a different number for each camera. How did you come up with the multiply by number?
@@thomastuorto9929 most of the cameras have different high gain conversion points, a few have the same. The Multiplicative factors (20 and 4) of the A7SIII and A7IV respectively, are just a convenient coincidence. The other cameras have conversion points that don’t have that convenience.
Erik, thank you so much for this comparison. I understood that you tested the dark channels of several cameras with an 32x (5 stops) amplification. You found the the points where some magic reduction in noise happened, where I guess is a reduction in bandwidth using a digital filter, which is applied (may be ?) to create a second gain stage or whatever magic. That was a real good information you provided for us photographers! Thank you. IMHO the real optical signal is on top of that dark channel noise and is impacted by several digital conversion errors (lin +1 bit uncertainty..) and the number of electrons freed up per photon of the light channel inside the photodiodes. That is the part I would be interest in for noise comparison of the signal channel. Do you have any data about that. Thanks a lot again. KO
Thanks Eric for an informative and visually believable comparison of these gain amplification break points. Are you relatively confident that the same behavior is exhibited when the sensor is processing in it's sweet spot, i.e. a properly exposed image as opposed to the back of a lens cap? So my interpretation for my A1 is, unless I'm shooting long exposure on a tripod at ISO100 or shooting handheld in very bright light where I can get 100 or 160 that I'm better off setting my min ISO range in AUTO ISO to 500 or 640?
Yes. That is correct. Unless you are shooting in ample light and you can effectively use ISO100-160, you’re better off shooting at ISO500 on the A1. If you’re doing wildlife or anything that typically requires higher shutter speeds, you definitely want to start at ISO500.
@@prashaanthdr7124 it’s only different between camera models. It’s always the same range for a specific model no matter what your base ISO is. For example the A7IV is 2 full stops. So if you’re shooting RAW stills and your base ISO is ISO100 then the second sensitivity is 2 full stops above ISO100 which is obviously ISO400. The same if you’re shooting video and using a Log profile where your base ISO is ISO800, the second sensitivity is still 2 full stops above your base ISO of ISO800 which is obviously ISO3200. The range is always the same. For the A1 it’s 2 + 1/3 of a stop, for the A7III, A9, A7RIII, A7C, and A9II it’s 2 + 2/3 stops, for the A7RIV and A7RV it’s 1 + 2/3 stops.
@@erikdurnall9585 I stand corrected, repeated the test righ now. But I had some previous tests with different results, oddly enough. Still have way less "worms" on 3200 compared to 2500, and there's definitely another big increase in quality by choosing ISO 6400 over ISO 5000. Did the latest tests without a lens at all, just camera body plus body cap, perhaps my Capture One activates different automatic enhancements, depending on (no) lens...
Honestly they’re all excellent. If you want the absolute cleanest image then the A7III is the best, if you want to best resolution with the most details then the A1 is the best, and the A7IV is really mix of both of those, it has better resolution and detail than the A7III and cleaner image output than the A1. The A7IV would be my choice.
PROPOSITION : Should we conclude we should make (available) SIMPLE graphic (same canvas) grids (a smartphone app!) for ANY camera, with just as many fields as iso/asa steps (50 to say 200'000). We coukd then color-marker over those fields that represent the camera's levels where a new treatment kicks in. Then: how much backwards steps it replaces. This we could simply for each camera cross over the settings that overlap and should be ignored. (Best would be to be able to eliminate them in the camera's own algorithms of auto-iso option !!!!!!). Maybe with 3 filters, because these "to eliminate/avoid" ranges, will probably differ not only just in regards to color noise, and dynamic range, but perhaps differ in regards of grain effect, too ? Maybe such an app/ (standardized) table exists already ? Perhaps something for DPreview.
Hi Erik, thanks for this very helpful video and the time you put into doing all the comparisons. I recently purchased the A7IV and also own the A7III. I have been noticing some noise on the A7IV as soon as I was shooting above 160. Started to wonder if I was just seeing things or what but I never seemed to have had this with the A7III. So I did the test and compared both cameras. The noise on the A7IV is noticeably worse at every ISO level. So I guess I was not seeing things. I suppose that means that there is factory issue with my sensor and I should contact Sony about it, would you agree?
