Canon R3

[Fast Glass]

I don't think you really understand how that is not relevant in this comparison.

That only works at full resolution, which doesn't translate into final output size. All that is telling you because of the higher res you are getting more noise. No disagreement there. But it's not super helpful in this comparison. My point there is not a lot of loss in efficiency with adding extra pixels, at least not in the 45 to 50mp realm. Modern tech has minimized this significantly.

As far as the FPS and resolution are concerned. Sony was able to do 30fps, with some caveats, at 50mp. Remember this is being touted as a sports camera, to me this is the new standard for a pro sports camera in this price point. 24 is just not being very competitive in my opinion. I'm sure the R3 is an absolute monster in every way, but even with using the 1Dx III. That was my major gripe, resolution. Even though I loved that camera to death in every other way. I was grabbing it more often than the 5D IV even though it has more resolution. I had said before if this had 30mp, I'd be so much happier with this combo and it could actually make it a viable all rounder instead of being as niche as it is.

And if Canon is indeed developing an 85mp camera, that would make sense for a high resolution monster. Fuji already make 100mp offerings.

I realize not everyone has a need for high resolution, but as tech progresses higher quality comes with it and Canon is still living in the past and Nikon and Sony defiantly got this part right.

After seriously looking at the Sony and Nikon offerings. I just can't switch, it's a personal choice just because I can't see myself shooting anything else. Canon gets everything else right. So I'll just hang in there, but I'll be all to happy when they release a 50mp or higher offering for the pro market.

[Kayaker72]

. Modern tech has minimized this significantly.

My last take on this. I assume this is about the noise discussion. I'll phrase it like this, all of these statements are true (specifically about read noise):

• Modern tech has minimized "this" (specifically read noise) significantly
• As an example the R5 has lower read noise than the 5DIII
• The R6 still has lower read noise than the R5 which does lend itself to certain scenarios
• Higher MP cameras tend to have higher read noise than lower MP cameras.

Again, all of those statements are true. The reason why I even brought this up is the exchange you had with Kari about aurora photography. Low light/nightscape images are a scenario where read noise becomes more of an issue, thus there would be a negative impact for Kari's type of photography of more megapixels. For blacks, where signal to noise ratios are low, lower megapixel cameras still show a benefit over higher MP cameras. Now part of that discussion was motion blur, but even that could be effected as if you dial in your ISO to have acceptable noise in the blacks, a higher ISO would allow for faster shutter speed and less motion blur.

The impact for well lit images drops off dramatically, as David's DXO example showed, 3300 vs 3000, not much of a difference. So, if you want to say that "modern tech has minimized this significantly" for noise in well lit scenarios, I am going to agree most of the time. But, for blacks, the Photons to Photos measurements indicate the R6 is about 1 stop better in read noise control which will absolutely affect low light photography.

Can read noise still affect well lit conditions, yes. If you think about it, "expose to the right" (ETTR) was about two things: using the whole dynamic range of a sensor, and avoiding read noise. The left side of a histogram has a lower signal to noise ratio (the 10/3 scenario from above) while the right side of the histogram has a higher signal to noise ratio (the 20,000/3 scenario from above). If you think about it, "expose to the center" would have its benefits as, in ideal conditions, that would give you the most latitude in post processing. But, it is "ETT Right" as the right has a higher signal/noise ratio. Has modern tech minimized this, yes, the noise floor is lower, but it still exists even in some well lit conditions, for example, if you underexpose. I have seen this even with the R5. Quick examples, but a series of birds against snow, I exposed for the snow, tried to pull shadows up. I still really like the images, but there was noise in the birds. Modern tech has not minimized this to the extent I had no issues.

Final point on this, I do have a noise issue with the R5. I am still trying to understand it as it is inconsistent. It isn't awful and I will still praise the R5 as a camera (it is great). Others have seen it as well and think that Adobe is not processing modern Canon files efficiently. But, it is luma noise, which could easily be read noise, and my other theory is that as the camera warms up, you start to see more read noise. So, if it isn't the software, then the high MP of the R5 is contributing to this with both higher read noise (compared to R6), and then more heat caused by moving more data. Even in well lit situations, but testers might not see this as it would only show up during intensive use.

To sum up, I find it misleading to state that there are not trade offs going to higher MPs. I am sitting here with my R5 and have several instances of these trade offs. While I was going to upgrade my computer anyway, the R5 certainly has hastened that decision as my current computer freezes up half the time downloading R5 files. Never an issue with my previous cameras. Resolution increases linearly, file size increases exponentially. Absolutely there are trade offs, with diminishing returns. Different people want and value things differently. The R5 is an amazing camera. I have no regrets, it is worth it, modern tech is amazing, but there are absolutely tradeoffs to higher MPs. There is a balance to be struck.

[Fast Glass]

Again, all of those statements are true. The reason why I even brought this up is the exchange you had with Kari about aurora photography. Low light/nightscape images are a scenario where read noise becomes more of an issue, thus there would be a negative impact for Kari's type of photography of more megapixels. For blacks, where signal to noise ratios are low, lower megapixel cameras still show a benefit over higher MP cameras. Now part of that discussion was motion blur, but even that could be effected as if you dial in your ISO to have acceptable noise in the blacks, a higher ISO would allow for faster shutter speed and less motion blur.

To sum up, I find it misleading to state that there are not trade offs going to higher MPs.

So you are not understanding this correctly at all. There is a stop in noise because there is over double resolution, but not in the real world.

If I were to print both images at 150 DPI, and then say the 45mp image has lots more noise that would be correct. But you are looking at an image that is larger n comparison.

If we both printed them at say 20x30, then you will not notice a huge difference in noise because that is more of a function of the output size and not zooming into all the noise. The 20mp is be slightly cleaner, but nearly like it would be if you viewed them both at 100%.

This is the part where you cannot use the readout noise as a guide. It just doesn't translate here because it's only telling you the readout noise at per pixel level. Not what it would look like at the same output size.

To prove this just take a typical night time shot with lots of deep dark shadows, either print them out on a large print or downres the R5 results to R6 dimensions. You will find there would be only minimal differences in noise.

I didn't say there is no trade off's, but they are not nearly as big as you might think or what just the readout noise would suggest at a pixel level.

To clarify what I said about modern tech minimizing the differences, I'm talking about the efficiency losses associated with adding more resolution. And they have gotten pretty dang good at adding pixels and not wasting light in the process. In a perfect world, we technically could just be adding as much resolution as processing power will allow or till the pixel wells start not being efficient at gathering light. Technically it wouldn't matter at that point. And in the case of 20 vs 45, that is really minimal the efficiency losses and the bigger question is not noise. It's do I need or want the resolution or not. And cost associated with that.

It's not about how many pixels you add, it's about how many you can add without wasting light. And modern tech has done a really good job of that. Besides that, a 50mp image is equivalent to 21mp 1.6 crop camera. It is not that small in the first place, many 1.6 crop cameras exceed this by a considerable margin. So it's not like we are even pushing the limits of sensor tech at a pixel level in the first place.