Q3 binned at 36MP vs A7C II

[adrianh]

Advertisement (gone after registration)

Didn't Mathphotographer show that the Q3 has an DR advantage at lower DNGs but M11 has not? That would be in line with Jesko was mentioning. 

[SrMi]

1 hour ago, adrianh said:

Didn't Mathphotographer show that the Q3 has an DR advantage at lower DNGs but M11 has not? That would be in line with Jesko was mentioning. 

A better DR is only possible if NR was applied when generating lower resolution. The mentioned differences in histogram and highlight recovery are also weird. I will run some tests later.

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[SrMi]

2 hours ago, Miltz said:

I just did some testing and I can confirm that there is definitely an increase in DR using M-DNG over L-DNG and resizing in Lightroom or photograph. I took 4 different images and all 4 show the exact same results. The L-DNG file is brighter and the histogram is different than the M-DNG file. I was able to recover more highlights and get cleaner shadows with the M-DNG vs L-DNG and then resizing in to the exact same size in photoshop. The histogram and image brightness surprised me but it makes sense. I'm glad I tried this out. What I can't measure is the amount of DR difference but if I had to estimate I would say around 1 stop which math photographer was able to confirm. (I know you don't like his methods but my results confirm his and disprove the resize later theory)

Thank you for running the tests. I will try to reproduce your observations.

[SrMi]

I tested L-DNG vs. M-DNG at ISO 2000 and ISO 6400 (exposure was the same for both resolutions). The camera was mounted on a tripod to guarantee repeatability. The raw histograms were practically the same.

After lifting shadows by one to two stops, I conclude that the noise in the deepest shadows was practically the same in both resolutions.

The results were expected, as I did not think that Leica would apply NR to their raw files.

Here is another test of my library at ISO6400, f/5.6, 1/60 (lifted by 1 stop in LrC). These are screenshots from 200% views in LrC after reducing the size of images to 6000x4000.

L-DNG files are, as expected, a bit sharper (may not be visible in practice).

L-DNG

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

M-DNG

[Miltz]

3.5 stop push

Highlights - 100

Shadows +100

L-DNG resized to M-DNG.

The shot below was completely in shadow as you can see in the image that was not pushed at all. After the image is pushed, the evidence is clear. The colors and detail are much better in the M-DNG. No amount of resizing will recover the better colors and detail in the L-DNG shot. Look at the colors, the box, the vacuum and the floor look so much better. The fine detail is also much better, look at the vacuum air port, you can't even see it on the L-DNG file, the letters b & w are completely gone in the L-DNG file. Keep in mind to be able to see 1 stop difference the images need to be pushed more than just 1 stop. As a matter of fact if you shoot in M-DNG and resize up to L-DNG you'll still have more detail in extreme shadows. For those wondering S-DNG looks identical to M-DNG.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[nicci78]

🍿

[nicci78]

Advertisement (gone after registration)

https://www.dxomark.fr/fr/Cameras/Compare/Side-by-side/Sony-A7IV-versus-Sony-A7R-V___1376_1385

 

looking at print size 60MP performed as good as 33MP sensor. 
At screen size 33MP beats 60MP. 
Once again what’s the point of using 60MP binned permanently at 36MP isn’t it even better to use a native 33MP sensor ? 

[SrMi]

1 hour ago, Miltz said:

3.5 stop push

Highlights - 100

Shadows +100

L-DNG resized to M-DNG.

The shot below was completely in shadow as you can see in the image that was not pushed at all. After the image is pushed, the evidence is clear. The colors and detail are much better in the M-DNG. No amount of resizing will recover the better colors and detail in the L-DNG shot. Look at the colors, the box, the vacuum and the floor look so much better. The fine detail is also much better, look at the vacuum air port, you can't even see it on the L-DNG file, the letters b & w are completely gone in the L-DNG file. Keep in mind to be able to see 1 stop difference the images need to be pushed more than just 1 stop. As a matter of fact if you shoot in M-DNG and resize up to L-DNG you'll still have more detail in extreme shadows. For those wondering S-DNG looks identical to M-DNG.

Can you provide more details? What was the exposure of both images (ISO, shutter speed, aperture)? What post-processor did you use? What do you mean by 3.5 stop push?

P.S.: I need that data to try to replicate your example exactly.

Edited September 6, 2023 by SrMi

[SrMi]

7 minutes ago, nicci78 said:

https://www.dxomark.fr/fr/Cameras/Compare/Side-by-side/Sony-A7IV-versus-Sony-A7R-V___1376_1385

 

looking at print size 60MP performed as good as 33MP sensor. 
At screen size 33MP beats 60MP. 
Once again what’s the point of using 60MP binned permanently at 36MP isn’t it even better to use a native 33MP sensor ? 

The screen size is when viewed at 100% and has no practical meaning in comparisons. Again, there is no binning in Leica cameras. It is impossible to bin from 60MP to 36MP.

 

Edited September 6, 2023 by SrMi

[Daniel C.1975]

vor 1 Stunde schrieb nicci78:

https://www.dxomark.fr/fr/Cameras/Compare/Side-by-side/Sony-A7IV-versus-Sony-A7R-V___1376_1385

 

looking at print size 60MP performed as good as 33MP sensor. 
At screen size 33MP beats 60MP. 
Once again what’s the point of using 60MP binned permanently at 36MP isn’t it even better to use a native 33MP sensor ? 

Don't get your point. 

It's about choices.

The binning mode in the Q3 is for me one of the major benefits. 

I don't always need 60mp and I do not always need 47mp (like I had with the Q2). The pictures above show a benefit on DR which is a nice benefit, but I am not sure how often that can be an advantage, but it's good to have if you need it. 

However, I've tested the Q3 briefly at ISO6400 binned and non binned  and exported the files from LR with standard development with around 16mp and literally saw no difference. It's the same quality - I've just checked on details and noise not on DR. 

This being said, I can use the binned raws for smaller files when I don't need the high resolution and can safe storage space and speed up processing, and when I need more resolution for print size or cropping I can go for the full size raws, and of course there is an resolution advantage in the lower Iso-values. I've also checked on it, and there is much more resolution compared to the binned raws or my 24mp SL2s. Maybe this resolution advantage diminishes at higher ISOs to a certain point. But who cares?

As said, I don't get your point. It's horses for courses. 

And the sensor of the Q3 is definitely top notch, as much as I can say so far. 

 

Disclaimer: If that binning is really binning and not just a downsampling of the files  - but... technicalities

Edited September 6, 2023 by Daniel C.1975

[SrMi]

Here are DNG files (L-DNG and M-DNG) and 6000x4000 JPEGs.

The modifications in LrC are: Exposure +.65, Highlights: -100, Shadows: +100.

Images shot at ISO6400, f/11, 1/4000sec.

There is no difference in noise in the deepest shadows. L-DNG's JPEG has a bit more detail.

https://www.dropbox.com/sh/rgjs19932ifmgto/AAApKMLbjftE23dxW2wTxJL2a?dl=0