flash wrote:
Did they test 18MP sDNG? Leica only claims an improvement with sDNG.
Gordon
Hi Gordon,
One would expect an improvement in noise levels pixel binning 60MP to 18MP. The point is that it's not an improvement compared to downsampling in post. So, at the end it's just a matter of preference getting the resized files out of the camera or not.
LBJ2 wrote:
The M11 Triple Resolution discussion starts about 45:40 in the video:
They both disagree with marketing on this topic as well as what some other reviewers have stated on the same.
At about 49:47 Josh says "to state this clearly...as of current firmware and Jan 29, 2022...there is no quality advantage, noise, detail, DR using mDNG or sDNG however there are performance advantages..."
Thanks. They say that
a 60-megapixel L-DNG downsized to the exact resolution of a 36-megapixel M-DNG has superior dynamic range, detail, and noise.
That is not what I see and, AFAIK, what is being observed by testers. Sean Reid's testing does not match the findings by David and Josh. Note that Sean Reid's testing is still ongoing.
I send the RAW file to @fred@ for 28mm Cron and 35mm lux with both M10P and M11. the lens profile are turned off. Based on what I saw, M11 seems brighter (about half stop) at ISO100 with the same setting compare to M10P. I don't feel M11 is worse regarding vignette at least with both profile off other than brighter overall.
I also send three 4 stop under exposure file at 60M, 18M and 36M to Fred as well. Based on my rough evaluation, 18M is cleaner at pixel level for sure which is not a surprise.
I will let the pro @fred@ to do the evaluation, I am too lazy and not really need to know those myself
flash wrote:
Did they test 18MP sDNG? Leica only claims an improvement with sDNG.
Gordon
For 60M downsize to 30M, it will give ~0.7 stop improvement. from 30M to 15M, it will give another ~0.7 stop. This is all therotic value from fundermental oversampling theory. Assuming noise is white/random but signal is the same during oversampling.
I don't know how Leica did it but based on their market material, it seems correct without surprise. We are really approaching something by number here and I will leave those to Bill Claff to do From a rough test I did, I clear saw improvement on 18M file AT PIXEL LEVEL but unsure how it compare to downsized 60M file.
I received @zhangyue's DNG files and was able to test for "vignetting" and "pixel binning vs downsampling".
All the DNG files received were first processed through DNG cleaner to erase any possible RAW cooking. I compared the resulting DNG files before and after DNG cleaner and didn't see any difference. @zhangyue@ had the lens profile turned off in camera.
To my eyes, the M11 files produce slightly higher vignetting compared to the M10-P when using the same lens, exposure and aperture. However, with real world images, I would not expect vignetting to look much different.
The conclusion is that there is no advantage in vignetting when using the M11. In fact, vignetting is slightly worse compared to the M10-P.
Lens used: Leica 35/1.4 FLE wide open.
Here are they side by side comparisons at 3m and infinity distance:
When comparing the M11 18MP pixel binned file (S-DNG) versus the 60MP file (L-DNG) downsampled to 18MP in post, I don't see any difference in the final result. So, no advantage in doing this process in camera except for getting a larger buffer depth and smaller files out of the camera.
The crops below were originally underexposed by 3 stops and pushed 3 stops in LR.
LEFT (M11 18MP), Right (M11 60 MP downsampled to 18MP)
At pixel level (100% mag.) - LEFT (M11 18MP), Right (M11 60 MP downsampled to 18MP)
I would say you underestimate the feature of less pixel without DR penalty but it is personal.
This is the cleanest 18M file you can get for FF right now. I remember there are a lot of A7Sx fans out there but M11 is even better. And then you can always turn 60M on if you want it. I forget to turn ISO back to iso64 for push files sent to you. The exposure was 4 stop under based on multi meter, but it is a dark green view.
Fred Miranda wrote:
In regards to comparing the 18MP pixel binned in camera versus the 60MP downsampled to 18MP in post, I don't see any difference in the final result. So, no advantage of doing this in camera except for the bigger buffer depth and smaller files out of the camera.
The crops below were underexposed by 3 stops and pushed 3 stops in LR.
I would say you underestimate the feature of less pixel without DR penalty but it is personal.
This is the cleanest 18M file you can get for FF right now. I remember there are a lot of A7Sx fans out there but M11 is even better. And then you can always turn 60M on if you want it. I forget to turn ISO back to iso64 for push files sent to you. The exposure was 4 stop under based on multi meter, but it is a dark green view.
It's a great feature and I'm glad Leica implemented it without deteriorating IQ compared to downsampling the file in post.
Knowing myself, I would always use 60MP though!
I'm already using the different DNG file sizes. Personally, I don't care much about theoretical measurements or detailed comparisons to my M10R. I just need a vague understanding how they compare and I'm golden. Slight exposure errors will mostly erradicate any differences in real world use anyway.
I like DR as much as the next guy/gal but I've never felt restrained by it. I learn what the camera is capable of and that's my shooting envelope. I've never, once, felt restricted by DR. I happily shot Velvia at 32 for years and years. Everything in 2022 is beyond my expectations and requirements. Seeing how it gors vs the M10R isa all well and good but there's more to any camera other than the sensor. Even if it were not any better there's lots to like in the M11 that make it worthwhile to me.
I'm certainly more guilty than many when I start to get bogged down and swept up in the hype that DR gets. It's become then new version of the megapixel war. And to be honest it's not what the M11 is really about. I shot th M11 and X1DII side by side today and the 7 year old X1D sensor still has the M11 beat, by a small margin.
I can downsize in post, except that I NEVER do. I just send to my required output in Lightroom, screen or print. So if I'm going to get to a smaller raw file, it'll be in camera. I went to a good friends moving away dinner. Don't need more than 18MP for those. Didn't have to think about a M10 or M10R debate. Just packed the one camera and made the decision on the fly. Wandering the city today at 60MP. It's something I'll use often. I think for some the M10R had too many MP. With the M11 you can set it at 18MP and shoot it this way forever.
I'm more interested in how my WATE performs compared to the M10R. It seems better at this point. Same with my 28 cron V1. More testing will tell the full story but so far so good.
@zhangyue thanks for providing the files, and thanks to Fred for the comparisons! Fun and interesting results 👍🏼
Next up is how to put this "Triple Resolution" function to good use. I know I can personally think of a few benefits and use cases and from the looks of it no need to worry about any IQ penalties either way.
Jan 31, 2022 at 08:54 AM
Steve Spencer Offline • • • • • • • Upload & Sell: On
Fred Miranda wrote:
I received @zhangyue@'s DNG files and was able to test for "vignetting" and "pixel binning vs downsampling".
All the DNG files received were first processed through DNG cleaner to erase any possible RAW cooking. I compared the resulting DNG files before and after DNG cleaner and didn't see any difference. @zhangyue@@ had the lens profile turned off in camera.
To my eyes, the M11 files produce slightly higher vignetting compared to the M10-P when using the same lens, exposure and aperture. However, with real world images, I would not expect vignetting to look much different.
The conclusion is that there is no advantage in vignetting when using the M11. In fact, vignetting is slightly worse compared to the M10-P.
Lens used: Leica 35/1.4 FLE wide open.
Here are they side by side comparisons at 3m and infinity distance:...Show more →
To my eyes the M10-P files are a bit brighter at the edges and corners even brighter in the center so the percentage drop off from center to corners might be higher. In between the center and corners the brightness is about the same. None of these differences are very big, however, so in the real world probably nothing to talk about here and whatever difference there is doesn't matter much.
Steve Spencer wrote:
To my eyes the M10-P files are a bit brighter at the edges and corners even brighter in the center so the percentage drop off from center to corners might be higher. In between the center and corners the brightness is about the same. None of these differences are very big, however, so in the real world probably nothing to talk about here and whatever difference there is doesn't matter much.
Although the DNG files had the same exposure, their histograms were not perfectly equal, so they were normalized as close as possible by tweaking exposure very slightly. When it comes to vignetting control, I don't think we would see any noticeable difference in real world image between these two cameras but at least this test suggests there is no advantage from the M11 sensor when it comes to vignetting level when using the same lens on the M11 or M10-P.
Fred Miranda wrote:
Although the DNG files had the same exposure, their histograms were not perfectly equal, so they were normalized as close as possible by tweaking exposure very slightly. When it comes to vignetting control, I don't think we would see any noticeable difference in real world image between these two cameras but at least this test suggests there is no advantage from the M11 sensor when it comes to vignetting level when using the same lens on the M11 or M10-P.
"The M11 also shows noticeably less vignetting than the M10-R"
Sean Reid's will look into it in more detail: 4. Tests of two 35 mm lenses on the M11, M10-R, and M10M: resolution, luminance vignetting and color drift.
"The M11 also shows noticeably less vignetting than the M10-R"
Sean Reid's will look into it in more detail: 4. Tests of two 35 mm lenses on the M11, M10-R, and M10M: resolution, luminance vignetting and color drift.
"The M11 also shows noticeably less vignetting than the M10-R"
Sean Reid's will look into it in more detail: 4. Tests of two 35 mm lenses on the M11, M10-R, and M10M: resolution, luminance vignetting and color drift.
Based upon what you posted on LUF, you own or have access to the M11, why not compare and present photos your own findings? If you present a different result, I'm sure that would be a very interesting discussion.
LBJ2 wrote:
Based upon what you posted on LUF, you own or have access to the M11, why not compare and present photos your own findings? If you present a different result, I'm sure that would be a very interesting discussion.
Proper testing requires a lot of time and effort. While waiting on Sean Reid's analysis, I am OK with relying on the experience of people who have used the camera for several months and with many more lenses, vintage and new.
I did a brief test with Summicron 28/2, looking at the brightness values in Rawdigger, and vignetting seems similar between M10-R and M11. My early conclusion is that with 28/2 there is not much difference in vignetting. However, I would not generalize my findings.
About vignette, Maybe it is the difference between m10r and m10p? I have to say I personally also don’t care too much on this vignette or DR. I am with Gordon here though I do appreciate m11’s clean file but I never really complain m10’s DR or any recent FF for that matter.
We can certainly continue discuss technical side of gear to understand thing better but for certain thing such as DR evaluation for different cameras and multi resolution, the way to do it is what Bill Cliff did on RAW level so that we can eliminate all software interpolation.
Fred Miranda wrote:
I received @zhangyue@@'s DNG files and was able to test for "vignetting" and "pixel binning vs downsampling".
All the DNG files received were first processed through DNG cleaner to erase any possible RAW cooking. I compared the resulting DNG files before and after DNG cleaner and didn't see any difference. @zhangyue@@@ had the lens profile turned off in camera.
To my eyes, the M11 files produce slightly higher vignetting compared to the M10-P when using the same lens, exposure and aperture. However, with real world images, I would not expect vignetting to look much different.
The conclusion is that there is no advantage in vignetting when using the M11. In fact, vignetting is slightly worse compared to the M10-P.
Lens used: Leica 35/1.4 FLE wide open.
Here are they side by side comparisons at 3m and infinity distance:...Show more →
Another vignetting test:
This time with the Leica 28mm f/2 Summicron wide open at 3m and infinity distance. (M11 vs M10-P)
Thanks again to @zhangyue@ for providing the DNG files.
The 35/1.4 FLE showed only slight higher vignetting for the M11 compared to the M10-P but I see a more pronounced vignetting with the wider 28/2 Cron. Perhaps we will see even higher vignetting with wider than 28mm lenses.
It would be interesting to see a comparison with the 21/3.4 SEM and CV 15/4.5 III lenses.
I'm surprising by these results based on the chatter about the M11 being able to provide lower vignetting compared to previous models.
👍🏼
