Samsung NX1 getting log-like setting

try by yourself with the files I shared. When you lift up the shadows of an image underexposed by more than 1 stop you can't recover details well and the picture looks ugly in that dark area.
Especially at ISO 400 the loss of details is dramatic compared to ISO 100...I wish I had shared that file as well before formatting the card. Sorry
 
Here's a couple 4k stills from uhd. Shot in Gamma DR with -10 sharpness and -5 contrast and master black level +12.
4k log files03 .jpg4k log files05 .jpg
 
Thanks for the DR test, diegocervo, it's quite good, I think I have a pretty solid preliminary conclusion.

* f/4 on the a7S has the buildings just about clipped, f/8 on the NX1 has them just slightly worse (points to a7S having 2 to 2.5 stops more range in the highlights)
* f/5.6 on the a7S is about a third of a stop from having absolutely no clipping (the sky has some color issues that will be gone closing just a bit), f/8 is perfect. NX1 is still fully clipped at f/11, and nearly OK at f/16 (points to a7S having 2.5 to 3 stops more range in the highlights)

So, the a7S+slog2 seems to have between 2 and 3 stops more range in the highlights.

Now, about the shadows...

* at f/22 on the a7S, the couch is noisy but could be cleaned and clearly has detail, at f/16 it's clean and definitely usable. f/22 on the NX1 has no noise issues and still shows detail, but is close to being crushed.

So, the NX1 seems to have at most 1 stop more range in the shadows, depending on your taste for noise reduction in post.

Join all that with my measurements here, and the NX1 seems to have around 4.5 stops in the highlights, and maybe 5 in the shadows, for a total of 9.5 stops, which puts it probably ahead of the GH4, but behind any other camera in the test, and in particular two or three stops behind the leaders (pocket+RAW and a7S).

All that's just out of the screengrabs you posted here. If I have time today I'll look at the prores files and see if they change anything.
 
I couldn't resist, I went through the prores files to see exactly where we stand... :)

Highlights: NX1, sky and buildings are acceptable at f/16, and not a bit earlier. Same level of "acceptable" on the a7S is at f/6.3. That's a 2.7 stops advantage in the highlights for the a7S.

Shadows: NX1, couch is perfectly fine and usable at f/22, with very little noise even after bringing it up a bit to see it clearer (though close to clipping in any case, so I guess at f/32 it would be gone). a7S is a lot noisier in the shadows, to get that same level of low noise it needs to be at f/11. Or at f/16 with denoiser in post. That's a 2-stops advantage for the NX1, 1 stop if you're not afraid of denoising in post.

So, in total, I'd say the a7S has around 1.3 stops better DR than the NX1, which puts the NX1 around the middle of the table, ahead of the 5D3 but behind the NEX-5N. And in any case, clearly ahead of the GH4.

Now, the interesting part:
* The NX1 seems to have the same sharpness as the GH4, but a lot less noise. I guess it's going to be a monster of a camera for green screen work... if the codec doesn't throw away color resolution.
* You're shooting 0-255, and ending up with a harsh highlight clip that happens at least one stop sooner than with any other camera I have tested. It's possible that the software you're using to convert H.265 to ProRes is treating the data as 235, and causing this ugly highlight clip. This happens with Sony footage that is not slog2, and severely limits DR if your workflow is not adequate (if you just load the file to Premiere, it clips around a stop sooner than it needs to, and in a much uglier way too; a 16-235 levels filter fixes this). I know there are not many options right now for H.265 footage, so, for the time being... check if 16-235 gives you more DR, it could very well be the case (also in the shadows, but it performs well there so my hopes are mostly in the highlights).


IF the 235 thing brings back one stop of DR --- AND -- the NX500 has the same image quality of the NX1 --- AND --- zyoptics brings out a lens turbo II for the NX mount... a Samsung may be waiting for me in my near future :)
(I'd be using the a7S for portraits and medium shots, and leave the NX500 on the Nebula 4000 with some lightweight lens -probably the cheap zoom, as a 16mm f/3.5 OIS prime- for wide shots)
 
Last edited:
Samuel, thanks for doing the math. :D
I converted the files from the nx1 with iffmpeg, which I think is better than wondershare.
Anyway, here I'm uploading the raw mp4 in case you want to try different converters.
https://dl.dropboxusercontent.com/u/2888648/NX1 Dynamic range test.zip

Soon I will test 16-235 as well.

Here is another clip shot with the NX1 that shows how good is h265 at dealing with movement.
It's a highway scene. I shot the same scene in the past with my sony AX100 in 4k. The two codecs, xavcs and h265 are night and day. The Ax100 had plenty of compression artifacts on edges and shadows that I can't see here.
https://dl.dropboxusercontent.com/u/2888648/NX1 Compression artifacts test.zip

Now, the only thing I'm really looking forward is a new computer to convert h265 faster! My macbook pro is taking ages to do it at 0.2x
 
I'm not getting anything extra from the original file, but that's to be expected, since I'm stuck at a similar workflow to what you had, because I can't just bring the H.265 file into the editor.
Which reminds me: no hurry, anyway... Time will make all of this a lot easier :)

Shooting 16-235 means you have 10.6M shades to work with, instead of 16.7M, but it may very well worth it *IF* you get one extra stop of DR in the highlights, and with a nicer roll-off.
(not the end of the world, in any case, since you're already throwing away 98.5% of the original data when compressing to 70 Mbps H.265; true, you keep the best 1.5% of the data, but well, just to put the 16-235 thing in perspective...)
 
Last edited:
Shenan said:
Samuel, has anyone sent you rolling shutter measurements in either mode (1080 and 4K) for this camera? I would love to see how it compares in real numbers with the others.
Click to expand...

Samuel H said:
No, not at all. All I know is that it's relatively bad, but how bad it is exactly, no idea.
If anybody has the camera and 10 minutes to shoot a test and send it to me... instructions here.
Click to expand...


I Samuel, I posted a link to a measurement I posted on dprevew forums on your rolling shutter thread here http://www.dvxuser.com/V6/showthrea...ing-shutter-put-a-number-on-this-issue!/page5
perhaps you didn't see it.

Anyway I have re-posted it in detail in a DVXuser thread here:

http://www.dvxuser.com/V6/showthread.php?333655-Samsung-NX1-Here-is-the-rolling-shutter-measurement

Basically
DCI 4k = 31.25ms readout
UHD = 33.33ms readout.

Feel Free to use the measurements in your list.
 
diegocervo said:
Here is a comparison between 0-255 and 16-235.
The latter retains highlights a little bit better, but I'm not sure how it would affect grading...we are talking 8bit here and I don't know if it's wise to limit ourselves further.
https://dl.dropboxusercontent.com/u/2888648/NX1 luma range 0-255 vs 16-235 f11 Iso100.zip
Looking forward to your comments.
Click to expand...

The 16-235 does look better. As long as the transcoding process is throwing away the extra information from 0-255 it never makes sense to shoot it. Doing so would be a false economy.
It's weird though that neither of the clips look similar to the examples earlier in the thread. In the earlier examples the couch was basically black even when the outside was clipped. But in these new shots both the outside and inside look well exposed.
Was it a lot brighter in the first examples and was it definitely set to Gamma DR?

It would be really good to see another comparison with the A7s with Gamma DR and 16-235 on the NX1. Your setting is a really good location to test dynamic range.
 
I forgot to look at that 16-235 test, thanks for reminding me...

It's better to run a full exposure ramp, but looking at the waveforms of these shots it seems they clip at exactly the same point in 0-256 and 16-235 mode.
 
Thanks to geffyb, I have all the rolling shutter measurements for the NX1, and they are really interesting:

Shooting UHD, it's the slowest camera I've tested, at 32.6 ms.
Shooting 4K is a bit faster, but still slower than anything else I've tested (well, it's actually a match with the a7S in FF mode): 30.9ms. I'd shoot 4K all the time, no point in using UHD that I can think of right now. You get more room for cropping in post, and a bit less RS.
And in 1080p mode... well, in 1080p it is basically four times as fast, at 7.9ms. This puts it right at the top of the table, leading much more expensive cameras by a very wide margin. Amazing. I wasn't expecting that.

Full table and comparison with other cameras over here.

I think right now, if you can't get an FS7, this is the camera to get for green screen, and also for fast action shots. If they add a gamma mode that matches the DR of my a7S+slog2, it's the all-round winner. I still need a leicaR-to-NX speedbooster in order to consider the NX1 for *my* needs. But WOW, on two counts by now.

So, about that 1080p mode... it must be doing either line-and-column skipping, or in-sensor 4x4 binning. How is aliasing/moire? And how sharp is it?
 
1080p is a little soft, but no false colour and very little aliasing. Need to test it further, 4K is best when sharpening is turned all the way down but it seems to soften the image in 1080p. Also seems it might have better DR than the 4K mode. I should send you some more samples, or you should rent the dang thing :)

Also, in regards to DR and sharpness I don't know if we can do full testing on it until the codec is supported by premiere/final cut.

Even though this is 120 fps, it's a good indicator. 1080p 24fps is a little sharper but a similar image quality: https://vimeo.com/118529486
 
I'm not renting it... unless I get a green screen project, then I will strongly consider it.

The reason: I'm not in a hurry, the NX1 seems superb but my a7S is working great too. If I were to get a shogun, I would probably get a Samsung instead. But I had no plans for that either, I just bought Magic Bullet Suite 12 instead.

Plus: all this waiting gives Adobe some time to get H.265 working in Premiere. I need both that and a speedbooster/lensturbo for my Leica-R lens set.
 
Samuel H said:
I'm not renting it... unless I get a green screen project, then I will strongly consider it.

The reason: I'm not in a hurry, the NX1 seems superb but my a7S is working great too. If I were to get a shogun, I would probably get a Samsung instead. But I had no plans for that either, I just bought Magic Bullet Suite 12 instead.

Plus: all this waiting gives Adobe some time to get H.265 working in Premiere. I need both that and a speedbooster/lensturbo for my Leica-R lens set.
Click to expand...

Can you think of any reason why the DR might be better in 1080? Is that a real possibility? From what you've seen so far, what do you consider the biggest minuses with the camera compared to the competition- say cameras in the sub $2,000 range.
 
You must log in or register to reply here.
Back
Top