My own Sony A7s test - 4K resolution, dynamic range, noise and GH4 comparison

dollsfoto said:
Just wow. For me this is definitely the most shocking part of your "review". I do not care that much about dynamic range, sharpness, colors etc, but this can be huge problem. Did you find workaround or does the camera makes this artifact all the time when shooting high contrast scene with backlit subject?

Anyway thanks for the tests!
Click to expand...


This might be a little bit of a over reaction :)

Dynamic range, colors, and the overall look of the camera does certainly matter when weighing overall image- if you're pixel peeping you might notice the highlight artifacts, and it definitely is one of the contributing factors (along with 8bit compression) thats makes the sensor suffer- but as most reviews/users have not pointed this flaw out, I would not start assuming this is something people will be noticing in your A7S footage.

If you look at this extensive testing of the F35: http://alfonsoparra-development.s3.amazonaws.com/uploads/docx/file/16/f35testev.pdf - which is a mainstay camera for macgregor - Alfonso states "As with F23 and other cameras like Genesis or Viper, the most valuable thing for me is the texture of the camera, something like a “film” feeling; the resolution, the grain, the color, the dynamic range conform a smooth image, with no noticeable video clipping and with a random discrete grain and white areas that disappear gradually till the limit."

Again, the overall "look" of the camera is determined by many many factors- and I for one see that this test tells me that the 2700 dollar A7S gives me a look close to a once $125k camera that short multiple hollywood films. In some critical aspects it does not hold up to the S35, but it others it actually surpasses.

Is there a feature/bug form that Sony has we can report this?
 
Last edited:
dcloud said:
thanks for the work macgregor!
does the a7s have better DR at 3200 on cine4 since its native ISO?

that highlight thing is a bummer
Click to expand...

No, the camera seems to apply the same gain in cine4 iso1000 as it does at slog2 iso3200 so iso3200 can't really be 'native' for both (whatever native means anyway). Iso3200 is just base ISO for slog2, any other observations of 'native' gain/iso depend a lot on how you are defining the terms involved.

Regarding the highlight edge problem, the fs700 has the same problem and I understand the F5/55 did as well. It was fixed in the F5/55 but never in the fs700 unless you shoot raw. Perhaps there is some hope for a fix if it was fixed in other cameras, but otherwise, it's just extra important not to let whites clip since it mostly occurs at clipped edges.
 
Last edited:
nyvz said:
No, the camera seems to apply the same gain in cine4 iso1000 as it does at slog2 iso3200 so iso3200 can't really be 'native' for both (whatever native means anyway). Iso3200 is just base ISO for slog2, any other observations of 'native' gain/iso depend a lot on how you are defining the terms involved.

Regarding the highlight edge problem, the fs700 has the same problem and I understand the F5/55 did as well. It was fixed in the F5/55 but never in the fs700 unless you shoot raw. Perhaps there is some hope for a fix if it was fixed in other cameras, but otherwise, it's just extra important not to let whites clip since it mostly occurs at clipped edges.
Click to expand...
i have no idea actually :) i dont have an a7s :D id like to know more. samuel H certainly posted some interesting stuff
 
I clipped the highlights on my short film I shot a couple of days ago. No highlight aliasing. It was not a very contrasty scene + SLOG, so maybe it happens when the contrast is too high?
 
hempo22 said:
I clipped the highlights on my short film I shot a couple of days ago. No highlight aliasing. It was not a very contrasty scene + SLOG, so maybe it happens when the contrast is too high?
Click to expand...
The FS700 has the same issue and yes, it seems to happen when there is a high contrast between something and the clipped area.
On the FS700, it doesn't happen when shooting 4K or the HD derived from 4K on the Odyssey 7Q. So my guess is that the issue won't be there when shooting 4K with the A7s.
 
Except fs700 4k is raw and bypasses sony's debayer and gamma/color processing, a7s 4k still relies on in-camera debayer and image processing which causes this problem. I would guess the issue most likely still occurs in 4k, unless it is a problem with sony's downscaling which it probably is not.
 
Great tests but hey, I spent $2500. I'm not expecting an F35 at that price much less anything from Sony's current higher end line-up. Pound for pound, it's a great camera I think. I'm also confused to the conclusion that Slog2 and Cine 4 have equal "usable range". Doesn't seem to be the case when I'm looking at it here and certainly not been my experience so far. If noise is that objectionable, you could remove it but that doesn't mean the DR must be tossed in the process, right? So again, thanks for putting in that work, you did a great job putting that together. Best test I've seen by far. (edited because I don't want to sound like I'm all that invested in the results and I not trying to knock anyone)

Expectations. Managed.
 
Last edited:
Nice test. Thank you.

Curious if you had Silent Shooting Mode engaged during your Nikon D800 test.

Yesterday, there was an interesting post on DPReview regarding poor light performance in that mode:

Silent Shooting Mode on A7s shows significantly poorer Low light Performance

Also, I don't have a dog in this fight , and contrast aside, but I'm seeing differences between the A7s and GH4 in the UHD Resolution tests.

Lastly, some might be interested in Alister Chapman's new and very comprehensive:

Exposing and Using S-Log2 on the Sony A7s

Edit: Added Chapman Link
 
Last edited:
Is there a mount converter to dumb ef? Wanna rent one and throw my rokinons on it.

Great test mac.
 
Two quick questions, please:
- Is the DR of internal 1080p different to that of 4K?
- Can you post the DR results of other cameras you have tested?
 
macgregor said:
Sony A7s
cine4 200ISO: 12 stops. 11 usable. +4 -6
S-Log2 3200ISO: 12 stops. 11 usable. +5 2/3 -4 1/3
PPoff 100ISO: 11 1/3 stops. 9 usable. +3 1/3 -4 2/3

Panasonic GH4
CineD 200ISO: 11 1/3 stops. 10 ⅓ stops usable. +3 2/3 -5 2/3
Click to expand...

I've been meaning to ask about how do you derive total range from +/- range since you first posted the numbers in the other thread.
Can you elaborate on that? I'm probably missing something as your numbers don't add up. +4 and -6 should give 10 usable stops, not 11, etc?
 
jcs said:
Current stop, +4 = 5, + abs(-6) = 11.
Click to expand...

Erm, still not following.
What is "current stop" when talking about range and not exposure?
Here is a simple case: +1, -1. That's a range of 2 stops for me: one above key, one below key. A total of two.
 
macgregor said:
a7sd800p1.jpg
Click to expand...

ETTR baby.
 
cpc said:
Erm, still not following.
What is "current stop" when talking about range and not exposure?
Here is a simple case: +1, -1. That's a range of 2 stops for me: one above key, one below key. A total of two.
Click to expand...

Actually, I think I got it. But I think it is a wrong way of expressing range.
If you count the "current stop" or the key, then a camera recording every incoming light level as a single value (say, 1) would have a DR of 1 stop/step/EV, while clearly it has DR of 0 stops/steps/EVs.
 
You must log in or register to reply here.
Back
Top