A7s sample footage

not sure..it just seems like sony has over hyped this cam for sure, I was planing on getting one, not so sure , really not feeling the images coming out so far ..maybe its just me
 
Ok, we nearly have the results for the riddle test:
http://vimeo.com/98791150

* He said my guess based on the highlight aliasing doesn't have even one camera right (you'll be surprised why)
* He said only one person has got everything right so far

That means that we already know:
1: F3
2: F55 slog2
5: F55 slog3

And the FS700 and a7s are 3 and 4, not sure which is which.

So, it turns out that the camera with the most highlight aliasing, by far, is the F55. And color changes a lot between slog2 and slog3 (I've never used it, this was probably known but I wasn't aware).

Since I don't want to be overoptimistic, I will guess that the a7s is 4, and it has more noise in the shadows than the FS700 (as we already saw in the APS-C test), but it also sees a bit more into the highlights.

edit:
Comparing with the F3, which was my favorite camera in the blind Zacuto Shootout from a couple of years back (tied with the Alexa) (here), and which has about two stops better DR than the C300 when recording internally (half a stop in the highlights, a stop and a half in the shadows) (here):
* if the a7s is camera 3, it has even better DR than the F3, both in the highlights (the drive on the left doesn't disappear into the window) and in the shadows (less noise)
* if the a7s is camera 4, it is probably on par with the F3: it has more headroom in the highlights, but more noise in the shadows
It looks like I'll be happy with the a7s, but I'll probably want to get an external recorder to reduce that shadow noise (C100 internal vs external comparison) (ninja 3 with 4K recording in early 2015?)

Also: more highlight headroom in the a7s than in the F3, and less in the shadows, means he was slighly underexposing the a7s; he wouldn't have done that if he thought that what Sony calls ISO 3200 is actually ISO 1600.
 
Last edited:
wallrc said:
I'm not quite sure what to make of that sample. Could it have been shot in 30p?
Click to expand...


Wow, I hope I'm not the only one who thought that last sample (people waling in the street) looks terrible. If thats the aesthetic of the camera in a street scene, this camera is most definitely a joke. DR in that image totally sucks. I sure do hope that was a false sample. No where near any level of quality I would think of working into a professional production. A C100 would walk all over that.
 
Last edited:
Here is another one:

https://www.youtube.com/watch?v=L1x3jPqntmM

Notice the ISO is at 3200. I expect thats not by mistake. Its most likely shot in S-Log and thats as good as it gets in this scene. Nice "blown out" everything. Man, I am truly bummed. When this came out...and again the term "game changer" applied...I got excited. I really thought for a second that we turned a corner. Just for a moment I thought we all might get the camera we thought was possible.

Sony, or anyone...THIS IS WHAT WE WANT:

$2500 price tag
13 + Stops DR in video
60 in 1080P
Log shooting at low ISO
Possible 4K option (hopefully internal)
Great codecs
Full Frame




PS...please...no one every use the term "game changer" again. Its like throwing a pile of steamy bad ju ju at it.
 
Last edited:
ReelWorksMedia said:
Wow, I hope I'm not the only one who thought that last sample (people waling in the street) looks terrible. If thats the aesthetic of the camera in a street scene, this camera is most definitely a joke. DR in that image totally sucks. I sure do hope that was a false sample. No where near any level of quality I would think of working into a professional production. A C100 would crap all over that in its sleep.
Click to expand...

It is clearly overexposed and shot at night and who knows what picture profile was used. You can make any camera clip highlights, its easy...
Looks clean to me, that's good.
 
Its the 3200 ISO limit (most likely)...it was an issue in this scene, as it will be in a lot. Its the fact that an average user did not get it right...either because he was required to use ND and didn't, or the camera was not capable of capturing everything with its available settings.

This sample looks no different than a 5D3.

https://www.youtube.com/watch?v=L1x3jPqntmM
 
Last edited:
cpc said:
Not sure why you think it's 3200. Might have been 12800 or whatever.
Click to expand...

The main reason I think its shot at 3200 ISO...is because its stated it was in the description of the video.



ps....sorry if I'm overly ranting this morning. I'm just expressing being a little bummed out...cause I really wanted to get into this camera and my mind has been changing about it recently. I do openly invite opinions to help me change it back. I would love this camera to be something really great. I just have my doubts now.
 
Last edited:
I am commenting on your comment on the Shibuya video, can't see iso mentioned there.
 
ReelWorksMedia: don't despair. Judging a camera by the worst footage you can find is not a good idea, unless you aim to use it in the worst way possible. If you think you can get the best out of the camera, you have to judge it by the best samples you can find.
In this case, with the 0-109 metadata thing, it is very easy for an uninformed user to post footage that looks absolutely awful, with bad DR and awful highlights.
 
ReelWorksMedia said:
Its the 3200 ISO limit (most likely)...it was an issue in this scene, as it will be in a lot. Its the fact that an average user did not get it right...either because he was required to use ND and didn't, or the camera was not capable of capturing everything with its available settings.

This sample looks no different than a 5D3.
Click to expand...

I doubt these people used SLOG2 and then graded it. They probably just used whatever video profile comes on by default and uploaded it. So it will be genuine analogue 3200 ISO (instead of simulated with SLOG2) meaning clipped highlights, and the super whites will also probably be clipped. This just means that any numpty with too much loose cash who picks up the cam and doesn't know what they are doing will still come out with camcorder footage, which is fine by me.
 
Mattoid said:
I doubt these people used SLOG2 and then graded it. They probably just used whatever video profile comes on by default and uploaded it. So it will be genuine analogue 3200 ISO (instead of simulated with SLOG2) meaning clipped highlights, and the super whites will also probably be clipped. This just means that any numpty with too much loose cash who picks up the cam and doesn't know what they are doing will still come out with camcorder footage, which is fine by me.
Click to expand...

I do hope you are correct. If you guys find any good videos that do showcase its ability first, please post.
 
We still don't know which is which (it can be camera 3 or 4), but this puts it at least in line with the F3 in terms of DR (and that means two stops better than the C300).
 
Samuel H said:
ReelWorksMedia: don't despair. Judging a camera by the worst footage you can find is not a good idea, unless you aim to use it in the worst way possible. If you think you can get the best out of the camera, you have to judge it by the best samples you can find.
In this case, with the 0-109 metadata thing, it is very easy for an uninformed user to post footage that looks absolutely awful, with bad DR and awful highlights.
Click to expand...

Hey Samuel, I have never personally owned a Sony camera. I have spent my career in other ecosystems. Would you mind explaining what is going on with the 0-109 metadata issue again. I read back through the threads and have a general idea, but I would like to know more and the best methods of handling that issue in Premiere. Would be a huge help.

Thanks.
 
My 5N records 0-109, but Premiere doesn't understand the metadata and thinks it is 0-100. After loading the files, highlights are horribly clipped. I have to apply a 16-235 levels filter, and after that the highlights come back to life. It seems that the a7s has the same metadata issue. I'm not sure if this is a Premiere-only problem, or if it is Sony's or Adobe's fault.

This video shows the before-after with the a7s: http://vimeo.com/98137975
 
Samuel H said:
My 5N records 0-109, but Premiere doesn't understand the metadata and thinks it is 0-100. After loading the files, highlights are horribly clipped. I have to apply a 16-235 levels filter, and after that the highlights come back to life. It seems that the a7s has the same metadata issue. I'm not sure if this is a Premiere-only problem, or if it is Sony's or Adobe's fault.

This video shows the before-after with the a7s: http://vimeo.com/98137975
Click to expand...


the the guy who took that video speculated that the A7s may go to 115, like Sony Pro cameras and that was the source of the extra range beyond 109.


"geco campos PLUS 15 hours ago
I really don't have that information, but Sony in some Pro cameras is working with 115IRE, so they are expanding the range in whites, but I don't know if they are using that in this camera, or is just part of the 15.3stops of dynamic range; but what I see is very hard to "blown" the images because there is allways detail in the back after postprocesing. No deep user manual or technical repair manual is out yet so I don't have more information about it; but is really good handling bright areas."
 
Seems pretty obvious most of these videos have been overexposed. If you want to favor the shadows when exposing you would lose?..........so like any other camera, you have to work within the range you have and expose accordingly. I'm sticking with my order and I'll return it if there is anything that really surprises me. After grading those clips (also ridiculously overexposed) that were posted, I'm pretty confident this is a nice improvement over what the fs100 was using and it looks like an improvement over what some of the other more expensive models will do in low light. Seems like everything Sony puts out, if you're going to shoot with a stock gamma or PP, gonna have a bad time. I was not impressed at all with FS100 until we loaded GLOG and then it was "damn, this is a good cam...."
 
Last edited:
Mattoid said:
This just means that any numpty with too much loose cash who picks up the cam and doesn't know what they are doing will still come out with camcorder footage, which is fine by me.
Click to expand...

:grin: Happens every time. Maybe I should get a cat.
 
You must log in or register to reply here.
Back
Top