A7s sample footage

So i'm behind here. Has their been a firmware update for the a7s?

Edit. I feel like an idiot. I turned on our a7s and checked the ISO in movie mode and it goes to 100, which is part of the upgrade. So, it was only the prototype cameras that lack the DR upgrade. Thank you Erik. That will teach me to post just before bed. Sorry.
 
Last edited:
Sorry. I have deleted the previous post. Looks like there is no firmware update on Sony's website...
 
Remmeber, the winner is not the one with more DR, but the one that gives the most balanced DR.
Lots of stops in the highlights is what we want.
The 21stop chart is great to measure the total DR, but I still prefer under/over exposure test to see where that DR is allocated.
 
After a day with the A7s I can say it's not for everyone. The dynamic range is huge but that extra sensitivity makes it a very sensitive camera if you know what I mean. You can't just point and shoot with S-log like it's a GH4 or MK3 raw. And the rolling shutter in full frame mode is atrocious. I was shooting off the shoulder (not balanced) with a 50mm lens and I got jello. With a lot of TLC it should look amazing but it has a lot of issues. I managed to induce the highlight aliasing and the purple smudges without too much effort. This thing is very light hungry, I wouldn't even attempt to underexpose S-log (well I did).
 
Ok, ok. I'll stop ;) It's all subjective, lots of ways to run with an image for moods and taste. What is clear is that when you have a vision, experience and ideally a crew and a budget, you can create fantastic images with just about any camera. Give Deakins a T2i and put a competent colorist at the helm and you'd get an image that worked within it's boundaries that would blow your mind.

Amen.

I looked at that DR post and it really seems like most of the commenters haven't used all the cameras. I happen to own all but one and have used them all. I wrote some of my experiences (I still vastly prefer my C300, overall, mostly for ergonomics and actual handling - they all have similar images) but my conclusion makes a similar point to you:

"Each camera has a purpose – including the Alexa/Amira. Look at Philip Bloom – who has pretty much everything short of an Alexa. He is shooing his latest work with an A7s, a GH4, I think a C100 and maybe his F55 thrown in there too. Oh yes, and a GoPro on a Phantom.

This arguing for one camera here or there based on numbers is absurd and pointless. It’s nice as a baseline to know what is where for lighting purposes – so you can plan your shoots and pick the right camera, but all these cameras have a purpose. They are tools in the kit. They are (mostly) affordable so many can afford to have 2 or 3 in a kit even as a one man band. No one camera is perfect, but generally the more you pay the less compromises you make. Opportunity costs: you can have cheap, good, or fast. Pick two.

At this point the camera is not the issue. They all look amazing in the right hands, charts be damned."
 
a7s within half a stop of the Amira. Awesome.

Also, FWIW, looking at the charts by eye, I would have said the a7s+slog2 has 14 stops and the GH4+cinelike has 10 stops.

Until we see the resulting images i'm highly doubtful. I only see two low resolution charts on that page - am i missing some?

One of the things Sony does very well is to roll off the highlights and the shadows to give the impression of technical range. In fact i guess all camera manufacturers do this. I have a transmission wedge, the same one dpreview use that has 41 1/3 stop wedges. I can backlight it with a remote phosphor light (no hotspots) and do the same test in a darkened room. You then decide from the resulting images at what point below clipping do you decide there is no detail. This is a very subjective decision because it's based on your tolerance for noise. I personally put this around 12/12.5 for the A7s, which i think is amazingly good. 12 stops of light in an 8 bit container is insane in terms of allocation of code values per stop.

Also WB is an issue, understanding that these cameras are more sensitive to green it depends on the process used to reconstruct the RGB and how much highlight recovery is going on. Step wedges are ideal conditions for demonstrating the benefits of this approach. The images on the site are not white balanced, are they? They're warm. The 4K RAW of the FS700 gains a whole stop doing this.

Now it could be that the transmission wedge i have is not accurate (although it's supposed to be for this kind of thing). The Xyla chart they are using may not be matched quite the same. So it's worth combining this range testing with other methods too. If the shutter speed is accurate then that in combination with aperture changes may allow you to cycle through a whole range of exposures.

Just adding some real world balance here...

cheers
Paul
 
^ All I can see are the 800-pix-wide charts that you get when you click in the thumbnails:
http://www.cinema5d.com/wp-content/uploads/2014/07/A7S_dynamic.jpg
http://www.cinema5d.com/wp-content/uploads/2014/07/GH4_dyn.jpg

So there's no way I could talk about noise, detail, or anything like that.
Still, the a7s chart image is remarkable, and the difference with the GH4 jumps out of the screen.

I'm waiting for someone with an a7s to shoot something like this. If nobody does before that, I will shoot it when I get mine (in late August, ughhhh).
 
Squig, no doubt, you have shoot outside your exposure comfort zone with slog2. Fortunately, I was already used to shooting like this from dealing with the BMCC/BMPCC so an overweight exposure wasn't a new thing for my eyes to gauge. Also, my dear lcd looks much dimmer than how the footage comes out now matter how bright it is so I'm still looking at 100% zebras and my histogram. It'll get better :)
 
Squig, no doubt, you have shoot outside your exposure comfort zone with slog2.

I'm just gonna forget today ever happened. There were a couple of shot tonight where I just knew I'd hit the sweet spot just looking through the viewfinder, no zebras or histogram. I'm gonna try to tweak my AC7 for S-log.
 
Back
Top