[ArtBenny Video] Sony A7s iso test from benny wong on Vimeo.
Today Sony announced the A7s in Hong Kong and Japan. And our reader Benny shared videos and image to SonyAlphaRumors. The video on top shows you the High Iso performance in complete darkness. He also sent me the first full size image samples taken with the A7s. These go up to 40.000 ISO and you can download them here at Wetransfer (10MB zip file). Please do not remove the copyright note if you share them (and kindly link back to SonyAlphaRumors as source).
This is the image taken at ISO 40.000 in complete darkness and with the [shoplink 26511 ebay]Canon 50mm f/0.95 lens[/shoplink]. I know it’s noisy but as you see on the video on top the room is really COMPLETELY DARK!
And this is ISO 800 with the small lights on:
If crazy High ISO is what you need than the A7s is for you. And remember that since today you can also buy a “cheap” native [shoplink 26501 ebay]Mitakon 50mm f/0.95 on eBay (Click here)[/shoplink]. Also note that the A7s hd a very early 0.01 firmware version. Image Quality will improve with the final firmware which will be ready right before the first July shipment.
A7s Preorder or pure notification store links:
USA: Sony US, Amazon and BHphoto.
Europe: Sony DE, UK, ES, FR, IT, NL, SE, FI, PL.
Asia: Digitalrev
Reminder: You will need the Atomos Shogun at BHphoto (Click here) to grab 8bit 4:2:2 4K video.
0 Responses
”
If crazy High ISO is what you need than the A7s is for you.”
It will depend on the price. 🙂
im interested in seeing still image dynamic range tests. as i hate the typical hdr look i normally use exposure bracketing and combine the images by hand in photoshop. so a ‘better’ sensor might save me a lot of time when i want to shoot high contrast scenes.
ya, until now still no mention on DR for both still and video.
ISO12800 looks pretty noisy to me and ISO40000 is useless. I expected better performance.
Only because it was underexposed in spite of the high ISO. In the second part of the video IMHO the noise was much lower at the same ISO values.
I think he was commenting on the stills? I downloaded them and the JPGs are bit noisy even at 12k ISO; however, there’s at least a stop improvement over the A7 to my eyes.
jpegs are not worth looking at, but the raw files look very good
Where are the raw files ?
Judging on jpeg is useless anyway.
There was a link but now it got removed.
thanks; maybe you or someone can repost them anywhere ?
I guess it got deleted for a reason…
But ISO12800 looks roughly like the NEX5N at ISO800 (but it can be downscaled a bit) so it is probably very much at the 1DX D4s level with a a bit less resolution and not as good AF.
what are you using to process/open the RAW files?
Metering has to be done properly while using the Canon 50/0.95. If it is metered for the center, the corners will be extremely dark (hence noise) because of the light fall off. This improves by f/1.4. Another fact(or) to consider is the very low contrast of an wide open f/0.95 lens. It needs contrasty light (kind of defeats the purpose of such a fast lens..), especially on digital. B&W film is a different story. 🙂
The first several dozen units in the hands of actual buyers will determine the “real” price, I think. With all these in-determination about the pricing from Sony, that is the way to find it. I only hope that the price cuts will not be dragged out by kit/lens promos and such and will be a straightforward cash slash. I am encouraged by the samples. 🙂
how about the fact that it is even possible? The only other camera that has the capability to do this is a 1DC . There was, NO light in the room!
under controlled conditions nobody will use such high iso settings. in any other scenario its all about ‘getting the shot’ and not the perfect looking image. this camera is not able to make miracles come true…
The room is completely dark dude. No other camera can focus under such circumstance. You might want to argue, what’s the point of shooting when you can’t see anything.
And you’re going to use which other camera instead?
Video up to 6400 looks very clean but a tad soft. Any way to get downloads of the OOC MP4’s?
Probably because it is 720p.
How do you figure that? The original is 1080p. I’m downloading the original (well uploaded to the web so not OOC AFAIK) right now.
Just watched the downloaded original from Vimeo and I stand by my statement that the video is clean but soft. Could be a focusing issue but I didn’t see anything that seemed sharp. The IQ a such high-ISO is pleasing (at least on part with what I see from Canon but cleaner) but not as sharp as I expected for a sensor that’s no doing any line skipping.
It is shooting at 1.8. I wouldn’t doubt the focus being off.
Sharpness is dependent on more than just focus. Factors include the angle of light, resolving of the details and contrast. In a practically dark room with a glimmer of overhead lighting at extreme ISO of course there isn’t going to be anything sharp.
Good point. Just observing. The footage on F-stop Academy is pretty sharp so I’m waiting to make any final judgements until more are out in the wild. 😉
So far the camera looks very promising with a few of these Sony head-scratchers.
With 16-18 Mpixels A7s would be perfect :p
Seriously? When was the time when you said: uh, I Need four more megapixels? ANd then again you can buy the Df then.
So a noisy picture.
Ok, it might be good for beeing an ISO 40000 picture, but it is still an ugly picture.
Better to switch on the light than upping the ISO 🙂
True, the camera better be named A7N (noisу)
Pity the test wasn’t done with a more standard light source. When the 5D2 came out it was a flickering candle a few feet from the subject. Den Lennie’s tests over on http://fstopacademy.com are a little more useful.
rather than seeing these extreme condition tests… why not show real world applications? I mean, for one thing, that video means nothing to me because I couldn’t care less about video, and even if I did, I would never shoot in that kind of condition. TO ME, I’d be more interested if suppose Sony showed some shots from a concert or perhaps an actual wedding… all I want is something useable @ISO6400 (with little to no pp), I really don’t need anything above that.
It´s just marketing crap.
Check out F-Stop Academy’s footage and BTS video. Yes, Sony is using it in their marketing but they actually were pretty hands off and I think the filming situations are much more realistic.
I shoot wedding & event work often in depressingly low light . I have found that for prints up to A1, I can deliver excellent results using my D4 and DF at 6400. Even my old D3s was capable of excellent 18×12″ prints at 6400ISO. I am pretty sure that the D610/Canon 6d would do the same.
Wedding work is one of the few areas where prints are still a major part of the game. As we know noise that looks annoying at 100% on a 28″ 4k monitor will never be seen in print same thing with processed reduced size preview gallieres on the web or DVD /bluray . I dont know what you shoot with but at least for some teh useable ISO 6400 arrived a few years ago
I’m looking forward to see better tests… These images aren’t good, I expected better. We’ll have to wait to see some raw files, I guess. The camera might produce usable images up to ISO 50000, but that’s something we’ll need better examples of.
The iso12800 of the zip with the cooker is so noisy.
I expected clean iso12800… too bad.
‘I expected clean iso12800’
do you also believe in santa claus?
it is only the bad jpeg processing (probably pre-production engine as well but Sony jpeg has never been really great), the raw files are very clean
On Sony’s website the claim the A7S has a new sensor package and color filter so maybe that’s thrown off the their JPG engine.
I’m sure that the A7S’ ISO 12800 must look like my 5N’s ISO 1600… So, almost clean. But these images look bad. I took pictures at ISO 6400 that look better with the 5N (after post-processing, and the images were purposefully slighty overexposed to start with).
The first image is with the Sony FE 55/1.8 not the Canon 50/0.95, the second jpg is at 16.000 ISO, not at 12.800.
I’ve said it before and I’ll say it again, Sony allows some the worst tests of its pre-prod models, go out and shoot some actual video. Shoot a Sony commercial, shoot a carnival at night, shoot people at the beach, shoot something that a buyer would actually do instead of some woman sitting in a dark room. Also, don’t let anyone use soft/low contrast legacy glass like the Canon 0.95, use Zeiss ZA’s and FE’s. After two really nice pieces from Den Lennie we get this, which really does nothing to convince me I should spend whatever they’re going to ask for the A7s. Where’s Brian Smith and the other Sony artisians?
this isnt an official sony video and the fact that its posted here on this site doesnt make it an important thing, dont you think?
More important than any official Sony tests!
It may not just be the low contrast (wide open) of the Canon 50/0.95 but the the AA filter stack in the A7s also could be a contributing factor. I see this with AA vs non AA filtered cams.
Really? Where? 🙂
Although this isn’t an official sony video – I would think they’d be careful who they gave pre production models to and would at least have an agreement to approve the tests prior to releasing online. At the NAB they didn’t allow testers to insert cards. Considering we have only a tiny handful of A7s samples so far, I find this all just a bit bizarre.
Yes they do…
Indeed, this is what winds me up about people who are able to get their hands on pre-release kit. They don’t appear to have the first clue how to test it or even WHAT to test. Who the hell are these privileged idiots? Some Sony employees Nephew?
The latest for me is people testing the AF on the A77II by just moving from one static object to another, they did the same “test” on the A6000 AF too….sigh….what a waste of an opportunity.
OK. The camera can go on an iso that high. The compression artifacts are completely mental, though. Noise should not be a bunch of square or rectangular blocks, just grain.
With the AVCHD codec, we’ve seen that cameras resort to macro-blocking to compress the footage and keep the bitrates within the codec’s specifications.
I am beginning to fear that XAVCs is not as beneficial for high iso footage as it could be. Hacked Panasonic cameras show grain, not blocks, when shooting in high iso situations.
https://vimeo.com/35307811
Download the video and see the ISO 6400 shots there. To me, they look considerably better than the A7s 6400 image granularity displayed in this sample (see the shadows in both clips)
Don’t get me wrong. I am really supportive of this camera and other Sony cameras, but the codec restrictions must move forwards, not stagnate.
GH3 is a good cam, it really is, but even that camera has some give when it comes to higher ISO samples (see details and lines):
https://vimeo.com/86743259
Neither of the two videos above are using the best patch (Moon T7) so more can be had with the Gh2 than what the samples above show.
It isn’t like 95mb/s SD cards are too expensive today or inaccessible as they were some years ago. We can definitively deal with a bit more mbps these days for the sake of detail and grading, no?
In that case high ISO video may be realistically limited to using an external recorder. We’ll see. It still seems like a 1-2 stop improvement in useable ISO for video over existing HDSLRs. I can’t imagine wanting to go much above ISO 12,800 and it looks passable up until that point (you can practically light with cell phones at that light).
My bigger concern is the video seems a bit soft. These are all really early reviews. Fun to speculate still to early to draw too many conclusions I think.
This video is particularly soft because of the light and maybe the lens too. We have to trust the guy focused properly to begin with and with such a large aperture, that the model isnt moving too much…
Not in pitch black dark. Take your camera and try shoot in the dark.
I have. The granularity is there but not macro blocking. It is not about how much noise you get, but how you get it.
I wonder… might be a combination of that and compression to mp4.
I would suggest, in Sony’s potential defense, that it could be the guy’s export settings.
I downloaded the video.
I just did a test in my office. One small light on about 1.5m behind me (25w desk lamp lowered to almost touching the desk so the light is subdued) aimed at my bookshelf which is 1.5m away and shot at ISO 1250, 2500 and 6400 with my GH2.
ISO 2500 is close to the look of A7s ISO 6400 but A7s looks a bit better with chroma noise but there is more blocks, whereas GH2 has smaller noise.
ISO 1250 is very clean on a hacked GH2 and it is like A7s on ISO 3200.
So, you could either say:
1) One has gained a stop and a half clean footage in however many years since the GH2 hacks.
2) XAVCs is holding the real hardware back. Imagine what can this camera really do if the firmware was not holding it back…
3) Also one needs to remember what the 50D, the 5DII and 5DIII can do with Magic Lantern to also get a perspective of how good could this have been….
yup. Silly putting out tests like this that leave people scratching their heads.
“Grain”, “granularity”, “macro blocking”… oh boy… video people starting to sound more and more the nut cases who try to go and replicate HCB and AA shots.
i would honestly love to shoot night video with this, even at 400k ISO you can tell what the image is enough to make it interesting!
one could almost imagine filming under the milky way with this camera.
I don’t know what to make of this really. ISO 40,000 images look better than anything I’ve seen from Canon and Nikon FF, but I’d never use that setting anyway. ISO 12800 doesn’t look groundbreaking to me although you can’t really tell without side by side comparisons.
iso1600 watch light flikering?
Hong Kong is a PAL 50hz country, but the camera was set to NTSC 60hz,
Great, I was thinking about buying the A7r. I love dynamic range + color bit more over anything else. Confused!!
A7r = for compact picture people
A7s = for video folks..or
now the A7s performs better for photos and video now, which makes the a7r obsolete???
Depends your needs. Do you want higher resolution and no AA filter? Then the A7r will do a great job, but it’s more difficult to shoot with it because the res is so high. Don’t need the super high res, want to take great low light pictures? The A7s seems like a good way to go. Just wait to see good samples from the A7s to compare the two cameras.
good question, does the a7s have a AA filter?
most likely, it’s a yes.
Due to the amount of resolution and video usually benefits with an AA filter more then photo, which this camera seems to aim for.
Actually you got it slightly wrong with your description:
A7r = high res, studio, landscape.
A7s = low light, video
A7 = general purpose
So if you’re a big cropper, the A7r extra resolution will be of benefit, but if you’re not then the A7s will probably be more of your liking.
Jonathan F. V. & Tan = Thank you for your replies.
That listing you did tan makes a lot of sense. I don’t crop at all, and I mainly I do photography, but I wouldn’t mind the extra video as cover, and the buffer against noise on the one time I’ll have to go over iso 800, because I don’t have a flash on me. hum.
I personally don’t like using a flash much, and I’m not doing portraits, so most times I just expose a bit longer, depending what I’m trying to do. The A7s would be a big plus for me.
I choose A7.. middle of these 2.. A7r-esolution and A7s-ensitivity
If someone’s a “big cropper” they should learn to compose.
thanks for posting, makes me want to buy one really badly.
Really? I came with a different conclusion. After seeing the sample images and video I no longer want to buy one. I was ready to go as high as $2,500 for the body. I hope we get to see some more samples soon. I had high hopes for this camera.
I guess people like you will NEVER be happy.
Touché my friend.
Douche my friend.
Really? you changed the T to a D and you think you are clever. Looks like you are the one that is unhappy. See ya…..
I changed the letters cause I’m fancy, like the canon rebel with kit zoom you be sportin.
Actually, I own a A7r, RX10, NEX-5N, I sold all my canon gear to get these, and I am very happy what they are intended to do.
What’s strange is that I sold my NEX-5n and my RX100 for Canon gear – because I couldn’t stand the color rendering.
You shouldn’t complain about the pics. For one, its at least 10 times better then your mass consumer grade canon rebel. For you thats a massive step up. But more importantly, there’s zero grain and if there is you just keep stepping backwards from your computer monitor- instant fix.
I never had high hopes for it, and the reason is simple. There is absolutely no progression in technology and sensor magic with this camera. If you want more light to come in, just make windows bigger, and that’s an evidence that is as old as the world is. If some day someone comes and says, “well, we have made a new redesigned sample of the 36 mpix chip, and now it makes 105000 iso clean, that is a progress. Here, we only do a low resolution chip with very large pixel pitch, what gives every pixel a double surface and the picture the resolution same as Nikon’s first 6 mpix D70, and thus the same picture quality with the improved dynamic range of a FF sensor. Is that the real deal. It is simply playing with our balls. Also, even if software rework inside cameras has progressed, no soft will ever be able to put something valuable where there is absolutely nothing. Sony has made a huge progression, but still is far away from the master in that cheating practice what is Nikon. Anyway, this camera reminds me the highly praised Nikon D700 with the 14 mpix chip that had, for that time it came out, huge high ISO facilities. But when I saw the first real world shots and saw images similar in quality and finesse as my former owed D70 and not even able to beat the finesse in IQ of D200, I forgot to buy one quiet fast, the price not having been a tempting one at that time.
Now, this said, if the camera was able in the dark to take stills that are free of noise at 105000 ISO, I would buy one, just for the pleasure to make the usual downsized prints of alround and travel shots we all usually do, and this includes bars and nightlife. If you see the samples posted here on SAR in another more recent post, you will see that the noise starts at 6400 ISO, fine noise, but, some detail loss can already be sen at this ISO range. This camera might have good video IQ facilities, but with fixed image, where you look at the same image for an long period, at full size and see all details, you will still be disappointed. I do not make any videos, or, smaller clips, but seldom. So, why indeed buy this camera. For me the resolution of the A7 with 24 mpix matches my needs, the difference between the 7 and the 7R is not that noticeable anyway. I do not think that A7S, when it concerns stills, makes anything better than 7 or 7R, even if beyond 6500 there is less noise, detail loss starts higher than 6500 ISO for all of them.
Is there any word on what the dynamic range will be yet?
Better to use your eyes than run after numbers. There are lots of other aspects that a DR measurement from DxOMark alone could show.
Did that question deserved such a jerky response?
Jerky would be sitting in a circle pleasing each other… in other words, what usually goes on here.
Yes , any question must be all start with “Sony is amazing better than everyone else” then and only then may you be allowed ti make minor critics lol
But numbers are consistent across platforms and give a good baseline judgement for comparison.
If Camera A has 10 stops of dynamic range in the test, and camera B has 11 stops in the same exact conditions, we can conclude that Camera B has more dynamic range than Camera A.
The tests are also done under more strictly controlled conditions, so the results are more accurate.
Numbers are worse because you assume someone else’s methodology is correct. When using your eyes you can still compare side by side… and you make conclusions only when you can do so with confidence. Also, even when done correctly, what’s the point of generating numbers for things we can’t tell with our own eyes? When we take photos we don’t show numbers to people, we show the photos.
DxOMark’s philosophy is to dumb everything down to the level that people who can count to a 1000 can understand. In that process a lot of meaningless numbers are produced as well. Sure they’re consistent across the board and naturally they will be as long as the numbers are calculated in a fairly deterministic way… doesn’t mean they’re not flawed.
It’s the same with photozone and a most places that show numbers. Even when done correctly, like with manufacturer MTFs, numbers aren’t indicative of the end result… that’s why manufacturers show image samples.
Also, anyone who has done computational image and data analysis knows how noise and DR measurements done by DxO are flawed. Their lens measurements aren’t even worthy of comments in a ps/xbox/wii forum.
Do you have something personally against DxOMark? Because you’ve brought them up a couple times now, and I don’t recall mentioning them in my original question.
All I asked was whether or not there was any word on the dynamic range, not what the DxOMark score was. Maybe Sony has said what the dynamic range is and I simply missed it. Maybe someone that has a pre-production model has done a comparison to other cameras.
“DR” is synonymous with DxOMark. I’d like you to point to other sites that attempt to measure DR like how DxO claims. Before the whole “DR” obsession, it was noise as measured by dpreview but people found out the limits of their measurements and moved on. Not that DxOMark’s noise measurements are actually very meaningful.
Also what good is what the manufacturers say? Sony says their raw files are raw but it’s not always the case. Nikon isn’t too different themselves. Again… using one’s eyes is the best policy. I can’t believe why photographers, out of all people, are so averse to it.
Pretty sure DR is synonymous with dynamic range, which is a very real, very measurable factor in a cameras performance.
What the camera manufacturer says is better than absolutely nothing at all.
And using ones eye is completely useless unless you have different cameras right next to each other, using the exact same settings, using the exact same lenses. And since I have been unable to locate such a comparison with the a7s at this time, hence my original question.
As far as DxO readers are concerned, DR is just a fancy word because if they believe DxO’s DR measurements they don’t understand DR in the first place.
So the real problem that you’re facing is not having access to the cameras. Even if someone else had access to the a7s, which other camera did you want them to compare against?
GH4 and 5dmk3 would be the two big ones. Follow that up with the Blackmagic cameras, Sonys own FS line, the other a7 cameras. Really, any of the cameras in a similar price range, and feature set of the a7s.
That’s why it’s easier to just have a number, because very few people have access to these laundry lists of cameras.
The range of cameras you mention is very wide and DxOMark even struggles to make sense comparing the D800 to the D800E… so, yeah sure, anyone can come up with numbers but the problem is making the numbers meaningful in real world usage.
Most definitely the the A7S will have lower noise in video because of the full-frame sensor combined with the full-sensor read out, because others that have full read out don’t have FF sensors and ones that have FF sensors don’t have full readouts. That’s easy to figure out.
The ability to do raw video will give cameras like the 5D3 an edge when extreme adjustments are involved. So only you would know if you’re the type that needs full flexibility (raw video) or the type that needs to be able to shoot stopped down with a single candle 🙂
Yes, that is a wide range of cameras, because there’s a lot of good cameras out there and I wanted to know where the a7s would fit into the lineup since it’s the only one (currently) with such high ISO capabilities.
And seriously dude, just get over DxOMark. There are other people that test cameras too.
Tell me one place that measures DR properly… that’s all.
i was at NAB and asked the engineers beside it. they said 12 stops.
Huh. I would have expected more, given the massive pixel size.
I was hoping this camera would let me film with no light at all. I stage all my videos in a sealed closet.
lol!
Yeah, I do the same and I just realized that every video — hundreds of them! — that I’ve shot for the last seven years is 100% pitch black. It makes you really wonder what the camera manufacturers are thinking (if they’re thinking at all) and why they’re neglecting the lightless camera market. Hurry up and make better cameras, idiots!
Just turn on the lights…
Time to film my private sexy video in the dark with the Sony A7s..
Too noise toooooo. And a hanged features camera. GH4 blow this away in all video aspect
incorrect. only noisy at the super high ISO’s. just add the slightest amount of light and you won’t need to use those ISO settings because its so sensitive
Take this… https://www.youtube.com/watch?feature=player_embedded&v=RgiqpIPbkQ0
I’d prefer to see samples of the A7s under practical situations, like a poorly lit restaurant or a birthday party, not stuff I shoot but always challenging lighting situations. I’ve tried the A7 camera and the images look great but the User Interface is horrid, like MS DOS 30 years ago.
Sony HAS to do something about redesigning ALL of their user interfaces.
For those of you who grabbed the A7S raw files before they got pulled Photo Ninja will open them—although Photo Ninja’s processing engine is a bit different than most so maybe not totally representative.
Kind of odd that the name of the girl in this video is Sony and she just happens to be doing a Sony preview.
her name is Fiona. she is hired by sony thats why had a Sony tag on her
yea, uh, no. watch the video again, her name is listed towards the beginning the bottom right.
haha this will be the perfect perv camera, I can already see loads of men snooping round parks during the night with them 😉
holy crap! WANT!!!!!
is that nipples??