Archive 2013 · Sony A7 - first impressions and some tests
p.11 #1Sony A7 - first impressions and some tests
The dynamic range is impressive. How many stops do you suppose it is in total?
Nov 27, 2013 at 11:10 PM
p.11 #2Sony A7 - first impressions and some tests
All the Sony sensors are about the same, except the SLT a99 - around 14 stops at base ISO. RX1 is 14.3 so that is a good estimate for the a7 - DxO data.
Nov 27, 2013 at 11:49 PM
p.11 #3Sony A7 - first impressions and some tests
That would indicate a 14 bit A/D on the sensor, compared to the usual 10-12 bit. Very impressive.
Nov 28, 2013 at 12:40 AM
p.11 #4Sony A7 - first impressions and some tests
zmix wrote:
That would indicate a 14 bit A/D on the sensor, compared to the usual 10-12 bit. Very impressive.
you don't need X Bit to have a DR of XEV, the Nex camera have somwhat over 13 with 12 bit.
About the advantage: Look at the images an judge for yourself, i think it is more than 2 stops.
DxOMark would suggest only 1 stop.
That would indicate a 14 bit A/D on the sensor, compared to the usual 10-12 bit. Very impressive.
you don't need X Bit to have a DR of XEV, the Nex camera have somwhat over 13 with 12 bit.
About the advantage: Look at the images an judge for yourself, i think it is more than 2 stops.
DxOMark would suggest only 1 stop.
Nov 28, 2013 at 01:36 AM
p.11 #5Sony A7 - first impressions and some tests
Technically the resolution of the A/D determines the dynamic range. Each additional bit doubles the dynamic range, in other words, increases it by one stop. However there may be usable signal below the noise floor, so by using DSP the range may appear to be greater. (And there is always DSP affecting the signal from the A/D converter). The phototransistors on the sensor are the source of the noise in the image, each one represents one pixel (in the undecoded RAW file). The noise they generate is known as "Brownian Noise" and is essentially random. Since the noise is random and the voltage generated by the illumination of each pixel (over time) is not, the noise does not add linearly, but the "signal" does, so even though sensor with more pixels will generate more noise, the signal level is proportionately higher and when that signal is scaled to a usable range the noise floor decreases significantly.
I'd like to see a dynamic range comparison between the "24mp" Alpha 7 and the "36mp" Alpha 7r.
I'd like to see a dynamic range comparison between the "24mp" Alpha 7 and the "36mp" Alpha 7r.
Nov 28, 2013 at 09:21 AM
p.11 #6Sony A7 - first impressions and some tests
zmix wrote:
However there may be usable signal below the noise floor
It isn't a very impressive noise floor if there is a decodable signal in there.
However there may be usable signal below the noise floor
It isn't a very impressive noise floor if there is a decodable signal in there.
Nov 28, 2013 at 03:00 PM
p.11 #7Sony A7 - first impressions and some tests
carstenw wrote:
It isn't a very impressive noise floor if there is a decodable signal in there.
In audio, where all of the research is now dominated by A/D and D/A conversion and DSP, it's been proven that signals can be recovered 30dB below the noise floor (that's ~= to 5 additional bits).
It seems that photography is headed in the same direction, so the narrative arc is somewhat predictable.
zmix wrote
However there may be usable signal below the noise floorIn audio, where all of the research is now dominated by A/D and D/A conversion and DSP, it's been proven that signals can be recovered 30dB below the noise floor (that's ~= to 5 additional bits).
It seems that photography is headed in the same direction, so the narrative arc is somewhat predictable.
Nov 28, 2013 at 03:11 PM
p.11 #8Sony A7 - first impressions and some tests
It's all relative in the numbers, I meant to add last post, Phillip, but the first images really show the benefit of the extra DR, you might agree.
Sony's approach to noise is now very 'photographic' in essence, they worry much less about smoothed out noise at any ISO, instead they want to provide the ability for photographers to produce fine images, recognising that noise is akin to film grain - a large part of the aesthetic.
On matters noise, I read this piece this morning re the a7:
'..(It) isn't the best full-frame sensor for low-light by any stretch. With the likes of the Canon EOS 5D MkIII out there - which is admittedly far more cash, and a lot bulkier..'
DxO don't yet have the a7 data released, but it is a safe bet it will be near identical to the a7r, which clearly waxes the 5DIII at, wait for it,...ISO 25600 in SNR and all other measures, at virtually identical actual ISO (a7r: 18357, 5DIII: 18589). SNR was 22.9 vs 21.3...what fine cameras these Sonys are...as they are so clearly superior at low ISO levels. The DR graph is the kind of narrative arc I can sure live with.
Sony's approach to noise is now very 'photographic' in essence, they worry much less about smoothed out noise at any ISO, instead they want to provide the ability for photographers to produce fine images, recognising that noise is akin to film grain - a large part of the aesthetic.
On matters noise, I read this piece this morning re the a7:
'..(It) isn't the best full-frame sensor for low-light by any stretch. With the likes of the Canon EOS 5D MkIII out there - which is admittedly far more cash, and a lot bulkier..'
DxO don't yet have the a7 data released, but it is a safe bet it will be near identical to the a7r, which clearly waxes the 5DIII at, wait for it,...ISO 25600 in SNR and all other measures, at virtually identical actual ISO (a7r: 18357, 5DIII: 18589). SNR was 22.9 vs 21.3...what fine cameras these Sonys are...as they are so clearly superior at low ISO levels. The DR graph is the kind of narrative arc I can sure live with.
Nov 28, 2013 at 04:11 PM
p.11 #9Sony A7 - first impressions and some tests
zmix wrote:
Technically the resolution of the A/D determines the dynamic range. Each additional bit doubles the dynamic range, in other words, increases it by one stop. However there may be usable signal below the noise floor, so by using DSP the range may appear to be greater. (And there is always DSP affecting the signal from the A/D converter). The phototransistors on the sensor are the source of the noise in the image, each one represents one pixel (in the undecoded RAW file). The noise they generate is known as "Brownian Noise" and is essentially random. Since the noise is random...Show more →
it sounds to me like you are completely ignoring shot noise here?
Technically the resolution of the A/D determines the dynamic range. Each additional bit doubles the dynamic range, in other words, increases it by one stop. However there may be usable signal below the noise floor, so by using DSP the range may appear to be greater. (And there is always DSP affecting the signal from the A/D converter). The phototransistors on the sensor are the source of the noise in the image, each one represents one pixel (in the undecoded RAW file). The noise they generate is known as "Brownian Noise" and is essentially random. Since the noise is random...Show more →
it sounds to me like you are completely ignoring shot noise here?
Nov 28, 2013 at 05:06 PM
p.11 #10Sony A7 - first impressions and some tests
For the purposes of my explanation, there is really no need to distinguish one type of noise from another because they are all random in frequency and will not add when the individual pixels are combined.
Nov 28, 2013 at 06:10 PM
p.11 #11Sony A7 - first impressions and some tests
zmix wrote:
For the purposes of my explanation, there is really no need to distinguish one type of noise from another because they are all random in frequency and will not add when the individual pixels are combined.
you have to add a lot of pixels together to make them all cancel out i'm afraid (and use appropriate software since this needs to be done after the image is off the camera), and there are plenty of images that will not get downscaled and may even get upscaled.
For the purposes of my explanation, there is really no need to distinguish one type of noise from another because they are all random in frequency and will not add when the individual pixels are combined.
you have to add a lot of pixels together to make them all cancel out i'm afraid (and use appropriate software since this needs to be done after the image is off the camera), and there are plenty of images that will not get downscaled and may even get upscaled.
Nov 28, 2013 at 08:36 PM
p.11 #13Sony A7 - first impressions and some tests
Phillip,
What was your methodology for these tests?
What was your methodology for these tests?
Nov 29, 2013 at 09:15 AM
p.11 #14Sony A7 - first impressions and some tests
more full resolution samples, developed from raw using LR 5
Minolta_MD_35mmf28_f8 (1 von 1)-2 by reevedata, on Flickr
Minolta_MD_35mmf28_f8 (1 von 1) by reevedata, on Flickr
Canon_FD_80200mmf4L_at200mmf4 (1 von 1) by reevedata, on
Canon_FD_80200mmf4L_at200mmf11 by reevedata, on FlickrFlickr
Sigma_24mmf28_atf11 by reevedata, on Flickr
Minolta_MD_35mmf28_f8 (1 von 1)-2 by reevedata, on Flickr
Minolta_MD_35mmf28_f8 (1 von 1) by reevedata, on Flickr
Canon_FD_80200mmf4L_at200mmf4 (1 von 1) by reevedata, on
Canon_FD_80200mmf4L_at200mmf11 by reevedata, on FlickrFlickr
Sigma_24mmf28_atf11 by reevedata, on Flickr
Dec 03, 2013 at 12:08 PM
p.11 #15Sony A7 - first impressions and some tests
my local camera shop has one left in stock. I went to mess around with it unfortunately the battery was dead so I didn't get a chance to actually play with the camera but I think I may buy one. A7 or A7r not sure yet but i liked the feel and look of the camera. I think I'll buy one around the time they release the 24-70mm f4
Dec 03, 2013 at 12:30 PM
p.11 #16Sony A7 - first impressions and some tests
I haven't read all 11 pages (stopped at 5), so forgive me if this has been addressed, but with the Canon FD 35/2 samples, I'm very surprised. The FD 35/2 is quite a decent lens...far better than its EF counterpart (well, pre-IS). I say this because mine looks better than that with the Speed Booster. It's not fantastic in the corners wide open with the SB, but it's better than that.
Dec 03, 2013 at 01:00 PM
p.11 #17Sony A7 - first impressions and some tests
I have scrolles several times. To me it saeems low ISO noise is the same on NEX-6 and A7?
Dec 03, 2013 at 03:43 PM
p.11 #18Sony A7 - first impressions and some tests
H.Lux wrote:
I have scrolles several times. To me it saeems low ISO noise is the same on NEX-6 and A7?
Assuming the A7 has similar performance to the RX1, it'll be about a stop cleaner with about a stop better DR at base ISO over the NEX-6.
I have scrolles several times. To me it saeems low ISO noise is the same on NEX-6 and A7?
Assuming the A7 has similar performance to the RX1, it'll be about a stop cleaner with about a stop better DR at base ISO over the NEX-6.
Dec 03, 2013 at 03:50 PM
p.11 #19Sony A7 - first impressions and some tests
douglasf13 wrote:
Assuming the A7 has similar performance to the RX1, it'll be about a stop cleaner with about a stop better DR at base ISO over the NEX-6.
I think my test showed about 1.5 Stops better Noise Performance and more than 2 Stops better DR.
See this post: https://www.fredmiranda.com/forum/topic/1255132/9#11964147
H.Lux wrote
I have scrolles several times. To me it saeems low ISO noise is the same on NEX-6 and A7?I think my test showed about 1.5 Stops better Noise Performance and more than 2 Stops better DR.
See this post: https://www.fredmiranda.com/forum/topic/1255132/9#11964147
Dec 03, 2013 at 03:54 PM
p.11 #20Sony A7 - first impressions and some tests
okay, last week I met up with somewone from a German forum and ran an extensive test on many lenses.
I made the mistake to test the lenses by brand and not by focals length so the sections are not matched, sorry
.
edit: some images are too large, so the forum resizes them. All images which have a width smaller than 950px are resized. Just click on them to see them in full resolution on flickr.
Edited on Dec 22, 2013 at 05:24 PM · View previous versions
I made the mistake to test the lenses by brand and not by focals length so the sections are not matched, sorry
.edit: some images are too large, so the forum resizes them. All images which have a width smaller than 950px are resized. Just click on them to see them in full resolution on flickr.
Edited on Dec 22, 2013 at 05:24 PM · View previous versions
Dec 22, 2013 at 04:59 PM