Sony A7S II is out!

[sunyata]

Well since most people can only perceive about 1 million colors and only one person has been discovered that had the wetware to perceive more, it might not matter.

 

If you have 8-bit systems, you can install all the log that you want, but you will always have 8-bit, 10-bit point was needed,  you can always correct it shortly with 8-bit color.

8 bit = 16,777,216 possible colors

10 bit = over a billion possible colors

Just to address both of these comments, even though they are disagreeing, the total number of colors per pixel is kind of a useless metric. Nobody will ever ask you "Hey, can you save that file out 1,073,741,824 colors please?", they'll just ask for 10bit, 12bit etc. This is because the total number of possible colors per pixel is not only impossible to remember, is not really what determines noticeable quality issues; that's contrast. You get 256 or less for 8bit, 1024 for 10bit etc. 

[maxmizer]

 

Proprio per affrontare entrambi questi commenti, anche se sono in disaccordo, il numero totale di colori per pixel è una specie di una metrica inutile. Nessuno potrà mai chiederà "Ehi, puoi salvare il file fuori  1,073,741,824 colori per favore?», Ti basta chiedere per 10bit, 12bit ecc Questo è perché il numero totale di colori possibili per pixel non è solo impossibile da ricordare, è non proprio ciò che determina notevoli problemi di qualità; questo è il contrasto. È possibile ottenere 256 o meno per 8bit, 1024 per 10bit etc. 

 

I just said you can put any curve log ( log2, log3...log65), but you will always have a small palette at 8bit, that is 256 colors X 255  gradations.

Do not you EVER do the grading of a movie on a par with a 10-bit as for example GH4 (from hdmi ) or other 10bit camera.

This bad even on FS5 who 10bit internal but saves 8-bit files (if I understand it).

I want 10-bit, because I do not want to see white patches in the sky..OK

[kaylee]

The slow motion looks really soft

[shemster]

The A7SII does seem like a great camera, although I do agree with you on some parts of your assessment. 

I think many reviews of the original camera were way off the mark concerning "clean low-light" at high ISO's. Even at ISO 12,800 I find it to be only just usable with very heavy post noise reduction. 

With additions like 4k, full HD 120fps, stabilisation and Slog3 - it's worth it for new buyers. For previous buyers, I'd hold on for the 3rd version (probs out in a year!). 

For professional video shooting, the FS5 seems like a much better proposition than all the A-series cameras. The extra money is worth it. 

Yes, Looking at the specs, it is a worthy upgrade of my c100 which I have been using for almost 3 years. I was expecting 4k for c100 mk2 but it is a huge disappointment when it doesn't. For the money I'm going to throw in, it has to be good enough to be my main workhorse for the next 3 years and FS5 seems to fit the bill currently.

[Don Kotlos]

I think as slog2, that the limitation to 8 bit it will see ...

I know that the issue is a mirror for larks...

There were two problems with SLOG2/SGAMMUT. First is limited space with 8 bits that can lead to banding and unattractive tonal gradations and the second was SGAMMUT that with its " twisted" colorspace required very fine color tuning. 

I am not that sure slog3 will help for the first problem but the sgammut3.cine will require less color manipulation and thus it might result in less problems. We will have to wait...

[Flynn]

Love Enzo

 

[Emanuel]

more than 10 bit versus 8 bit?

People are too focused on 10-bit versus 8-bit, when the-real-difference between 4:2:2 and 4:2:0 is higher than 10-bit 4:2:2 vs 8-bit 4:2:2.

[Emanuel]

 

Im not talking about anything that comes from post. im talking about camera output. more on that later

 

just talkin about in camera color rn tho

nothing could be further from the truth. i want a high level of color accuracy for reproducing sets, costuming, and prop elements (which were very expensive and time consuming to make). i want nature to look like nature, i dont want to have to avoid shooting something as common as a F------ BLUE SKY for any reason, i want people to look awesome, and healthy, like actual living creatures with blood running through their veins

fyi im capable of being extremely critical of the alexa, everybodys favorite movie camera. im not a mark; i know what i want and im gonna keep bitching until i get it, thats my job as a consumer. after TONS of bitching from us all about slog2 on the a7s, heres slog3 on its successor –– no coincidence. anyway, i digress, just saying that youre a bit unclear on what im "looking for"

 

these video reinforce my point: ive seen em, not impressed. at all. im not into retro luts or affected color grading, thats a fine option –– but i described above what i want from a digital video camera

but then again –– some feature filmmakers are going to map all their colors to teal and orange in their edit, and for them this shit doesnt rlly matter too much. slog2 is great for black and white work ive made some beautiful stuff

 

 

So, we have here a very demanding viewer... LOL : )

You are right. This or that camera can give you less than what you're looking for.

But.

There's no 'pictures' without post. This doesn't exist. And if exists, you can't complain. Much effort on post is mandatory. And that is the territory where you can provide the deterritorialization for reterritorialization* of your material, as you wish (e.g. that Japanese sample/video). For a simple reason, flat is higher because comprehends the wider. Hence the fashionable logs nowadays. Straight out of the box means nothing. What are we expecting for? Godard said motion picture is the truth 24 frames a second. He was right but only in a half. The other half is that actually is a big lie whatever frames you decide to tell.

How can't they give you what you're looking for? Trust me, you're just looking for a decent capture device and post. It is useless to say I like this or that camera in a way or another. The best you can say to yourself is if that camera is decent for your purpose/proposal or not. Believe me that you'll find a lot of decent cameras in the track. The trouble is when we blame technology when this one is already available. It only requires much work to find out the best balance and know-how from all of it. No teal + orange stamp (others would call it look of their own which makes them happy with) will affect the countless options on your choices of and for your set(up).

BTW, Ed is very good & versatile.

E :-)

 

*

https://en.wikipedia.org/wiki/Deterritorialization

https://en.wikipedia.org/wiki/Reterritorialization

[TheRenaissanceMan]

People are too focused on 10-bit versus 8-bit, when the-real-difference between 4:2:2 and 4:2:0 is higher than 10-bit 4:2:2 vs 8-bit 4:2:2.

I've heard people touting that on the forum like it's an established fact, but no one's ever given me a single example that demonstrates that 4:2:2 is more important than 10-bit. My eyes, the articles I've read, and the math, on the other hand, tell me that 10-bit is a way bigger quality difference.

[Emanuel]

I've heard people touting that on the forum like it's an established fact, but no one's ever given me a single example that demonstrates that 4:2:2 is more important than 10-bit. My eyes, the articles I've read, and the math, on the other hand, tell me that 10-bit is a way bigger quality difference.

Who said 4:2:2 is more important than 10-bit? The difference between 8-bit 4:2:2 and 8-bit 4:2:0, it is more significant for your grading. Much different statement. Of course, for certain type of grading. No mention of raw or alike where even 10-bit is short by default. People too much focused on the 10-bit holy grail tend to forget ; ) But, banding was the topic (hence my "the-real-difference" of my post; numbers help but like machines need i-n-t-e-r-p-r-e-t-a-t-i-o-n :-D) :

http://www.eoshd.com/comments/topic/9390-sony-a7s-ii-is-out/?do=findComment&comment=106562

Speaking of math, let's see with a very simple case: 1) 100,000; 2) 1,000,000; 3) 10,000,000. Where's the bigger difference? Now, take these figures as bucks and your single expense as any number between 100,000 and 1,000,000. Which step will make the whole difference?

Here goes your single example on pictures, posted and reposted more than once:

http://www.eoshd.com/comments/topic/9355-canon-announces-development-of-8k-cinema-eos-camera-and-120mp-dslr/?do=findComment&comment=105935

[gelaxstudio]

Which camera are you talking about, the a7s2 or r2?  The r2 doesn't do 1080p120 and the a7s2 is contrast detect only.  It's "binning" by the way. 

 7RII and 7SII both~under $3200 price range~any other camera to compare？

[maxmizer]

 7RII e 7SII entrambi ~ sotto $ 3200 fascia di prezzo ~ qualsiasi altra fotocamera a confrontare?

In the market of that price range there is nothing ...
all other competing brands are at a much lower price ...
(With 4K video) nothing can compare, if you do not need full frame or low light

It buys another brand much cheaper if you do not need these two specifications.

[sunyata]

Who said 4:2:2 is more important than 10-bit? The difference between 8-bit 4:2:2 and 8-bit 4:2:0, it is more significant for your grading. Much different statement. Of course, for certain type of grading. No mention of raw or alike where even 10-bit is short by default. People too much focused on the 10-bit holy grail tend to forget ; ) But, banding was the topic (hence my "the-real-difference" of my post; numbers help but like machines need i-n-t-e-r-p-r-e-t-a-t-i-o-n :-D) :

http://www.eoshd.com/comments/topic/9390-sony-a7s-ii-is-out/?do=findComment&comment=106562

Speaking of math, let's see with a very simple case: 1) 100,000; 2) 1,000,000; 3) 10,000,000. Where's the bigger difference? Now, take these figures as bucks and your single expense as any number between 100,000 and 1,000,000. Which step will make the whole difference?

Here goes your single example on pictures, posted and reposted more than once:

http://www.eoshd.com/comments/topic/9355-canon-announces-development-of-8k-cinema-eos-camera-and-120mp-dslr/?do=findComment&comment=105935

Oh my, I never saw that one. There is a little flaw in his test that probably explains the huge difference between the demonstrated quality of 8bit 4:2:0 vs 8bit 4:2:2: he's recording externally to the Shogun with the 8bit 4:2:2 and internally to the GH4 for the 4:2:0. I'm gonna bet a dollar that's why they look so different. Maybe it's a data rate thing, maybe its smoothing happening with the Shogun, Maybe it's the GH4's implementation of 8bit 4:2:2, I don't know. Below is just a radial saved as 8bit dpx, then two conversions to respective chroma subsampling schemes. One is yuv420p 8bit and the other is yuv422p 8bit,  Same very high h264 constant bitrate settings for both. Re-imported, gamma curve applied and saved out as a sRGB png.

 

 

[gelaxstudio]

In the market of that price range there is nothing ...
all other competing brands are at a much lower price ...
(With 4K video) nothing can compare, if you do not need full frame or low light

It buys another brand much cheaper if you do not need these two specifications.

I know Iphone6S is much cheaper and shoot 4K and 240P video ，but tool for work？

I think the A7RII or A7SII are the best tools for work under $3200，hardly see other camera can make such good balance between image quality and efficiency，BMD、GH4、C100 all with obvious disadvantages

[Shield3]

 7RII and 7SII both~under $3200 price range~any other camera to compare？

There must be a language or comprehension problem.

Fact:  The a7r2 only does 720p 120 FPS, not "FHD 1080".

Fact:  The a7s2 is reported to only do contrast detect AF, not phase detect.

You mentioned a camera that did all of these.  One does not exist.

Your words:  "Overpriced becaus no any other competitor can do what it do~ 14 stops DR，120P FHD without banning ，body IS，super lowlight  and  Phase Detection AF on canon 、nikon lens！LOL！

Again.  "120P FHD without banning" = The a7s2 reporting does do 1080p120 without "binning".  The a7r2 does not.

"Phase Detection AF on canon , nikon lens! LOL!" = The A7s2 is contrast detection only AF.  The A7r2 does do this though.

Not sure why I'm bothering here...

[animan]

IBIS at 4k?

[paulraymaekers]

Oh my, I never saw that one. There is a little flaw in his test that probably explains the huge difference between the demonstrated quality of 8bit 4:2:0 vs 8bit 4:2:2: he's recording externally to the Shogun with the 8bit 4:2:2 and internally to the GH4 for the 4:2:0. I'm gonna bet a dollar that's why they look so different. Maybe it's a data rate thing, maybe its smoothing happening with the Shogun, Maybe it's the GH4's implementation of 8bit 4:2:2, I don't know. Below is just a radial saved as 8bit dpx, then two conversions to respective chroma subsampling schemes. One is yuv420p 8bit and the other is yuv422p 8bit,  Same very high h264 constant bitrate settings for both. Re-imported, gamma curve applied and saved out as a sRGB png.

 

 

please show the same test in colour? :-)

[Emanuel]

please show the same test in colour? :-)

Very true, on 4:2:2 vs 4:2:0... those figures mean color!

In any case, don't forget we're speaking about banding, not exactly full grading...

[KarimNassar]

Below is just a radial saved as 8bit dpx, then two conversions to respective chroma subsampling schemes. One is yuv420p 8bit and the other is yuv422p 8bit,

the reason there is no difference is because you are using a grayscale image so there is only luma information in your image so nothing gets subsampled.

It is the foundation of how chroma subsampling works:

"Chroma subsampling is the practice of encoding images by implementing less resolution for chroma information than for luma information, taking advantage of the human visual system's lower acuity for color differences than for luminance"

only color information gets discarded when using chroma subsampling, in your images there is none.

4:2:0 4:2:2 4:4:4

^ first number is the luma component, as you can see there is no reduction no matter the chroma subsampling.

[sunyata]

It's true that some colors degrade more than others, but because the source radial was generated in RGB space and converted to Y'CbCr, you will still see chroma subsampling artifacts (it still get's subsampled in Cb and Cr to compose the final image). I was comparing with the other test, which looks grayscale, but in the past I've preferred to test this with random animated colors just to see where most breakup happens. I've updated the image to use a lot of red, typically the worst color.