Sony A7R III review - the BBC fixed Sony's colour!

[wolf33d]

On 1/14/2018 at 3:00 PM, Andrew Reid said:

To be fair, Sony 24-70mm F2.8 GM is a brick as well

24-70 F4 on the other hand... tiny. But is it any good? Anyone used it?

You can also have a look at the new 24-105. Nice versatility and generally good comments online. But I am not a big fan of F4 lenses so I have not used them on the Sony. 

18 minutes ago, Andrew Reid said:

Just got the Sigma 24-35mm F2.0, it's an amazing looking image

The big surprise is this...

With the Sigma MC-11 adapter it tracks AF in video mode like a Sony FE mount lens.

The Metabones adapter does not do this.

Going to try a few more lenses on the MC-11... The adapter is just superb with Sigma's own stuff on the A7R3

Now we are talking. 24-35mm F2.0 is great so are many Sigma lenses. I am very interested in getting more feedback on how those sigma lenses perform with continuous AF with the MC-11 VS Sony FE lens on the A7RIII. 

Also, Sigma announced that they have some Sigma native FE lens in development. First one is rumored to be 35mm. I like the fact that they come to this system because Sony FE pricing is rarely engaging...

[Andrew Reid]

The Sony FE lenses tend mostly to have the more quiet AF motors, but some of the high-end GM lenses do make a clicking sound.

The Batis 25 and 85 are great... and silent.

Very impressed with the Sigma 24-35mm...As well as world's fastest full frame zoom lens, the quality is as per Zeiss Batis 25mm F2 prime... Which is more expensive by the way, and only fits one mount... But with the Sigma, you are getting a 24, 28 and 35mm prime in one lens...that goes on A7R3, 1DX, 1DC, GH5, you name it... and a handy 50mm field of view when you switch to Super 35mm mode to boot. All at F2.0 and pin sharp, with Dual Pixel AF quality autofocus in 4K on the A7R3.

Most of the Sigma lenses I've tried so far do all make a noise when focussing though, so you won't want to use the onboard audio with video AF... Same situation as on a 1D X Mark II, just for the record  The 18-35mm seemed a lot quieter though, almost silent?!

It's just very very positive to finally have Video Autofocus with Canon mount lenses... Albiet just the Sigma ones. The Canon lenses DO NOT autofocus in video mode, and neither do Tamron or the older Sigma non-ART lenses. Throw them on the scrap.

When did they introduce this BTW? MC-11 has been out for ages but I haven't seen a word out of the other bloggers and sites about C-AF in video mode + Sigma lenses. Dickheads lol

Metabones still works better in stills mode with Canon lenses... but reliability is shit with Sigma ART. So you may end up having to get both adapters. At least the Sigma is competitive priced. Mine was 250 euros brand new.

[horshack]

4 hours ago, webrunner5 said:

I don't think there is Any dual native thing going on in Any Sony A7 series camera. There used to be the misconception that it had a different base ISO in Photo and Video. That was proven wrong also long ago. I would love to be proven wrong though. Yeah there is an extended ISO, that is not really Dual ISO's. Almost Every camera has that feature.

This was written a Long time ago on DPR. And Jim Kasson is a wiz at all this stuff and DR.

 

"JimKasson wrote:

JDLaing wrote:

What is the native ISO on the Sony a7II? 100 or 200?

Do you mean base ISO? In that case, it's 100.

I this I consider this sensor to have two "base ISO" - one for both conversion gains. ISO 100 is one of them, the other seems to be ISO 650.

I'm not sure how "base ISO" is or should be defined though, but for me it's usually been the point with the least gain amplification. And for this sensor (and the A7S) sensor there would be two such points

Let's define the base ISO of a digital cameras the ISO setting for which the gain at the output of the analog to digital converter (ADC), measured in counts per electron, is the lowest. In most cameras, it is the setting where the upper end of the linear region of the pixel voltage/electron count curve produces a full scale output from the ADC.

Let’s say that the full well capacity (FWC) of the a7RII when the ISO is set to 100 is 50,000 electrons. That’s close to the real value. You will see that the exact number doesn’t affect the conclusions of the calculation to follow. Full scale on the a7RII is 16372. But that is before the black point is subtracted. The nominal black point of the a7RII, like all the a7x cameras, is 512, so let’s say that full scale is 15860. You will see that the exact value is not important either, but it’s nice to keep things real. Thus, the gain, at ISO 100 is 15860/50000 = 0.317 counts per electron (you may be used to seeing the inverse of this number, or 3.15 electrons per count).

At ISO 500, just before the conversion gain is changed, the gain must be (500/100)*0.317 = 1.58 counts per electron.

By looking at the read noise curves here, you can see that the conversion gain change is a little over one stop. Actually, you can get the answer to the entire question by inspection of the top curve, but I’ll leave that analysis as an exercise for the interested student.

Let’s be generous and say that the conversion gain changes by a factor of three. So the gain that we’re looking at must be (640/100)*0.317/3 = 0.676. That is larger than the gain at ISO 100 by a factor of more than two, so the a7RII does not have two base ISOs.

With the a7S, it isn’t even close, since the conversion gain change happens at a much higher ISO.

Jim"

Jim and I converse all the time. The explanation you quoted from Jim doesn't have direct relevance to the discussion because "dual native ISO" isn't what the name implies - and btw there's no such thing as "native ISO" since conversion gain happens at all ISO levels. Again, it's simply Pany's name for the Aptina technology, which is a dual-gain configuration and has been implemented in multiple sensors prior to the GH5s, from both Sony (A7s, A7rII, A7rIII) and Nikon (D4 and forward).

3 hours ago, Don Kotlos said:

If the "native" term is used as loosely as Panasonic marketing team then yes that is correct More correct is that dual gain has been around for quite some time, used by many cameras.

Generally, native ISO should mean an ISO setting without any gain applied, but Panasonic defines two native ISOs as two base ISO settings that yield similar noise profiles:

The varicams get the closest to actually having dual native/base ISO, but the EVA1 (and I am guessing the GH5s as well) does not (http://zsyst.com/2017/12/panasonic-eva1-first-look/):

 

Whereas the noise profiles from Sony cameras come very close (A7s is even closer to true native ISO than EVA1...):

This tells you how well Panasonic plays the marketing game. 

 

"Dual gain" has been around for a while but it's not implemented in that many cameras, at least those using APS-C and larger sensors. Nikon first implemented on their sensor starting with the D4, and Sony with the A7s. Btw, gain is applied at all ISOs, including base ISO.

[Trek of Joy]

On 1/14/2018 at 2:00 PM, Andrew Reid said:

To be fair, Sony 24-70mm F2.8 GM is a brick as well

24-70 F4 on the other hand... tiny. But is it any good? Anyone used it?

I like my 24-70z, its weaknesses are well documented, but it covers a really handy focal range that until recently was really the only compact option in E-mount. And the plastic kit lens is much worse. Below was at 26mm on the a7r2, I cropped out the rope barriers on the bottom of the frame which also gets rid of the worst part - the corners. That being said - the 24-105 looks absolutely stellar, its everything I wish my 24-70z could be. If you can get a good copy the 24-70z's are going for cheap used these days. When I'm done traveling around the world, I'm promptly dumping my copy for the 24-105, as its so much better on the wide end and across the frame.

chris

[webrunner5]

3 hours ago, horshack said:

Jim and I converse all the time. The explanation you quoted from Jim doesn't have direct relevance to the discussion because "dual native ISO" isn't what the name implies - and btw there's no such thing as "native ISO" since conversion gain happens at all ISO levels. Again, it's simply Pany's name for the Aptina technology, which is a dual-gain configuration and has been implemented in multiple sensors prior to the GH5s, from both Sony (A7s, A7rII, A7rIII) and Nikon (D4 and forward).

"Dual gain" has been around for a while but it's not implemented in that many cameras, at least those using APS-C and larger sensors. Nikon first implemented on their sensor starting with the D4, and Sony with the A7s. Btw, gain is applied at all ISOs, including base ISO.

Ehh then you do you call the ISO that has NO Conversion Gain applied? Base ISO? Sort of what I would call it.

[horshack]

2 hours ago, webrunner5 said:

Ehh then you do you call the ISO that has NO Conversion Gain applied? Base ISO? Sort of what I would call it.

Every ISO has a conversion gain, including base ISO. If you read back over Jim's post you'll see him calculate the theoretical conversion gain for "base" ISO 100 on the A7rII.

[webrunner5]

38 minutes ago, horshack said:

Every ISO has a conversion gain, including base ISO. If you read back over Jim's post you'll see him calculate the theoretical conversion gain for "base" ISO 100 on the A7rII.

He is a Little above what I know LoL. Well maybe that I Want to know. I was a Ham Radio operator for years. So have a grasp of it, but not into intense thinking anymore.

I just try to remember if I have peed in the last 2 or 3 hours now. And I hate times changes, I loose track of When Lawrence Welk is on!

 

[Domi V Edwards]

@Andrew Reid  Sorry forgot to tag.  Just purchased the Procolor / pro log packet Sunday to use with my a7r iii. Do I have to repurchase once you update?

 

Thanks

[Andrew Reid]

Hey Domi. Just get in touch when the article about the update is out and I'll send you the new one. (The update is only for the A7R III)

[Oliver Daniel]

17 hours ago, Andrew Reid said:

Just got the Sigma 24-35mm F2.0, it's an amazing looking image

The big surprise is this...

With the Sigma MC-11 adapter it tracks AF in video mode like a Sony FE mount lens.

The Metabones adapter does not do this.

Going to try a few more lenses on the MC-11... The adapter is just superb with Sigma's own stuff on the A7R3

 

Awesome!

I also have the Sigma MC-11 adapter. 

How does the autofocus perform in stills with the Mc-11 adapter? 

Are there any greyed out AF options when using this adapter with photo or video? 

[Andrew Reid]

It works superbly for stills if you engage AF-C, which uses phase-detect AF.

In AF-S it seems to struggle. But single-point AF-C on a back button or half-press is like having AF-S anyway.

I found a similar thing with official Sony lenses and even the RX1R II, where AF-C was far superior to AF-S mode.

The Metabones adapter doesn't work at all with some Sigma lenses (was out of luck entirely with 35mm F1.4 ART!) but you should keep it for Canon stuff. I am not having much luck with Canon EF lenses on the MC-11. How about you Oliver?

Not tested Eye-AF yet but no greyed out options that I've noticed with a fully compatible Sigma lens on the MC-11.

I now prefer the A7R III to the GH5 and D850 purely for the amazing AF in video mode with Sigma lenses and Sony E-mount stuff. I have a lot of these lenses and it means I can sell a few Canon lenses when I get rid of my 1D X II.

Video AF on the GH5 is very much mediocre and had expected bigger leap with GH5S but not to be!

[Arikhan]

@Andrew Reid

Sigma Germany is about 1 hour far away from my house and it offers the possibility to test there every Sigma lens you want.

I tested the 35mm 1.4 and the 85mm 1.4 (Canon version) with MC-11 on the A7R iii and the A6500. Both lenses work superbly on both Sony cameras for video - AF is quite smooth. Eye AF (photography) works very nice too (with both cameras). In my eyes, both lenses are more responsive with the A7R iii than with the A6500....

The weight and form factor of the lenses (especially the 85mm 1.4) is not ideal, because - compared to the 2 Sony cameras - the lenses are HUGE...Without vertical grip not ideal to hold in your hands for a longer time when packed on mirroless Sony E-Mount...

But imagery is really exceptional...For my personal taste, the 35mm 1.4 offers a little bit of "neutral" colors, not so much microcontrast (for photography)...But that's splitting hairs in this league...

[Andrew Reid]

The Sigma lenses are all superb. As you point out the only thing I'd change is the size & weight.

The FE mount ART range should be fantastic.

But will they still be bang for buck?

[Arikhan]

When looking at the results you can get with them and depending on your needs (resolution, color rendition, considering CA and other aberrations, etc.) for sure. At least for photography...When talking film, there are people who don't want this level of high resolution and detail offered by Art lenses...

Another problem with Sigma lenses is their use with DSLRs: When trying to use Sigmas with eg Nikon, you have sometimes to try more copys to get them work with your camera - because of AF inconsistencies, even after calibrating lenses with USB tuning and internal AF fine justage within your camera. But - as said - this is only a problem for some DSLR cameras.

Please consider, the Sigmas are not perfect - some of them need serious correction in camera and post. No problem when talking RAW photography, but sometimes a bummer when talking film. Zeiss is mostly better, but there is another price point to consider too.

[Oliver Daniel]

22 hours ago, Andrew Reid said:

The Metabones adapter doesn't work at all with some Sigma lenses (was out of luck entirely with 35mm F1.4 ART!) but you should keep it for Canon stuff. I am not having much luck with Canon EF lenses on the MC-11. How about you Oliver?

I now prefer the A7R III to the GH5 and D850 purely for the amazing AF in video mode with Sigma lenses and Sony E-mount stuff. I have a lot of these lenses and it means I can sell a few Canon lenses when I get rid of my 1D X II.

Video AF on the GH5 is very much mediocre and had expected bigger leap with GH5S but not to be!

On the A6500 with MC-11, the Sigma ART's work well from what I remember - I'll give them another go as I've been using a Sony lens the past year for gimbal autofocus. 

I don't have any Canon EF's, just Sigma ART's and manual EF lenses. 

I wouldn't be too interested in the A7R line usually, however I'm being asked more and more to bundles photography with video these days. If the video autofocus performance is up there - that's a huge asset. Might rent it and give it a whirl. 

Thing is, I've been putting the GH5 on a pedestal and really invested my time and money into making it sing. So I guess my fear is hurting that really focused process. 

For stills, I do prefer Canon. They're not mirrorless, however the images look so satisfying in comparison to the "emptier" Sony /Panasonic image rendering. 

DECISIONS. 

[Andrew Reid]

12 hours ago, Arikhan said:

When looking at the results you can get with them and depending on your needs (resolution, color rendition, considering CA and other aberrations, etc.) for sure. At least for photography...When talking film, there are people who don't want this level of high resolution and detail offered by Art lenses...

Granted they do not look like vintage lenses in the rendering - quite uniform bokeh and very sharp.

However my Cooke S4 PL lenses are even sharper, so I guess film people DO need resolution

12 hours ago, Arikhan said:

Please consider, the Sigmas are not perfect - some of them need serious correction in camera and post. No problem when talking RAW photography, but sometimes a bummer when talking film. Zeiss is mostly better, but there is another price point to consider too.

What sort of correction would you do to videos shot with Sigma ART glass?

I don't do any. The distortion on the 35mm F1.4 for example is practically non-extistant.

33 minutes ago, Oliver Daniel said:

On the A6500 with MC-11, the Sigma ART's work well from what I remember - I'll give them another go as I've been using a Sony lens the past year for gimbal autofocus. 

I don't have any Canon EF's, just Sigma ART's and manual EF lenses. 

I wouldn't be too interested in the A7R line usually, however I'm being asked more and more to bundles photography with video these days. If the video autofocus performance is up there - that's a huge asset. Might rent it and give it a whirl. 

Thing is, I've been putting the GH5 on a pedestal and really invested my time and money into making it sing. So I guess my fear is hurting that really focused process. 

For stills, I do prefer Canon. They're not mirrorless, however the images look so satisfying in comparison to the "emptier" Sony /Panasonic image rendering. 

DECISIONS. 

Yes I can see why you'd not want to chop and change too much when the GH5 is working well and it's good to focus on something that works.

But full frame, good video PDAF even with Sigma EF lenses, 42MP stills, very good ISO 12,800 and two 4K field-of-views per prime lens thanks to S35 mode with instant switch-a-doo button, is quite a lot the GH5 lacks.

May want to wait to see what the A7S3 brings though if you need to replace the GH5 outright, with another 10bit beauty.

[Arikhan]

@Andrew Reid

24 minutes ago, Andrew Reid said:

What sort of correction would you do to videos shot with Sigma ART glass?

I don't do any lens correction to videos...But there are colleagues arguing to see some issues (in video) when comparing to the much more expensive comparable Zeiss lenses. When considering IQ, AF capabilities AND pricing, some Sigma Art primes are really hard to beat when used with Sony mirrorless cameras.

[Django]

On 1/16/2018 at 3:16 PM, Andrew Reid said:

 

The Metabones adapter doesn't work at all with some Sigma lenses (was out of luck entirely with 35mm F1.4 ART!) but you should keep it for Canon stuff. I am not having much luck with Canon EF lenses on the MC-11.

 

I remember trying out the MC-11 with Canon 35mm 1.4 L on A7R2 and getting pretty decent AF results to my surprise. I think it's definitely lens dependant (and there is a Canon EF compatibility list) what Canon lenses have you tested?

[Oliver Daniel]

On 17/01/2018 at 1:53 PM, Andrew Reid said:

Yes I can see why you'd not want to chop and change too much when the GH5 is working well and it's good to focus on something that works.

But full frame, good video PDAF even with Sigma EF lenses, 42MP stills, very good ISO 12,800 and two 4K field-of-views per prime lens thanks to S35 mode with instant switch-a-doo button, is quite a lot the GH5 lacks.

May want to wait to see what the A7S3 brings though if you need to replace the GH5 outright, with another 10bit beauty.

Yep, this is the problem. 

I love full frame. I love DPAF. I love cameras that are great in lowlight. I love great hybrids. 

Yesterday, I was hired to do some content for a new sports product. In the end, did more stills than video! (with a GH5!) 

I'm being really strict with new purchases - I willingly get another model / different camera / tool if it's ABSOLUTELY NECESSARY. 

The GH5 is doing it's job very well for the moment. 

[markr041]

I would like to return to two of the original premises of this thread - that Sony color is now "better" because you can shoot in HLG with the REC2020 color space and that HLG is easy to grade. Note that REC2020 and REC709 are both standards and both of these color spaces are available. That is, it has always been true that SONY accommodated color standards set by others (REC709). S-Gamut is their own, but not REC709.

First, I assume that we are still having to view video in REC709. If we view video in HLG, then shooting in HLG is great. But, few will do that. 

So, the fact that one can shoot in REC2020 color can hardly (easily or automatically) result in good REC709 color. REC709 color is a subset of REC2020, and REC2020 color does not nicely translate into REC709 color just because you view it on a REC709 screen. Nor is translating REC2020 to REC709 in post simple at all. Thus, it is not easy to grade HLG to make a video that looks good in REC709 unless you know the exact math that makes REC2020 translate to REC709.

Of course "looks good" is the issue here. Let's start with accuracy or realistic as one notion of looking good. REC2020->REC709 is not easy if one is looking to get the most accurate color for REC709 screens. Maybe someone likes the look of REC2020 as viewed in REC709. But to say that look is "better" is completely subjective.

If you prefer the gamma curve in HLG, you can play with that. But then why not combine the HLG gamma with REC709 color if the ultimate viewing is going to be in REC709?

OK, here is a comparison of the same scenes shot in HLG and in REC709. Four scenes. For each, the first was shot in HLG (HLG gamma and REC2020 Color) the second in Standard (REC709 gamma and color). If you watch this on a standard TV or monitor you can assess how HLG clips look viewed in REC709. If you watch this on a TV and *set* it to HLG, you can see the HLG clips as they are meant to be seen and get HDR. The HLG clips underwent no post-processing; the REC709 (Standard ) clips had to be re-rendered to merge with the HLG clips but no other post-processing was done. That is why there are no titles in the video - to avoid processing.

This was shot using the Sony FS5, which can also now shoot in HLG. I chose the standard one, not the Sony HLG variants. Does anyone think the HLG clips have better color?

I will later post a video that displays the transformed HLG clips to conform to REC709, using Catalyst Browse. One can see how ell HLG converted lokks in REC709.