Quick review of A7S and GH4

[jcs]

Quick Review A7S vs GH4

 

Where the  GH4 is better:

• Build quality
• Viewfinder
• Flip out screen
• User interface (by a mile)
• Battery life (by a mile)
• Color science
• Skin tones
• Rolling shutter (by quite a bit)
• Detail (internal 4K and 4K to HD)
• Noise grain (mostly pleasing monochromatic)
• Built-in flash for stills
• When lens size is included, a much smaller package
• Can shoot best quality on < 64GB Class 10 UHS-1 cards (A7S requires >=64GB cards for XAVC-S)
• Price

Where the A7S is better:

• Low light (it's very good however the noise grain is colored and multi-spectral; not as pleasing as GH4's mostly monochromatic fine-grain noise)
• Shallow DOF (add a SpeedBooster/focal-reducer to the GH4 and this isn't much of an advantage)
• Audio (no buzz; not tested thoroughly yet though)
• If a 5D3/full-frame owner: can use existing lenses with a non-focal-reducer adapter (with similar DOF)
• Super 35 mode allows use of many NEX lenses (some aliasing)
• Auto ISO

After having worked with 5D3 14-bit RAW and RED RAW, where the unmodified RAW looks pretty good straight from the camera, Slog2, Cinegammas, Cine-D/V, etc., start looking like gimmicks. None of those gamma curves are providing a real increase in dynamic range. In certain cases they may provide more information for 8-bit compression, however they can be tricky to use and must be exposed carefully. I've had the best luck shooting Natural with the GH4 (and no crazy curves), though iDynamic is helpful sometimes (appears to be local pixel group based processing vs. a simple curve. The A7S has a similar feature (disabled if any picture profile or effect is active, though)).

 

Rolling shutter is very challenging on the A7S in full frame mode (worse than the 5D3): not very usable for handheld without a lens with IS and/or a solid rig. The A7S menu and button system is a mess. The FS700 has a better design! (many folks find the FS700 non-intuitive to use).

 

Both cameras have aliasing and moire in slow motion. Perhaps similar to the FS700 in 240fps mode (not as bad as FS700 480+fps modes). I haven't done a direct comparison, however the FS700 120fps and less slow motion is better than the GH4 and A7S (all slow motion modes).

 

While the A7S isn't quite a one-trick-pony (low light), the GH4 is a much better all around camera. I haven't had the A7S as long as the GH4, so there may be other elements that make the A7S more compelling, yet to be discovered.

[Inazuma]

Very good out line. Wonder if you can test the battery life of the a7s? Just leave it recording until it turns off.

Also, does the gh4 picture degrade at 60fps? Or you mean above that?

[64mulford]

Great comparison on paper. Looking forward to hearing and seeing more.

[jgharding]

Great post, this is what I like forum communities for!

 

I'm considering both, I'm leaning to A7s at the moment as I like low light, s-Log 2 a lot more than cinelike-d, and I prefer the colour so far. Seems like there are quite a few caveats to using the A7s though... hmm...

[gelaxstudio]

Oh come on !

I thought you drawed your conclusion by some tests or videos review! 

How can you judge something that you did't even test it ?!

 

There are some points you mentioned that I disagree with you :

 

The picture profiles on A7S is way more advance,you can change internal color performance of the camera.Simulate the canon-like colour is also possible,while GH4 can't do as much as A7S can.

So  I don't think the Color science and skin tones are advantages on the GH4. Actually the color performances on the A7S is not bad!

 

Moreover, A7S has much better dynamic range,people hardly see the differences between F55 and A7S in terms of DR! (see this video :

[Jacek]

Quick Review A7S vs GH4

 

Where the  GH4 is better:

• Noise grain (mostly pleasing monochromatic)

Where the A7S is better:

• Low light (it's very good however the noise grain is colored and multi-spectral; not as pleasing as GH4's mostly monochromatic fine-grain noise)

 

That's interesting. From test footages I thought it's the opposite (noise grain). Have to look closer.

And what about slog2? Again, test footages looked like there is significant difference in DR..

[pablogrollan]

Isn't it kind of contradictory to say the A7S is better in low light but that the noise is better in the GH4?

 

From what I've seen, the GH4 gets quite noisy at 1600 ISO, making it the highest usable ISO, while the A7S hardly shows any noise at 6400 ISO... Are you comparing the "quality" of the noise at 1600 on the GH4 vs. the A7S at 51000 ISO? Kind of unfair, don't you think?

[jcs]

We're shooting with the 5D3, FS700, GH4, and now A7S. My review is based on actual camera usage, not other's reviews.

One of those cameras (not the 5D3) shot this: '?do=embed' frameborder='0' data-embedContent>>

 

I'm sure 50Mbit/s XAVC on the A7S is better to grade than 24Mbit/s AVCHD on the FS700 (both are H.264- the primary difference is bit rate in this case (both 8-bit 420). The FS700 has much better slomo.

 

Regarding Slog2 and DR- no gamma curve can help DR. All they can do is redistribute stops, so that in some cases we can count more visible stops. Actual DR won't change (noise floor to sensor clip). I'm looking for accurate color representation and I've already seen strange green/yellow in mid-tones with Slog2 on skin. To be fair, I don't use Cinelike-D on the GH4 due to the strange color handling (Natural is looking very good).

[jcs]

Isn't it kind of contradictory to say the A7S is better in low light but that the noise is better in the GH4?

 

From what I've seen, the GH4 gets quite noisy at 1600 ISO, making it the highest usable ISO, while the A7S hardly shows any noise at 6400 ISO... Are you comparing the "quality" of the noise at 1600 on the GH4 vs. the A7S at 51000 ISO? Kind of unfair, don't you think?

 

The GH4 does have more noise, and I don't use it above ISO 1600. Noise isn't always a bad thing, especially for 8-bit material. It helps reduce banding, acting like dither. The GH4's noise looks tuned to behave like dither: mostly fine and monochromatic. The A7S noise is typical colored noise, and IMO doesn't look very good. In studio shoots, we have control of lighting, so the GH4 does very well. We purchased the A7S for situations where we can't control the lighting.

 

We're still testing the A7S, however so far I haven't seen any low-light A7S material that wouldn't be helped with Neat Video. YouTube/Vimeo tests folks are seeing online are effectively noise-reduced by low bitrate H.264 compression.

 

Regarding XAVC-S and IPB: XAVC-S is an IPB codec on the A7S, the same as the GH4's H.264 variant (which also offers ALL-I for 1080p). XAVC-S is another flavor of H.264 (mostly a bitrate bump and spec support for 4K, 422, 444, and 10+ bits per pixel (not used by the A7S)).

[QuickHitRecord]

Thanks for sharing your observations thus far. Do you think that you could you could post a lens cap ISO test comparison between the two cameras? 

[maxotics]

jcs, great review!  However, like others, I find it very difficult to believe that the full-frame, large sensel design of the a7S wouldn't show significantly better color saturation over a MFT sensor camera at anything above 1600 ISO.  I understand if you prefer higher ISO images to go gray, so to speak, rather than guess color, but for those needing to match color I would think the a7s would be the best option, by a mile.  

 

You mention applying Neat to the GH4 footage to get it to match the a7s, but the real question is can you apply Neat to a7s footage to get it to match ISO 400, say?

 

Also, DOF is no small matter, at least to me.  You mention that, but it doesn't seem to rank high as the menu stuff, build quality, etc.  The kind of lenses you can put on the a7s is no small thing.  

 

But again, great review!  As you say, you're still working on it! :)

[GMaximus]

None of those gamma curves are providing a real increase in dynamic range.

They do
 

While the A7S isn't quite a one-trick-pony (low light), the GH4 is a much better all around camera.

"Who said it to be so? Let the sword resolve it! Contest! And God be the judge!" (which means we want side-by-side footage :) )

[sanveer]

This discussion is getting very interesting. Btw, whoever is doing this testing, could you not make the picture profile too flat, so that it does not come apart while grading. Also, in all the videos that I have seen of the GH4 and the Mark iii, I feel, its dynamic range is about 2 stops or more, than the Canon 5D Mark iii in video. In stills too, it does seem to resolve noticeably greater dynamic range.

[pablogrollan]

The GH4 does have more noise, and I don't use it above ISO 1600. Noise isn't always a bad thing, especially for 8-bit material. It helps reduce banding, acting like dither. The GH4's noise looks tuned to behave like dither: mostly fine and monochromatic. The A7S noise is typical colored noise, and IMO doesn't look very good. In studio shoots, we have control of lighting, so the GH4 does very well. We purchased the A7S for situations where we can't control the lighting.

 

We're still testing the A7S, however so far I haven't seen any low-light A7S material that wouldn't be helped with Neat Video. YouTube/Vimeo tests folks are seeing online are effectively noise-reduced by low bitrate H.264 compression.

 

Regarding XAVC-S and IPB: XAVC-S is an IPB codec on the A7S, the same as the GH4's H.264 variant (which also offers ALL-I for 1080p). XAVC-S is another flavor of H.264 (mostly a bitrate bump and spec support for 4K, 422, 444, and 10+ bits per pixel (not used by the A7S)).

 

I'd like to see how the GH4 at max ISO holds against this... Maybe under ideal conditions and low ISOs, GH4's noise is more grainy, less blocky, etc. (I have no reason not to trust your judgement) but the truth is that many recent cameras -even GM1, Sony A600, etc.- are relatively noise free at 800 ISO. It is at high ISOs where noise starts to become a problem, and I believe that problem happens way sooner on the GH4.

 

The same could be said of GH4 vs. BMPC 4K noise. It is much more "filmic" and organic on the BMPC... up to 400 ISO. After that, who cares, the BMPC footage becomes unusable while the GH4 still has headroom...

[Quirky]

"Who said it to be so? Let the sword resolve it! Contest! And God be the judge!"

 

...and because the virtual swords of verbosity are non-lethal, I predict a high page count for this thread.  :P

 

Meanwhile, I tend to agree with Philip Bloom, get both, and make use of their respective strengths, and work around to cover each others' weaknesses. Budget permitting, of course. Not much point in bickering which one is better, because it's always more or less subjective. The numerous online 'test videos' won't change that.

[jgharding]

I don't think discussion about noise between the two makes a lot of sense really. Look at James Miler's ISO 80,000 clip, no micro 4 3rds camera can do this that i know of. We're talking about physics as well as processing, it has four times the light gathering area. You can gain a stop or so with a speed booster, but not 3 or 4!

 

True, log doesn't actually change the chip, and it's hard to expose, but it does have the effect of protecting highlights if you use it right

 

EDIT: ah that video has been posted above while i was writing..

[Michael1]

Oh come on !

I thought you drawed your conclusion by some tests or videos review! 

How can you judge something that you did't even test it ?!

 

There are some points you mentioned that I disagree with you :

 

The picture profiles on A7S is way more advance,you can change internal color performance of the camera.Simulate the canon-like colour is also possible,while GH4 can't do as much as A7S can.

So  I don't think the Color science and skin tones are advantages on the GH4. Actually the color performances on the A7S is not bad!

 

Moreover, A7S has much better dynamic range,people hardly see the differences between F55 and A7S in terms of DR! (see this video :

>

)

 

While GH4 can't even defeat C100 in terms of DR!(A video review done by Digital Screen Imagination can proved that)

 

In a A7S review from  Newsshooter.com,the author W. Ashley Maddox wrote :

It(A7S) had a natural grain look to it vs. the muddy blacks that I am seeing in the GH4. Compared to the FS700’s AVCHD, the XAVC codec is way easier to grade

In a GH4 review done by cinema5d,they wrote: 

When you move to 800 ISO and above the footage becomes blocky (noisy + codec), especially on flat single surface elements.

 

So Noise grain on the GH4 is bad ,the same issues won't happen on the A7S , I think mainly because GH4 use IPB ,and it is a bad video codec for 4K~

Same issues are also mentioned in  related topics from EOSHD Forum

 

Obviously ,the XAVC-S works better than the IPB 

[Nikkor]

We're talking about physics as well as processing, it has four times the light gathering area. You can gain a stop or so with a speed booster, but not 3 or 4!

Stops are exponencial, 1 stop more is 2 times the light, 2 stops is 4 times the light. The difference between FF an m43 is 2 stops, if you get a 0.5 speedbooster the difference will be 0.

I'm not really following all the reviews on the a7s, but it seems that it's only better than other FF cameras after iso 6400, and iso 6400 looks like shit on every camera so... (this is in stills, in video the a7s will look a lot better because it oversamples from 4K to 1080p, that alone is a 2stop noise advantage over other FF cameras, but you can do the same with a GH4)

[johnnymossville]

I'm not taking anything away from the A7S. I think it's ground breaking in low light.  but I'm seeing a lot of footage where a fast lens on a lesser camera can get pretty much the same shot, with a bit more noise of course. For inside shots we have things called lights.   Still, it's a great cam, I get it. 

 

All things considered, the big drawback is no internal 4k.   Sony will probably bring it next year, and if I get an A7S now I'll be laying out another $2500 next  year for that model.  4K ready just isn't 4k now is it?

[rak_heri]

Which one has the best dynamic range? In the comparison, it seems to assume both GH4 and A7S have same dynamic range.