Sony a7R II vs III

[Macrero]

I am considering buying a Sony a7R II, but the new mark III is (was) tempting me.

The only two advantages (to me) of the a7R III are the 14 bits RAWs in "silent mode" (in the Mark II RAWs drops to 12 bits, I could live with that) and the Pixel Shift. It seems like interesting feature, though truth be told, I am not sure if it really brings any significant improvement in resolution and detail.

So I downloaded the samples RAWs from the Dpreview's studio shot comparison site and for my surprise the a7R II with e-shutter actually shows more fine detail than the a7R III in pixel shift mode

*Pics removed due DPR copyright policy. You can download the RAW files from DPR if you want to see how they compares.

Am I missing something? Given that the III would cost me almost 3X than a used/good a7R II it is a no-brainer for me, but maybe there is something wrong with the Dpreview sample (though very unlikely).

If that's really the result of Sony pixel shift, I don't quite understand what's all the hype about Pixel shifting... I played also with an Oly RAW in pixel shift mode and while you get a 80 MP file, it comes out smeared and I was unable to get a satisfactory result even after a lot of sharpening and processing...


Best,

- Macrero

[Lou Jost]

Macrero, I don't know what's going on there, but many other tests show that Sony pixel-shifting works very well, though these tests may not have used electronic shutter for the un-shifted results, and this mistake would make it difficult to say how much of the improvement is due to pixel-shifting and how much is due to the use of the electronic shutter which eliminates camera vibrations. On the other hand, Ray on this forum showed that the Sony pixel-shifting introduced false detail in his tests.

You mention that you tested the Oly pixel-shifting and found it poor. That has to be due either to a bad camera sample or something moving in your set-up. I've taken thousands of hi-res 80Mp shots with the Oly and when vibration is controlled, it works perfectly and resolves features that can't be seen at all in the standard-res versions of the same photo. I can't stress enough how well this works, even for microphotography.

Vibration must be controlled to a much higher level when using pixel-shifting than when doing ordinary macro work. Movement during pixel-shifting produces sharp artifacts (false detail, usually patterned detail in the Oly), and stacking programs will pick up that false detail. In contrast, occasional environmental vibrations during capture of a normal stack might blur a frame, but that frame will often be ignored during stacking since it doesn't have sharp detail.

Another thing that has to be controlled better when using pixel-shifts is rail drift. Very slow drift that would not affect normal photos could ruin pixel-shifted photos. This would be especially true of Sony's implementation of pixel-shifting, because (unlike Olympus) Sony waits 1s between each shifted shot. If there is drift, this will generate false detail.

[Macrero]

Lou,

while I don't doubt that the guys at Dpreview know their stuff, there is the remote possibility that there has been something wrond with their tests. I'd wish I had the cameras to try them...

The Oly pixel-shift RAW was also a sample I downloaded, so it's not the last word either. The "resolution" was there, but as I said above, image came out smeared/blurry. It needs a lot of sharpening which inevitably generates artifacts and noise. And even after that, image still look worse compared to the Sony's native 42 MP.

Best,

- Macrero

[Lou Jost]

I also am trying to decide between cameras right now. The Oly has sold me on the value of pixel-shifting. I wish I could compare its hi-res files to images of the same FOV taken on the Sony A7Rii. we'd have to take care to compare them at the same fraction of picture width, not comparing 100% crops of each.

Pixel-shifting, when it works, will make FOVEON-like images, but with better quality and far higher resolution, since every pixel-location on the image gets full RGB information.

Edit: I am currently leaning towards the Pentax K1 (first version).

[Beatsy]

I use A7rii for macro and micro and I'm entirely content with it - even with the drop to 12-bit dynamic range in fully silent mode. I've tested that and can't find any significant differences between 12-bit and 14-bit output - or compressed vs uncompressed RAW (an option on all the high-end Alpha cameras).

I have an A9 too, but only use that on the microscope or for field macro, not for studio macro. 20-fps RAW is fantastic for quick stacking on the scope! It would be ideal for "handheld stacking" in the field too, but I've yet to master that technique - more practice needed.

Note: I use my cameras for a wide range of amateur/enthusiast genres (landscape, street, wildlife, arty stuff, etc) as well as a smattering of "pro" photography (i.e. paid for - weddings, events, portraits and products). These uses were the main factors in my camera choices. I'd suggest that your potential uses outside of macro/micro should be what influences your decision on getting the ii or the iii too because the differences *within* a macro/micro context are pretty minimal (except for battery life and cost - and pixel shift of course, though frankly, I don't value that). I already had two A7rii's and the A9 when the A7riii was released, so it was an easy choice for me to NOT get the A7riii - it didn't offer enough improvement over what I already have for my use-cases. I'm holding out for A9r instead, if Sony ever release that.

Not sure if this will be of any use, but here's the result of a test comparing 12-bit vs 14-bit and compressed vs uncompressed-RAW with A7rii. I may have posted this before - but couldn't find it with search. Images are 50% crops. I think I see a *very* slight improvement in micro-contrast with 14-bit uncompressed compared to 12-bit compressed (and I examined the original images at 100% and 200% magnification), but it's such a small difference it could just be natural variation between "takes" IMO.

[Macrero]

Lou,

I'd like to try a camera with pixel-shifting and see for myself, I was considering also the Oly Pen-F and will probably end up getting one in the near future, the price has already dropped quite a lot. But for now I think I'll go with the a7R II, I'm not paying 1.500+ EUR difference just for the Pixel-shifting of the a7R III. Coming from an a5100, (which is a nice little camera, but plays in another league), the a7R II will be a big enough change/improvement for me.

Beatsy,

yep, difference is negligible, 12 vs 14 bits RAW wouldn't make a difference in resolution/sharpness. 14 bits will be superior in high-dynamic range pictures, but if needed, you can use the EFCS and get 14 bit RAWs with the mark II, not a first world problem either.

I will use the camera for studio stacking only, so I don't quite care about FPS, IS, AF, etc... The lack of touch screen in the a7R II is pretty disappointing though, considering that the a5100 which is a "low-end" model has it, but I could live with that.

Best,

- Macrero

[Lou Jost]

Beatsy and Macrero, thanks very much for the advice. I do value the pixel shifting very highly, not only for resolution improvements in ordinary stacks but also because it reduces the need to buy a FF monochrome camera for my specialized laser-illuminated monochromatic photolithography lenses (by far the world's best lenses in terms of line pairs per picture width).

I cannot now afford a FF monochrome camera or an A7Riii (though I lust after the latter's pixel shift ability, though I also fear the dreaded line artifacts caused by the special focusing pixels on the sensor in certain lighting conditions). I spent all my money on lenses. I now have a basic set of lenses for several different formats: MFT (for field work and highest resolution per mm of sensor), FF, 6x7cm (Mamiya RZ series, dirt cheap!!), and 4x5 inches.

So for me the choices for a FF camera are A7Rii and Pentax K1 Mark 1. Both cost the same and both have fully electronic shutters (the Pentax has this only after a firmware update). The Sony has better adaptability to non-native lenses but the Pentax has pixel-shifting. Beatsy, it is very helpful to know that the difference between 12 bit and 14 bit images is barely noticeable. Thanks for that demonstration. I'm surprised you mentioned that pixel-shifting is not important to you. I bet you'd like it if you tried it.... Pixel shifting has a big effect on noise reduction and can give a significant increase in resolution.

[Beatsy]

Beatsy ... I'm surprised you mentioned that pixel-shifting is not important to you. I bet you'd like it if you tried it.... Pixel shifting has a big effect on noise reduction and can give a significant increase in resolution.

It's an optimisation too far for me Lou (with compromises attached). I don't have any issues with noise and the colour fidelity I get at pixel level is already plenty "good enough". So what if I *can* have a 42 mpix image that's perfect down to pixel level - what on earth am I going to do with it? I very rarely print my macro images - and the largest I did print were A2 (for a public display). I'm simply not going to use all that resolution and fidelity at the output stage, so it's a waste of time and effort to capture it at the input (for me).

I suppose the one good argument is gaining the ability to crop in severely to show more detail from parts of the main image - and I've done that with a few images posted here. But I can achieve the same by using a higher mag and higher NA lens to image the relevant areas directly anyway. And I can already crop in 100% and still get a good image at the lower mags.

I'm not decrying the technique, but this is why it's of little interest to me.

[Lou Jost]

Your photos are so good that I had always imagined you might hang them in galleries some day...

[JohnyM]

@Macrero
Im using A7RII routinely in my close-up / macro / micro photography.
Almost always with electronic shutter, i find that 12 bits plenty good enough.
Im using EFSC only if i need to use flashes.
In fact, we often print those images on 150cm x 100cm with 150ppi and everyone is very pleased with the effect.

My subjects are living plants - stacking them with wemacro (3 seconds / shot ) is already pushing it as is. I cant imagine adding additional 4 seconds for pixel shifting (subject movements and lots of retouching).

[Lou Jost]

Yes, that's a good point. I also often photograph living plants and run into problems with long stacks.

EDIT: I think the Pentax images are processed externally in post, so they might not take longer (as long as shutter speed is fast) than ordinary stacks. The PEN F takes at least 16s which is huge, and the capture part takes at least 8x the shutter speed. I think the Oly E1 ii is much faster but these figures are hard to find on the internet.

[Lou Jost]

Pentax cameras like the K1 can be converted to accept Nikon lenses by changing the mount. For the K1 there is also an Adaptist Multi-Mount which is out of stock at the moment but may be available again in the future, according to a letter I got from the Adaptist owner, Alston. This mount allows one to mount many different brands on the same mount! Nikon, Oly OM, Contax-Yashica, Konica, Four-Thirds, and of course Pentax K, with infinity focus "in most cases". Very clever.

https://adaptist.weebly.com/pentax-k-mu ... rsion.html

In most cases the Nikon mount requires adding a provided spacer ring to the lens' Nikon mount. For Nikon, infinity focus is possible but won't be at the true "infinity" setting of the lens; it will be closer. This will have the effect of ruining the correction done by floating elements, so for such lenses, a do-it-yourself job may be required:

http://forum.mflenses.com/viewtopic.php ... b947356d52

But for lenses that focus by pure extension, like the Printing Nikkor, the Adaptist mount will be perfect.

[Pau]

But for lenses that focus by pure extension, like the Printing Nikkor, the Adaptist mount will be perfect.

In fact it wouldn't be needed, why do not just use Pentax tubes or bellows?

[mjkzz]

oops, the A7III (not R version) does not have pixel shift feature, looked everywhere, so can't answer your question in another thread.

[Macrero]

oops, the A7III (not R version) does not have pixel shift feature, looked everywhere, so can't answer your question in another thread.

Oops indeed, I thought you had the R, sorry for my poor reading comprehensions