Thursday, 11 October 2018

experiments with Sony A7 Full Frame

thought I'd take advantage of shitty light to do some testing, because I've long wanted to compare my GH1 in RAW to the (now old) Sony A7 full frame.

To do this I chose:

  • Panasonic GH1 + Panasonic 25mm f1.8 
  • Sony A7 + Pentax thread mount SMC50mm f1.4

Both shot wide (which is a small EV advantage to the Sony, but a small softness disadvantage to). I chose to also keep shutter speed to the same in both. My "base" speed was 160th determined by the lighting and the selection of 1600ISO on the GH1 (as I find 3200 verging on unusable)

Ok overview

even at this (lack of) magnification that noise and banding effects from the sensor are present in the RAW (although not in the JPG), as an aside I'll say I've not shot this lens at this ISO on the camera before, so I'll need to determine if this (noise) is exacerbated by the lens ... as has been observed in other camera + lens combinations at high ISO

Straight out of dcraw (of course I shot raw) with zero processing (and allowing dcraw to determine everything (which I may say it does very well) including colour balance (to remove camera bias from the equation) )  we see this:

... and my well experienced noise levels in the shadows (restricting the low light high ISO worth of the m43 cameras) are obvious. Clearly looking at the side grain of the wood we see greater ability to resolve features in the wood ... as well as very favorable colour.

Reducing the image of the A7 to the same pixel dimensions (simple rescale) to make it a cleaner comparison:

... reveals that the additional A7 captured detail remains and the noise levels diminish in significance (and well shadow noise is still crummy on the GH1)

I then employed Nik Define (going back to pre-resize) to have a quick go at handling the noise. I chose the setting to only adjust contrast noise by a small margin (24) to not erode details but cranked colour noise right up to 160 (out of 200). I did this to both directly and then (again) resized the A7 image back to 4016 pixels (to equalize image size for ready comparison)

This further equalises them ... but still there is an advantage to the A7 and well that sensor noise just isn't going away is it.

Lastly I've reisized them again further to what I consider is normally visible in a print when viewing them on a decent screen (full screened) or indeed on web forums.

where one just can't get rid of that sensor noise (without also obliterating image details), which brings me to the OOC JPG's of the two, because Panasonic does a more effective job of obliterating the noise in the OOC JPG (but at the loss of details of course ... so here are both JPG's sized the same

... of course the GH1 (and indeed earlier Panasonic cameras) were often criticized for their lack lustre OOC JPG and this is a good example of why (and why I prefer RAW working with them).

I then moved into less dim (but still challenging) light and took some images with Depth of Field and aperture in mind. I was interested in effects on

  • contrast (as this also changed with stopping down)
  • depth of field (at aperture)
  • shutter speed (constant aperture, differing ISO to get the same speed)

So comparing the P25 at 1.8 vs 2.8 lets start with both on 1600 ISO.

we get a clear increase in constast (that's a good thing) and of course a reduction in shutter speed (because the aperture closed down) from 400th to 125th ... no surprises there.

And comparing the 25f1.8 to the 50f1.4 we see that the Sony yields less noise (as observed above)

as well as both shallower DoF (expectable) and a bit less contrast. As you can see in the data below the pictures I've resized the Sony to be the same dimensions as the Panasonic.

So what if we up the ISO on the Sony to compensate for the dropping of the aperture to f2.8, which should give us the same DoF as the Panasonic at 1.7

which it does. So shutter speed is now 400th between them while the A7 (on the right) is at 4000 ISO ant the Panasonic on the left is still the 1600 ISO image. Now we see that DoF is relatively equalised (as expected from calculation of aperture diameter) but both contrast and noise are still better even though we've gone up that stop (or more).

It gets worse if we then don't resize down the A7 which renders WAY more detail (even from this 1970's lens)

...and is thus more amenable to some post processing without image degredation. You can see its even clearer at the bottle caps.

where print is actually almost readable on the A7 captured bottle

As it happens even stopping the Panasonic down to f2.8 (which will drop the shutter speed or increase the ISO to much worse noise levels and increase the DoF) simply doesn't help

So the Panasonic now matches the contrast (but not the detail) of the Old Pentax lens on the Sony , but at the expense of reduced shutter speed, increased DoF (meaning less background separation) and if you happened to be on the edge of shutter speed (you know, to prevent motion blur with moving subjects like people) you'd need to up the ISO and get more sensor noise.

For instance if I'd had the A7 with me on this evening I could have done a much more clear capture of this portrait (late in the evening, with dimmed lights.

I hope this helps others answer some of the questions about why some people in some situations prefer Mirrorless FF (on the Sony A7) vs m43. I know its swings and round abouts but I've been a proponent of m43 for so long I thought I should provide some balance as to why the new Mirrorless FF cameras are actually pretty attractive in some situations.

Obviously this does not address lens availibility, focal lengths and weights...

No comments: