Friday, 28 March 2014

shallow portrait lenses 43rds vs Full Frame (part 2)

Way back in 2011 I wrote a post about comparing portrait length lenses on micro 4/3 vs full frame. I had always meant to do a part two of that article, but things got in the way of that (as my friends will attest).

So with the excuses out of the way I have felt moved to make the second part of that comparison because the topic still seems quite interesting to people. I see on a board that I follow that people are interested in comparisons between the Nocticron on a micro4/3 and a cheaper Full Frame system, which is more or less quite related to what I did back in 2011. To sum up the points I made back there I felt that the evidence I found supported:
  • there is a essentially two stops difference between 4/3 and Full frame to get the same Depth of Field.
  • that there were significant contrast differences between a 50mm @ f1.4 on a 4/3 and a 100mm @ f2.8 on a full frame
  • that the Canon 5D Mk1 had a better detail than the GH1
  • that there was better highlight and shadow details in the 5D than the GH1

To recap:

The image from the Olympus 100mm at 2.8

and then the 4/3 camera with the 50mm at 1.4

Stopping the 50mm f1.4 down to f2.8 we find it to be more or less equal to the contrast of the Olympus 100mm lens at f2.8.

but of course with much increased DoF (as expected).

Which brings me to observe that so far it is my experience that simmilar F-stop numbers produce similar contrast. Personally I find the 50mm @ f1.4 to be too milky and too soft. The close examinations in that article show that flare and reduction of contrast are the culprits.

However back in that article I wrote that it would be interesting to compare the new Oly 45 f1.8 ... let me quote:

Perhaps the new Olympus 45 f1.8 lens would solve these problems and yield a lens which would
  • have the contrast I find in the Olympus 100 f2.8 on full frame,
  • have at least similar (if not better) sharpness
  • be more compact and light weight than the FD f1.4
The lens will set you back about $400 as it seems to be panning out, which is quite attractive. The initial discussion of the new Oly 45f1.8 indicated prices would be much higher, but if it does come in at $400 that will put it right on par with the typical 100mm f2 lens.

Since the Oly is a 45mm not a 50mm, you will need to get closer to the subject as there is a significant difference between 45mm and 50mm. It may work out near enough, but I suspect that you'll get a tighter DoF out of a 100mm lens on full fame than even the Oly45mm f1.8 simply because I got exactly that when comparing the 50 and the 100.

With both images I tried to focus on the back edge of the cap. Using the magnify zoom focus assist on the GH1 it was easy to get that spot. Looking carefully its clear that the 50mm lens on the 4/3 the focus zone is much deeper even at f1.4. By the distance from the ear to the eyelashes, they are starting to go out of focus on the 100mm at f2.8 but not on the 50mm at f1.4. This means also that you won't get that on the Olympus 45 f1.8 either.

The New Material

I have a couple of things I wanted to show, first is the differences in more detail between the two images in two ways:
  1. same diameter aperture
  2. same f-stop aperture
Recalling that F-Stop is a ratio of lens focal length over aperture diameter (so a 50mm lens with a 25mm diameter aperture would be a f2 lens).  Again I have an old post explaining that on my older server here. As Depth of Field is essentially related to diameter not strictly aperture having the same diameter aperture between 4/3 and Full Frame works out to be f1.4 on the 50mm on 4/3 and f2.8 on the 100mm on the Full Frame.

So at the same diameter

and next at the same aperture:

I notice two things seem related at different settings: 
Contrast at fstop and Depth of Field at aperture diameter.

I see that at the same F number the contrast levels are almost the same and yet the background Depth of Field is deeper (IE more focused). I'm not sure if this transfers to other lenses or if its just here.

Ok, the next thing I noticed in panning around the images was how much better (cleaner) the 5D image was to the GH1.


Back in that earlier article I observed that the 5D held shadows and highlights better than the GH1 did. I am sure that this is going to be the case across all the micro4/3 but would love to actually see results (not just fanboyism). In that comparison I used ACR to do my conversions, and still noticed some differences. The other day I used DCRAW to do the conversion because I wanted to see what was actually in the RAW files, not what cunning reconstructions that Adobe can do (which they do so well).

As you (should) know, both these cameras use a typical Bayer Array, which contains 2 times as many Green pixels as it does Red or Blue. There is no actual RGB pixel, this is constructed during 'demosiacing' from the R G and B pixels found in the Bayer array (as in figure to the left). There are of course various algorithms for this each with various advantages.

This has the basic effect however of making the Green channel much stronger (less noise) than any other channels. If you have not done this before I suggest you load any image into Photoshop and look at each of the colour channels. Of course if you have a foveon sensor camera, where you will not see such differences in RGB ...

This noise does make a difference on a print, especially if you print large. Some years ago (before blogs existed) I wrote a post on my observations of this here as well as some other observations on channel noise differences between JPG and RAW here. While the focus of those articles is not the same as here, there is some overlapping information of relevance.

So, starting again with the same diameter (to keep same DoF) and looking at the Red Channel from the images (converted as 16bit TIF) we see this:

And the levels of detail in the darkness of the red are dreadful on the GH1 while allso the hightlights are blown to saturation too.

Now, lets look at the images obtained with the same F-Number (and of course the shutter speed was slowed by 2 stops as the aperture was brought down too):

We now have a much happier comparison between the two images, although still the GH1 has a bit more noise than the 5D, its a lot better than it was at f1.4, which is curious.

To compare the other channels I have put together this pair of animated GIF files cycling through the images:
Colour - R - G - B



So its pretty clear that the GH1 files do not have the same capacity to be free from noise or saturation as the 5D does. This of course will be noticeable as a grubby noise in the shadows when comparing the two, or a softness (as all current noise reduction algorithms I've seen produce).

I would love to make this comparison with a newer lens like the Olympus 45mm f1.8 to see if it does a better job of the contrast.

the JPG versions

I realise that some people are JPG only shooters, so I thought I should include the JPG's that came from the cameras. These were both with settings as "standard" as possible.

5D @f2.8

its red channel

which isn't too bad really.

Then to the GH1 @f1.4

its red channel

same mushy shadows as I obtained with RAW processing, perhaps worse.
So in my view, even the marvels of in-camera processing just make it clear that working in RAW has been a benefit to the Panasonic.


Does this mean I don't like my GH1? Not at all, its still exactly the camera I wanted it to be. Its light, it does great video and takes a host of lenses. I love it. All this means to me is that I'm more aware of the actual differences that will result in use between the cameras.

Does this mean I'll be ditching my 5D? Well no, while I hate it for being a fat slug it produces good images. Fat slug you say? Ok, lets look at a few images.

This is my 5D:

and from exactly the same location (the camera that took the pictures was on a tripod, it didn't move and it was using a manual focus lens which I also didn't move) my
EOS 630 film camera:

Did you notice how much more blob there is in front of the camera and how much more bulk there is behind the film plane (yellow line)? Yet the 630 gives me shit-hot fast AF, fast 5 frames per second motor drive (twice the speed of the 5D) and lasts about a year of shooting on a single battery (well ok, you need to feed it film too). Its lighter than the 5D and almost as convenient to carry as my GH1 is really.

I happen to like digital cameras but dislike DSLR cameras for their bulk. I hate how fat DSLR cameras are in comparison to 35mm film cameras.  Of course the new Sony a7 seems to have solved that problem.

Below is the crew of cameras used in making this post (obviously not taken with any of these cameras):

When going on holiday just walking around I'm far more likely to have my GH1 with me than a 5D. (In fact I'll probably take the 630 along because I prefer the way negative reacts when shooting in harsh light). However when I'm shooting an important event like a wedding I'll be sure to be taking along a Full Frame camera like the 5D.

Also the GH1 (by being a smaller 4/3 format) allows me to make better use of my telephoto lenses, or even own telephoto lenses which I could not make use of before (such as my FD200mm f4 which only weighs 440g and is really cheap as it won't fit on any DLSR).

In the end to answer the question I posed in the opening of the first post (you remember that one, back here), while its close, the cameras are not fully "fungible" ... each has distinct differences, each brings something different to the table.

To quote an old saying: if all you have is a hammer, everything looks like a nail. So by having both full frame and 35mm (well and large format too) and understanding the differences, I can pick which tool is right for the job.



Unknown said...

I think it all comes down to what kind of shooting styles you prefer. For me personally, who shoots mainly on the move, a smaller camera makes more sense, but if I need to shoot stuff that moves, I would grab my 1D Mark III.

Shallow depth of field is overrated :)

By the way, the Canon EF 100mm f2 is absolutely underrated. It should have a red ring on the lens barrel. Perfect portrait lens on full frame.

obakesan said...

Hi YuLin

I agree all that. I sometimes think that the general discussion of desire of shallow portrait ignores that you have to work for it to get it. I remember in working conferences (shooting on the fly) that I'd pick f4 or f5.6 on my 100mm lens so that any focus error would not leave me with the subject just (tantalizingly) out of focus.

I sold my EF100 years ago to fund a TS-E 90 (which I've now also sold) as it was better for what I wanted to do with it.

When I'm on the move my micro4/3 is my first choice and my full frame stuff gets pulled out perhaps once a year.