Sunday, 24 November 2013

clarity in digital (a retrospective with new data)

Well this is here because I became interested in the Sigma 30mm f2.8 lens. Before buying it I thought I'd do a little reading about it. So ... that's why this post started.

*Bigger is Better *More is Less *Less is More

Seems like an Orwellian double-think paradox.

For nearly 15 yrars I've been banging on about digitial cameras and film cameras. Being an IT specialist there is a great appeal to digital for me, but with 30 years of using film there was much experience and a high bar for digital to attain.

I have owned, tested and examined many photographic systems. The reason "why" was always to understand the tools and know what their limits were so I could make the best choices for my needs (which includes my budget).

The image to the left was taken using my 35mm outfit on Kodak slide film many years ago. My interests in digital lay in, convenience, speed of access, lowered costs. Since my first digital camera (in 2000) things have improved in iterations, going from .7Megapixels to my current camera being 12Megapixels.

When I bought my Panasonic G1 (12Megapixels) I owned Canon EOS digital gear (as well as film gear) and was getting tired of lugging the EOS digitals around (10D and 20D). I really appreciated many of the features of the G1 (excellent EVF and all that this brings to the table, tilt and swivel much bigger rear screen, much lighter). After some extensive testing I sold the EOS gear and have only kept the 35mm film cameras because they are just not worth anything to sell and as it happens have been useful photographing a wedding.

That was back in March 2009 and after spending time in the analysis I reached a few conclusions.
  • bigger format was better for wide and normal
  • bigger format was better for contrast and clarity
  • smaller format made better use of telephoto lenses
While looking for clarity and resolution in my images I used tools like Norman Korens lens testing ideas and charts. These enabled me to see effective resolutions on both my Film and Digital camera systems.

Systems is the key word here ... but I'll get to that in a moment.

Looking at the data it supported what I was seeing in my images and that was that my G1 was a better system for clarity and resolution than my EOS system was. When applied to my 6x9 roll film and 4x5 sheet film systems I could 'see' the numbers showing me that they not only resolved more (despite lenses having lower lp/mm) but also had better contrast and texture.

When it comes to capture "Bigger is Better"

Back when I first had my 20D I was disappointed with the images from it (while all the while people were going ape over it) because they just weren't clear enough with EF-S lenses and my EF lenses just didn't look and feel the same on it. My 24 was no longer a wide and my 50 no longer a normal.

Somehow these lenses simply looked better on full frame.

Owning the G1 seemed to be unilaterally better than the 20D (EVF for critical focus, shorter flange distance opened up wide choices of cheap creative alternative lenses, actual resolution was higher, slightly smaller sensor gave a more convenient x2 rather than x1.6 factor...) and somehow also made it clearer to me that Full Frame had something that neither APS nor 4/3 did (plenty of posts on that here).

Quantitative data

Fast forward to last night and I was looking up data from a lens testing site about a Sigma lens I'm interested in. I learned that since I was doing much of my testing (2009) a group has developed a very convenient lens testing metric - the Perceptual Megapixel (announced in 2012).

This is very useful metric, perhaps the most useful metric I've ever seen with relation to digital imaging. For it brings together the entire system (as did the Norman Koren method) yet gives a single number which can be applied in parts of the analysis of lens on almost any camera. Why is it useful? Well previously the only 'metric' on lenses was stuff like LP/MM measurements (quite subjective) and MTF curves (for full on experts), as the guys over at DXO say:
Can you define MTF? Do you know if an MTF20% of 50 lp/mm is better or worse than an MTF 50% of 30 lp/mm? And when reviewing an MTF chart, can you distinguish which curve is best? The short answer is probably no.

Since I bought my G1 (and swapped it later for a GH1 to get video and then added a GF1 last year) I have seen nothing compelling to encourage me to "upgrade" my micro4/3 cameras to any newer model. Seeing this data now gives me some metric to fall back on to support what I was seeing (but it seemed that few if any others were seeing NB:
cameras were getting more megapixels but the images just seemed to have the same levels of detail.
To wit lets look at a few lenses from the dxo site. As I was interested in the Sigma 30mm f2.8 I found that review first (and from there the P-Mpixel):

So the Sigma has 8 Perceptual Megapixels, the 'illustrious' (and slightly expensive @US$529) Panasonic Leica 25mm f1.4 gives you only 10Mp (from my 12MP camera) and the oft dribbled about on forums Oly 45mm (@US$400) gives you 9Mp. Hmmm ....

So, looking at a few more of the 'top micro4/3 lenses you simply have to have' we see:

9, 9 and 6P-Mpix. Seeing that rating on the 12mm f2 makes me even more comfortable that my bang for buck (based on what I saw in images) x0.79 adapter was good enough.

I get sick of the dribble I hear about how good the original Olympus 17mm f2.8 was (and how it was sooo much better than the zoom, and don't even start me on the 'stellar' 20mm) seing this low rating only confirms what I saw with my eyes. Well to my eyes it wasn't great and now there is data to back this up.

So looking at the performances its not ball tearing stuff is it. I mean none of these lenses will warrant a move up to 16Megapixels in your camera ... you'll just get more mush. In fact it shows that none of these "super zooms" even give you much more value than the original kit 14-45 zoom that came with my G1. It puts some metrics behind the arguments made by people that the Olympus micro4/3 anything zoom on the OM-D anything will compete with Full Frame ... it just can't and this new understandable data metric seems to support what you see when you look carefully at the DPreview comparison images.

So Less is More (more money back in my pocket).

In comparison to Full Frame some quite humble lenses (and my EF24 f2.8 is always bagged by people from the Kingdom of Wang) as well as one EF-S zoom

So no matter how many MP your camera has (and the 7D has an 18Megapixel APS sensor) if you use that lens you'll get effectively no better than 7Mpix (or about what I get on my micro 4/3) while if I used the older original 5D I'd get the full 13Mpix that the sensor had with either my EF24 f2.8 or my EF50 f1.8.

Worth noting is that the EF24 f2.8 when tested on an APS camera gave no more than 7P-Mpix (due to higher magnification demands of the smaller sensor)

This has helped explain to me what I was seeing when 35mm images taken with my EF50mm f1.8 were sharper and clearer than what I ever got with any of my smaller sensor digital cameras.So as well as better shallow normal I also actually get higher definition too.

The older EF24 f2.8 is not available new anymore, but as a used lens in excellent condition you'll find them for less than US$300, which is much better bang for buck than the Olympus 12mm f2.0 at US$800 and will give you higher resolution to boot!!

These numbers make it clear that if I actually want higher resolution and better contrast I'll need full frame (cos APS and micro4/3 just don't have the numbers). This re-enforces that with the lenses being as they are there is no point in having a compact camera more than of 12Megapixels.

So again Bigger is Better, and Less is More.

Does this mean I am disappointed with my micro4/3 move?
No, because it still gives me a light weight camera to mount many and varied lenses on.

My initial choices for moving from my 20D to the G1 was to have a high quality lightweight camera system that was flexible. My GH1 and GF1 are still exactly that. I now know that do get better performance from my micro4/3 system than I did from my APS system as well as better options for low price and creative lens options.

To me pictures have never been about Megapixels and you'll find in my writings for many years that I've often felt that a good and sharp 6Mpix was enough for most needs. So now with this DXO metric I'll say that a 6P-Mpix is good enough :-)

I'll leave you with an image which is an old favorite of mine, it was taken with the EF24mm f2.8 on film. Its certainly got way less than 8P-Mpix in it, but due to the nature of the way things works prints nicely up to 70cm. Please click on it for a larger view

but I like it not because its sharp but because it has feeling. Its an nice photo of my friend and excited dogs down on the beach on a nice day. Knowing how digital blows out in the highlights, I know that I'd have never been able to get this image with a digital, so I'm glad I had only this antiquated equipment.


So I'm really glad that there is now a metric to help me to make appraisials faster. I know that my micro4/3 has limits and I'm willing to live with them. Understanding the limits makes it clearer to me why I won't benefit in the upgrade. In fact something I have not touched on here is I will benefit by not upgrading because more megapixels in the files just means more storage. More storage costs, longer backup times, bigger SD cards .. and all for nothing that can really be seen in the images.

So I'm going to buy the Sigma and I reckon that it'll be a nice addition to my GF1's compactness (cos that 14-45 zoom is a bit bulky). This will give me a nice lightweight travel and snapshot camera
GF1 (380g) + 14mm (55g) + GWC1 x0.79 wide adapter (70g) + Sigma 30mm (135g).
It'll weigh 250g less than just a 7D or a 5D body and give me effectively a 21mm f2.5, a 28mm f2.5 and a 60mm f2.8 Sure it wont have the Megapixel count but who can see that? The camera only has one eye, but those who only read data and chase the megapixel madness are just as blind as King Wang

A wonderful quote on the DXO site:
The Perceptual MPix measure confirms certain rules of thumb such as a 12 MPix full-format camera is sharper than an 18 MPix APS
Just say More Megapixels and the followers of King Wang chant:
Long live the King!
King Wang King Wang King Wang King Wang

but to me Megapixel madness is
King Wang King Wang King Wang King Wang King Wang King


Yu-Lin Chan said...

I can't claim I understand the way DxO test the lenses, but the Sigma 30mm f2.8 can resolve much more than 8MP. On my 16MP NEX-6, the resolving power of the lens is unbelievable given the price point. It puts most expensive lenses to shame.

Charles Maclauchlan said...

not being an engineer I never spend much time reviewing these types of charts. It is interesting though. If I follow what i think to be their point, almost all modern cameras are limited by their lenses. Is that correct?
I'm confused though by the different ratings for the same lens in different mounts. Your 30mm in 4/3 vs APS for instance. It could be that the lens construction is different however i kinda doubt it. I also see dramatically different results for a Canon mount vs a Nikon mount

obakesan said...

Hi Charles, yes you got it. Its often that the lens is the limiting factor. So if you take that same lens (say one made for full frame) and then crop out the bits that APS or 4/3 sees and the scale it up to those pixel densities, instead of getting more clarity you just get more mush. The EOS EF300mm is one such example. Looks great full frame but starts to soften noticeably when using only the portion projected on 4/3 sensor sizes.

I've tried this on both film and a friends 5D with similar results.

However as its effectively a 600mm on my 4/3 camera it probably looks better than a 300 + a x2 teleconverter.