Canon Ixus 860, 8MP, tiny sensor
Panasonic G1, 12MP, "4/3" sensor
Canon 500D, 15MP, "APS-C" sensor
Canon 5D-II, 21MP, full frame sensor
All of them had their respective "standard" zoom, set to 28mm equivalent.
All of them set to around F:5.6 or F:8, and 800 ISO, which I find to be representative. Except the IXUS, which I let use its auto-ISO setting which selected 400, wisely since such small sensors suck above that.
Full scene:
Crop from Ixus:
Crop from G1:
Crop from 500D (T1i):
Crop from 5D mark II:
(Big pic when click.)
Update: it's coincicental that the megapixel sizes in my cameras follow the sensor sizes neatly. You might well have a tiny sensor with 14MP and a full frame sensor with 12MP.
At the top is a neat diagram I found which, except for being slightly bigger than real life, perfectly illustrates the four sensor sizes above (two of them even mentioning the same cameras!). (The EP-1 is "4/3" like the G1.) (Don't try to find meaning in monikers like "4/3" and "1/8", they are complete gibberish with no relation to anything real.)
Update: the Canon 5D mark 1 was the king of high-ISO performance for two years. It was impressive that you could get usable pictures at 1600 ISO, albeit rather noisy. No other camera came close for a long time.
But the 5D Mark II kicks it all across the field. At 1600 ISO I don't see any noise at all! In fact I basically can't see any quality difference going from 800 to 1600. And they did that at the same time as almost doubling the resolution! That is f***ing impressive.
Update: even more so 6400 ISO, which I had kind of dismissed, but shouldn't have. At default setting, it's good. But it gets even better if you turn off high-ISO noise suppression (custom settings) and handle noise in the computer. (And this is from JPG, I haven't tried RAW yet.)
I thought only Nikon D3 could do this, and the D3 has only nigh half the pixel count.
Frankly, this goes beyond my expectations. When they boosted the resolution that much, I was sure they'd blown the chance for really good high-ISO performance.
Note: for a long while, I thought that noise was the only penalty for using the high ISO settings. But there are other issues too. Most importantly, you loose sharpness/resolution/detail, and that can't be recovered. The dynamic range also suffers some, and colors may lose some clarity, though with the new generations it seems to be minimal.
Apart from the color cast the G1 and 500D seem to be comparable. What's the difference between 3/4 and APS? I see it's 2MP, but that isn't much.
ReplyDeleteI also got motion sick going between images as they rotated.
APS-C is about 2/3 of 35mm full frame. "4/3" size is about half. This is *linear* measurements, not area.
ReplyDeleteIn practice such a difference is rarely noticed, it's true.
I guess it matters what you print; as I rarely do billboards, the smallest would be fine, but for the lack of manual controls, and external flash. The G1 is most appealing, sans recession.
ReplyDeleteGood idea, Eolake, to show the same objects with different cams who have different sensorsizes!
ReplyDeleteBut why do you picture through a closed window? So you give away sharpness you can show?
By the other hand you give us two pictures in one: It seems you show a statuette like Victor Issa's 'Inviting-Joy'-Statuette inside ... I like it ;-)
Have a nice day!
Walter
Yes, that's the statuette.
ReplyDeleteJust for fun, I have tested it, and apart from reflections, I could not seen any quality difference when shooting through a window.
Thanks for your test-happening, Eolake!
ReplyDeleteSo what do I/you learn from this? A glassfilter named 'window' close to the lens and nearly parallel to the optical axis is not so harmful for the center of pictures! If the window-distance is a little greater, the things come with there own little show!
What do you learn? You need not to clean your window ... this week or so ;-)
Nice day go on !
Walter
Me, I learned that while pocket cameras makes amazing quality pictures up magazine-double-spread size, in really big sizes, bigger sensors does make their mark.
ReplyDeleteAnd that the bulk and price of the 5D2 and "L" (high-end) zoom lens (24-105mm) really does justify itself in resolution when you need that.
Note the 5D2 image, clean as a whistle.
ReplyDeleteYes, I must admit I was positively surprised by that. Impressive.
ReplyDelete... And I just updated the post... 1600 ISO is clean as a whistle too. Noise? What noise?
No noise! That's terrible, what's the point of 1600 emulation if it's not all grainy!
ReplyDeleteWell, you can always use 6400.
ReplyDeleteOr I think it can even be pushed to 25,000.
That is just crazy!
ReplyDeleteYep. Nutty as squirrel poop.
ReplyDeleteAnd that the bulk and price of the 5D2 and "L" (high-end) zoom lens (24-105mm) really does justify itself in resolution when you need that.
ReplyDeleteSo to chuck a cat amongst the pigeons ... I've a bagful of 'L' series lenses and have been pondering on whether to buy the 50D or really upset the bank account and splash out on 5D? Seriously wondering whether or not I'm likely to see any real difference in the end result, assuming that I do my bit properly?
With L lenses, I think you'll be very pleased with the 5D2.
ReplyDelete(I have about 4 L lenses, which was one of the reasons I finally went for the 5D2.)
ReplyDeleteUpdated again, with 6400 test. Well, paint me green and call me lazy Hulk.
ReplyDeleteWith L lenses, I think you'll be very pleased with the 5D2.
ReplyDeleteTend to agree with you ... but a new baby boy bouncing around the place makes it very easy to feel guilty / over-extravagant where things like this are concerned :-)
The really crazy part of all this is when I try to put it into perspective ... I 'retired' (early, I hasten to add!) a few years ago after working self-employed as a photographer since the mid-80's ... during which time, out of all the photographs I produced, substantially less than 1% were ever needed / enlarged above a 10" x 8" print ... and the vast majority were from 35mm film stock, not roll-film!
There is that. You need a lot larger prints to see the difference.
ReplyDeleteConsider the 500D (T1i) instead of the 50D then. I have it too, great cam, same image quality, only more compact.
ReplyDeleteIf I'd known that the 500D was in the works, complete with video, I would have waited instead of buying the new 50D in October '08, but that's the technology gamble. The next move is definitely toward full frame, 5D2. The 50D has enabled some low light shots that would have been impossible with the old 450D, the last of my consumer DSLRs. The extra pixels are a big plus if I want to crop out a section of a photo to use.
ReplyDeleteThat looks like a nice, cozy little apartment. In London it would probably cost about five hundred thousand pounds.
ReplyDeleteI was so lucky with that. I bought it in 2002, and at that time, for some reason this small area was behind the price explosion. I bought it cash for £34.000! Now it's over 70k.
ReplyDeleteIt has everything I need, and it's a five minute walk to town, and the neighborhood is not bad.
Note: I thought things photographed with wide-angle tended to look bigger, but it seems to have the opposite effect here. The room is bigger than it looks on the picture. It's not a coffee table, it's a big, solid dining table seating six people.
ReplyDeleteWell, it couldn't be called a coffee table because it's too high, you've got chairs there. It's way too big to be a coffee table! ;-)
ReplyDeleteYes, but there's a sofa too, and I *do* drink my coffee there. As well as use it for a quick photo studio surface. So I guess it's a multi-purpose table.
ReplyDeleteWhich I really love by the way. The clean, simple, heavy, strong form. The legs are flush with the surface both vertically and horizontally. There's not one line in it for the purpose of making it beautiful, but because it's quality and simple, it *is*. Or at least to my taste.
I often keep it bare because it is so nice on it's own.
(I would need a bigger space to make a photo of it which shows it well, though.)
You're definitely not using it as a coffee table then, I mean aren't you supposed to have a bunch of pretentious art coffeetable books and magazines like The Times Literary Supplement, the New Yorker, etc., sitting out to impress your many visitors? ;-)
ReplyDeleteMy usual kind of visitor is more impressed that I'm not trying to impress.
ReplyDeleteBut for others, I have a variety of spreads to select from, from Perfect 10 to Non-Illustrated Science (with the articles in the original languages).
Four Thirds is the ratio width to height of the sensor and lenses behave like twice the focal length because of the small size has half the diagonal.
ReplyDeleteUnder the same logic 35mm would be called three halfs (doubt that will happen).
I thought I remembered some reference to obsolete video tubes widths. Found this on wikipedia here:
ReplyDelete"The name of the system comes from the aspect ratio of the image sensor used in the cameras, which is commonly referred to as a 4/3" type or 4/3 type sensor. The common inch-based sizing system is derived from vacuum image-sensing video camera tubes, which are now obsolete. The imaging area of a Four Thirds sensor is equal to that of a video camera tube of 4/3" diameter."
... So I'm guessing the video tube naming tradition is the origin, and the aspect ratio fit is a nice bonus.
Notice how big the tube had to be (34mm) to fit a 22mm diagonal imaging area.
I can't seem to find a magazine with the name "Non Illustrated Science."
ReplyDeleteI was joking.
ReplyDeleteTry SRAW1 when at 6400 ISO, it's even more impressive. This is one of my 5D² tips that you can find here.
ReplyDeleteThanks for the tip, I'll take a look.
ReplyDeleteWhat is "SRAW1"?
sRAW1 is the reduced RAW (10 MPix), more info following my link...
ReplyDeleteXavier.
Actually, 4/3rds does not specify an aspect ratio, just the sensor diagonal. This is why the GH1 is actually in-spec for Micro-4/3rds despite using a multi-aspect sensor which delivers the same ~21mm diagonal in 3:2, 4:3 and 16:9 aspect ratios.
ReplyDelete