Megapixels - Are More Better?
Monday, July 14, 2008

This content has been moved to the new SeriousCompacts.com.

Click here to go to the new location

Posted by Amin Labels: ,

Comments (23)

Loading... Logging you in...
Dashboard | Edit profile | Logout
  • Logged in as
+1 Vote up Vote down
Hiding Pup's avatar

Hiding Pup · 873 weeks ago

There's one thing I like about higher pixel count sensors: you can run software like PTLens/LensFIxCI through the images, and the corrections they make to barrel distortion etc. are utterly invisible. This doesn't happen on a 5mp image where you'll be able to see where the corrections have been made. For me, the ability to crop a larger megapixel image without too much loss in resolution, together with the ability to fake a distortion-free lens are compelling arguments to get a higher megapixel camera than a lower one, particularly if the higher one has a bit of optical image stabiization thrown in as well, thereby reducing the need for high ISOs...
+1 Vote up Vote down
David Jenrette's avatar - Go to profile

David Jenrette · 873 weeks ago

I have a D2x and a D3., each with about 12 megapixels. If I make a full frame 11x14 print with both of them, the D3 gives a magnification of 11 times while the D2x nees to be magnified 16 times. Just as a 4x6 enlargement of almost any camera looks good, the D3 image looks that much better to me because it is magnified less.
+1 Vote up Vote down
Andrew's avatar

Andrew · 873 weeks ago

Something about John's comparison doesn't sit well with me. In reality, you would never use the same focal length lens on both of those cameras to achieve the same picture. I understand what the comparison tries to show but the method would never be used in practice and therefore becomes irrelevant. If I'm going to use a 22mm focal length on an FZ50 shot, them I'm going to use a 70mm lens on a comparable 400D shot (my math may be wrong here since I have no idea what the sensor size of the FZ50 is). Why would I limit myself to such a small area of the sensor?
+1 Vote up Vote down
Stephen Gillette's avatar

Stephen Gillette · 873 weeks ago

This post adds interesting nuance to a topic discussed ad nauseum. More pixels can be a good thing. Less can also be good.

Some of us are compelled to translate the end product of a given camera into numbers, in order to compare the "performance" with that of other cameras. If the numbers are "better"--higher or lower in a given metric--these folks are happier. Their cameras reflect back on them as wise consumers, and status is enhanced. Part of me sits in this camp.

Others want the big picture (metaphorically). For them, the camera is a tool. Numbers-based performance means little. "What kinds of pictures can I take, and what will they look like?" is the larger concern, and the resulting pictures are what matters. A larger part of me sits in this camp.

A camera can be used as a tool primarily to test itself, over and over. Or it can be used as a tool to create images in order to learn about ourselves and to communicate our discoveries with others. (Call that "art" if you wish.)

This post has confirmed how smart I am to pack a Canon G9 when I shoot. And how smart I am to pack a Fuji F20 when I shoot. Turns out, for a non-numbers kind of guy, I am smarter than I thought. Thanks for that, Amin!
6 replies · active 873 weeks ago
Loading...
+1 Vote up Vote down
amin's avatar - Go to profile

amin 67p · 873 weeks ago

Stephen, I think your comment gives exactly the right context for this topic. Although you are neither in the "numbers-based" camp nor the testing camp, you're one of the few folks around who can answer this question based on practical experience.... so I have to ask: If you needed to take an ISO 1600 image at 35mm (equivalent), f/2.8, and 1/20s (advantage of image stabilization excluded), which in your experience would give you an overall better 8x10" picture, the G9 or the F20?
+1 Vote up Vote down
Stephen Gillette's avatar

Stephen Gillette · 873 weeks ago

Rats. I'm being cornered...

In real life, I would shoot the G9 at ISO 400, rely upon image stabilization, and be happy. The G9 is really not the camera for ISO 1600 shots needing 8x10" prints. But while the F20 would do okay at ISO 800 if a 8x10" was the objective, it would not do great. And ISO 1600 would result in some heavy loss of detail. (Which might be fine, depending upon the subject. We are talking low light, after all.)

I'm thinking results would be close to a toss-up. With IS, I'd go for the G9 and ISO 400...
Loading...
+1 Vote up Vote down
amin's avatar - Go to profile

amin 67p · 873 weeks ago

I did corner you there, didn't I? =)

Without IS, close to a toss up is an interesting answer. That sort of practical estimation means more to me than the sort of experimental data John Sheehy posted, though I like looking at the latter as well.
+1 Vote up Vote down
Andrew's avatar

Andrew · 873 weeks ago

After reading the dpreview thread and thinking about it more, I think the takeaway is that yes, more megapixels crammed into the same size sensor doesn't necessarily degrade picture quality if final output size remains constant. But, shouldn't one advantage to having more megapixels be the ability to print larger? I bought the F20 with the belief i should be able to comfortably print an 8x10. If i got an F100 today, I would think that my comfortable print size should increase as well. I think it's obvious that more megapixels will benefit a given print size, but i should also be able to print BIGGER if needed. Now is that the case in today's megapixel race? Will an F100 give me beautiful 13x19 prints?
+2 Vote up Vote down
Stephen Gillette's avatar

Stephen Gillette · 873 weeks ago

Okay. I did a quick test...the Canon G9 vs. Fuji F20, ISO 1600, indoors with dim enough natural light to allow for 1/20 sec. @ f/2.8, both at 35mm (equivalent). I then printed at 8 x 10", on my wife's Epson 1280. (The G9 image not only has higher resolution, but was processed from a raw file. This test does not strictly conform to the scientific method!)

First, before results, some waffling. How close is "close to a toss-up"? Hmmm...

The F20 newly impressed me with its noise control. At normal viewing distance (don't ask, I'm not orthodox here), there is little visible noise. I printed on matte paper, which is a bit more forgiving than glossy. The G9, as expected, displays better detail. The F20 gets more total points at this size, however, for the noise handling, than the G9 does for the enhanced detail.

So the F20 wins. The subtle lack of detail is forgotten when you see the great job the Fuji processing engine has done in regards to noise. But is this really a surprise? The Fuji 6MP sensors have built their reps on low noise. Close to a toss-up? For many casual shooters, yes. For the visitors to this site, I suspect most would pick the Fuji 8 x 10" hands down.

(For black-and-white, however, the race might be closer. And low light is the realm of the rods, not the cones, in the anatomy of vision.)

So I fall back on my original disclaimer. Namely, with IS enabled, I could have shot the same image at ISO 800, and probably 400. This would have given the advantage to the Canon on all fronts, I believe. But, playing by the rules (Amin's rules), I was reminded that I should dust off the Fuji and pack in on my belt a little more often when I venture into the nocturna...
Loading...
+1 Vote up Vote down
amin's avatar - Go to profile

amin 67p · 873 weeks ago

Appreciate the test Stephen!! By any chance, could you send me the G9 RAW file and the Fuji original JPEG? I'd be interested in trying to process the G9 file with Noise Ninja and doing my own 8x10 prints to see whether I can close the gap.
Loading...
+2 Vote up Vote down
1001noisycamera's avatar - Go to profile

1001noisycamera 98p · 873 weeks ago

It looks like the original dpreview thread has been yanked. I followed Sheehy's posts and tests on this topic in the past. While the ideas and the tests themselves are great, educational and entertaining, the argument they are trying to prove/disprove does not compute.

The reason? You just can't use a piece of the sensor in the way he is comparing it.

In other words, if the Canon 40D sensor was to be the size of the FZ50 sensor, then it wouldn't be the sensor it is now. It would have been designed and engineered to fit its size.
2 replies · active 873 weeks ago
Loading...
+1 Vote up Vote down
amin's avatar - Go to profile

amin 67p · 873 weeks ago

I tend to agree with you 1001. On the other hand, imagine a camera that slapped together four G9 sensors to form a single 48MP 1-inch sensor.
+1 Vote up Vote down
Stephen Gillette's avatar

Stephen Gillette · 873 weeks ago

For those with a bit of whimsy and a sense of adventure, you can shoot four--or more--overlapping shots of just about anything and get the 4 x G9 effect on the cheap, right now. Stitching software is abundant, or layer the shot together manually, if you enjoy time spent in Photoshop. Really huge files are easy enough to generate. Less easy to work on and archive. (Got extra hard drives?)
Loading...
+1 Vote up Vote down
amin's avatar - Go to profile

amin 67p · 873 weeks ago

It seems as though they changed the URLs and broke incoming links, but the thread is still there.

http://forums.dpreview.com/forums/read.asp?forum=...

I'll go through and fix the links when I have a minute.
+1 Vote up Vote down
sonomichele's avatar

sonomichele · 873 weeks ago

I'm not trying to make fun of this thread, but this might amuse you!
Why stop at 48MP?
http://forums.dpreview.com/forums/readflat.asp?fo...

Michael
Loading...
+1 Vote up Vote down
amin's avatar - Go to profile

amin 67p · 873 weeks ago

That's funny. Somebody needs to tip Engadget!
Loading...
+2 Vote up Vote down
1001noisycamera's avatar - Go to profile

1001noisycamera 98p · 873 weeks ago

Thanks! That was great timing for this post, it came right before the new 15-megapixel Samsung NV100hd TL34hd 1/1.72" camera.
0 Vote up Vote down
wolfie's avatar

wolfie · 873 weeks ago

Having seen the 14MP Samsung cameras just announced it sadly obvious that the megapixel madness has gone up another level.
0 Vote up Vote down
Anonymous's avatar

Anonymous · 872 weeks ago

I decided to post this here instead of emailing Amin so I can keep my anonymity and my job. I work at a high-end camera store and Friday of last week our Nikon rep came into our store and gave us a bit of information about a brand new Nikon P&S camera. The P6000 is coming and it will have RAW and higher resolution. However, he said that Nikon is coming out with a high-end P&S camera above the P6000 that will use the 10mp APS sensor from the D60 and be geared toward pros and serious enthusiasts. This is a DEFINITE FACT, no B.S. We could not get anymore info out of him as far as body design, or lens focal length. We'll just have to wait and see. He did say that he is pretty sure it will be announced at or before Photokina.
1 reply · active 872 weeks ago
Loading...
+1 Vote up Vote down
amin's avatar - Go to profile

amin 67p · 872 weeks ago

Has this rep given you a correct scoop in the past on a different product? This is certainly something I hope comes true. It's only a matter of time before it does. Prior to Photokina would be unexpected but great!
0 Vote up Vote down
Anonymous's avatar

Anonymous · 872 weeks ago

Yes. Not 100% sure it will be ready for Photokina, but we hope it will be.
0 Vote up Vote down
maths's avatar

maths · 808 weeks ago

The following page has a comparison of 1Ds3 and D700 at iso 12800, both files shot with the same aperture and shutter speed through the same lens (Nikkor 85 f1.4 via adaptor). There is no significant difference between the two results, supporting the correct proposal that sensor area matters and pixel size doesn't:
http://cyberphotographer.com/megapixelmyth

Post a new comment

Comments by

Subscribe to: Post Comments (Atom)

 
Copyright 2007 | Andreas08v2 by GeckoandFly and TemplatesForYou | Design by Andreas Viklund
TFY Burajiru