kbg32
neo-romanticist
It's an interesting comparison between these two cameras, but I wouldn't exactly say that in general 6 megapixels is as good as 10. It depends on the camera and the type of chip. The Epson Rd-1 and Nikon digitals use chips manufactured by Sony. Quality has a lot to with the size of the chip, the algorythm that records the image, and how it is processed. All this comparison shows me is the difference between these two cameras, and an interesting comparison in that. The Epson can produce some nice clean files at high ISOs it seems.
kaiyen
local man of mystery
Marbrink,
Assuming that the metering systems on the Nikon and Epson are equally accurate (which they probably aren't, I know, but let's assume they are at least close), jlw basically says that while he did set the shutter speed to different speeds, he did so in shutter-priority mode. That means the aperture would've adjusted (automatially for the Nikon, manually on the RD1), which would've been an equivalent 1600 exposure.
allan
Assuming that the metering systems on the Nikon and Epson are equally accurate (which they probably aren't, I know, but let's assume they are at least close), jlw basically says that while he did set the shutter speed to different speeds, he did so in shutter-priority mode. That means the aperture would've adjusted (automatially for the Nikon, manually on the RD1), which would've been an equivalent 1600 exposure.
allan
Issy
Well-known
Nikon Bob said:
At ISO 400? You do know the C5050 uses an internal noise reduction algorithm at ISO 400, right?
ampguy
Veteran
very interesting test jlw, the RD1 definitely works well in low light. Lot's of factors already mentioned could be responsible for the Nikon's noise factor, but I would think high on the list would be what you set the Nikon's high-iso NR reduction at:
• Normal
• Low
• High
• Off
if you recall?
• Normal
• Low
• High
• Off
if you recall?
dcsang
Canadian & Not A Dentist
Throwing out all the body of the post and just referring to the title where in the same camera/lens/etc. were used save for different size (but similar manufacture brand) sensors; the answer would be "that depends".
If you're printing nothing but 4x6's then you're correct and 10MP is no better than 6MP.
If you're going to crop and upsize the image then you're wrong and the 10MP sensor will offer you the ability to do this vs the 6MP sensor.
Dave
If you're printing nothing but 4x6's then you're correct and 10MP is no better than 6MP.
If you're going to crop and upsize the image then you're wrong and the 10MP sensor will offer you the ability to do this vs the 6MP sensor.
Dave
N
Nikon Bob
Guest
Issy said:
I use both cameras at their lowest iso settings and I have tried the Pana at 400 iso with not that pleasing results. Not much experience with digital but I assume most, if not all, digital cameras have some sort of NR reduction algorithms as part of their colour engine. On some you may be able to turn off or at least reduce the NR effect. You can on the Pana. I am thinking that they both have the same size sensors and that going to 10mp from 5mp left the files noisier at low isos. Just an impression nor necessarily a fact. I just prefer the Oly files.
Bob
jlw
Rangefinder camera pedant
marbrink said:
How so? Do you mean in terms of their appearance, or how I shot my photos?
One reason I was careful NOT to refer to this as a "test" was that I can't even be sure the light levels were exactly the same from one night to the next. It's pretty common, especially at this particular theater, for the director to look at one night's rehearsals and then tell the technical director to brighten up various scenes for the next night's.
What I did was look at histograms occasionally and try to use settings that would put the highlights as far as possible to the right without clipping -- "expose to the right," as the saying goes.
Incidentally, here's a follow-up (which also applies to another question): After I saw these results, I went back into the D80's custom functions and changed the "High ISO Noise Reduction" setting from "Normal" (where it had been when I made these pictures) to "High." Subjectively, it seems to help, without costing much fine detail -- although so far I haven't had a chance to make another back-to-back comparison like this one.
jlw
Rangefinder camera pedant
dcsang said:
Which was exactly the point I was trying to make. All other things assumed to be equal, I had assumed that the newer camera with the higher pixel count would produce pictures that subjectively looked better.
What I was surprised to find was that, in this specific case, all other things were NOT equal -- specifically the perceived "smoothness" of the images -- and the result was that the images from the lower-pixel-count camera looked better -- at least to me.
Well, looking at the pictures I shot, I'd have to say that I would have been very reluctant to upsize the 10mp picture. Even when resampled to the same size, it looked significantly rougher than the 6mp picture; enlarging it further would only have made it look worse.
I think I know what you're getting at -- all other things being equal, a 10-mp image should provide more useful image data than a 6-mp image.
What I'm saying is that, in the real world of practical picture-taking, you can't always assume that all other things ARE equal!
As I said up front: My original assumption was that my new 10mp camera would just naturally "take better pictures" than my old 6mp camera, and I was very surprised to find that under these specific conditions that was not the way it turned out!
I'm not claiming that I proved anything that would apply to different cameras or different picture-taking conditions. All I'm saying is that it pays to be careful about your preconceived notions. The moral of the story, assuming there IS a moral to this story: Be familiar with your equipment and know what it can do, rather than just assuming that X (whatever X is) will always work better than Y (whatever Y is.)
My next step: Try out the D80 with its "High ISO Noise Reduction" custom function set to "High" instead of "Normal," and see if that gets its EI 1600 results to look more like the R-D 1's results.
marbrink
Established
jlw,
Did you use the same aperture for both shots?
Did you use the same aperture for both shots?
jlw
Rangefinder camera pedant
marbrink said:
I honestly don't remember. I know from the EXIF data that I was using the D80 at f/1.8, but the R-D 1 doesn't record the working aperture in its EXIF data (since there's no way for it to get that info from the lens.) I was using a 50/1.5 Nokton and the light was fairly dim, so I'm guessing I was somewhere between f/1.5 and f/2.
More to the point, I'm not at all sure the light level was the same on both nights. During the first dress rehearsal (the R-D 1 one) not all the lights had been focused and not all the cues had been finalized; by the second rehearsal (the D80 one) this work had been finished. Normally, focusing the lights makes the stage a bit brighter, since less of the light from each instrument is going to waste.
I'll concede that I may have given the D80's example picture a bit less exposure than would be optimal, and this may have made its noise performance in the dark areas seem a bit worse. But you can't say that the facial tones, for example, are underexposed, and these also show the "granular" texture that surprised me.
I also want to point out that although I showed only two example images, I shot more than 250 pictures each night -- and when I said that the D80's results were subjectively more "textured" than I expected, when compared to the R-D 1's results, that was based on my overall impression of the entire "take."
I'm a bit surprised at the amount of interest (and in some cases what almost seems like acrimony) that has been generated by what I intended to be nothing more than a simple, casual observation of some real-world results that I found surprising.
For all of you who feel I didn't give the D80 a fair shake, or that I should have posted something better documented and controlled, well, I promise I'll do something closer to an organized test when I get the chance -- just don't hold your breath on when THAT will be!
In the meantime, although I still maintain that my approach was too casual to make this sort of scrutiny worthwhile, I'm attaching the original ACR histograms of both images (with identical values set in the conversion dialog boxes) for you quantitative-methods fans. Yes, these histograms definitely provide a "smoking gun" for the contention that I underexposed the D80 shot (although I don't know what else I could have done without clipping detail on the highlight side) -- so if you want to use this as evidence to conclude that everything I wrote in my original post was BS, then feel free! Me, I'll keep shooting my pictures my way, and you can shoot yours your way, and then we should all be happy, right?
Attachments
Pherdinand
the snow must go on
It is not only a matter of pixel numbers. There's a software behind the sensor. It can matter a LOT for the same sensor, in what camera it is installed and what does the software do on the image before you see the result.