I'm curious what others think about his first observation. That higher pixel densities require more care to look sharp at 100% views is no surprise. The statement that higher pixel densities require more technique even when both images are viewed at the same reproduction size is surprising: he's essentially saying that images with slight motion blur from high-mexapixel cameras don't downsize cleanly, and not only do they not look better than a lower mexapixel camera, he claims they look worse (less perceptually sharp).
IMO, that possible lack of sharpness due to interpolation artifacts can be overcome in PP. What do others think?
p.1 #3 · Ming Thein on pixel density and sharpness
Al, I think that Ming Thein (not Thien) was only saying that if you downres a 36MP to something greater than 1/2 the resolution, i.e. >18MP, then the fractional downres doesn't let the interpolation come out as sharp as a sharp 18MP camera, for example. I don't find this contentious, but I also don't find it interesting. I bought my D800 fully knowing that it only makes sense if you maximise resolution as much as you can.
p.1 #4 · Ming Thein on pixel density and sharpness
This doesn't make sense to me. Digital cameras are sampling devices of an analogue world. If you shake a D700 or D800 an equal amount, you'll still always have more recorded detail with the D800 even if it's blurred.
Let's think about a scene with a really small window that has a painting you can see through it.
Even with a small but equal amount of motion blur, you'll see more details of what that painting might look like on a 1:1 pixel level from the D800 shot than the D700 shot.
p.1 #5 · Ming Thein on pixel density and sharpness
And how does that relate to the downsampling process?
Essentially, what Ming Thein is saying, I think, is that the downsampling process cannot match the amount of detail natively available at the lower level, unless it exceeds it by a certain amount. This makes sense, because it is not only about how much detail is captured, but also about how it is turned into discrete values when downsampling.
p.1 #7 · Ming Thein on pixel density and sharpness
Well, I think he's saying regardless of the downsampling process, it can't match the sharpness of the natively lower-megapixel sensor. Which I'm not sure I believe, in any sort of practical sense. I don't think there's would be any detail (resolution) being lost, and I think any difference in sharpness (acuity) could be corrected for in PP. But I have not the done the tests, so I can't say for sure.
To put it in specific practical terms, Thein is suggesting you need to shoot the higher-MP camera at a higher shutter speed handheld to get equivalent results, and I'm not sure why that would be the case.
p.1 #8 · Ming Thein on pixel density and sharpness
FlyPenFly wrote:
So he's just saying, use a good downsampling process appropriate for the image? Yeah, thanks.
Heh, not quite He is saying that if you downres 36MP to 24MP, you might get a better image from a native 24MP camera.
Imagine downressing 36MP to 35MP. Is it sharper? I doubt it. The native 36MP will be sharper than the slightly interpolated 35MP. He is simply talking about where the crossover point is.
I don't think it is a super-interesting point. If you need 36MP for some of what you want to do, get a D800/D800E. If not, get something smaller.
p.1 #9 · Ming Thein on pixel density and sharpness
alwang wrote:
Well, I think he's saying regardless of the downsampling process, it can't match the sharpness of the natively lower-megapixel sensor.
Well, no, he does say that if you downres 36MP to 12MP, say, you will get better results. He is just talking about where the crossover point is.
To put it in specific practical terms, Thein is suggesting you need to shoot the higher-MP camera at a higher shutter speed handheld to get equivalent results, and I'm not sure why that would be the case.
Not to get equivalent results, to get results which are as good *at the pixel level*, i.e. to get as sharp pixels. This makes total sense.
He also talks a bit about what if the 36MP are slightly fuzzy, and the 12 or 24MP are sharp? This to me is uninteresting and academic. If you buy a D800, you figure out what you need to do to get sharp pixels, and accept it if you can't for whatever environmental reasons. You don't sit around comparing to 12MP and 24MP cameras and try to decide which is better. I mean, does he go out thinking "today I am going to slightly shake the camera when I take photos, so I'd better bring the D600 or the D700".
p.1 #11 · Ming Thein on pixel density and sharpness
does he go out thinking "today I am going to slightly shake the camera when I take photos, so I'd better bring the D600 or the D700". good one, Carsten
p.1 #12 · Ming Thein on pixel density and sharpness
carstenw wrote:
Heh, not quite He is saying that if you downres 36MP to 24MP, you might get a better image from a native 24MP camera.
Imagine downressing 36MP to 35MP. Is it sharper? I doubt it. The native 36MP will be sharper than the slightly interpolated 35MP. He is simply talking about where the crossover point is.
I don't think it is a super-interesting point. If you need 36MP for some of what you want to do, get a D800/D800E. If not, get something smaller.
I'm not buying, at least not yet. Assuming proper technique (naturally, we can't let incompetence be a factor), BOTH a native 24mp image and 36mp image down sampled to 24mp are interpolated images. In BOTH cases more data is being compressed and interpolated into discrete chunks. The native 24mp is no more pure than the other. I guess the question is whether 2 successive smaller interpolations (scene to 36mp to 24mp) better or worse than one larger interpolation (scene to 24mp). Certainly the quality of the interpolation matters, but given optimal interpolation methods (if they actually exist), should there be a difference?
p.1 #13 · Ming Thein on pixel density and sharpness
carstenw wrote:
Not to get equivalent results, to get results which are as good *at the pixel level*, i.e. to get as sharp pixels. This makes total sense.
This is a quote from the article, with my bolding:
This is one of the reasons why less pixels might actually produce a perceptually sharper/ crisper image for a given reproduction size, providing that this size is reasonable for the amount of resolution you’ve got in the smaller image.
What he says makes total sense if we're talking at the pixel level. If we're normalizing for the same reproduction size, I don't buy it.
p.1 #14 · Ming Thein on pixel density and sharpness
carstenw wrote:
Heh, not quite He is saying that if you downres 36MP to 24MP, you might get a better image from a native 24MP camera.
If you have no idea what you're doing, it's certainly possible.
Imagine downressing 36MP to 35MP. Is it sharper? I doubt it. The native 36MP will be sharper than the slightly interpolated 35MP. He is simply talking about where the crossover point is.
p.1 #15 · Ming Thein on pixel density and sharpness
I seem to recall theSuede saying the opposite in terms of larger pixel counts and down sampling, but I guess we need him to chime in, because I can't remember exactly.
p.1 #16 · Ming Thein on pixel density and sharpness
What I take away from this, also a part of my own experience using the 7D, is that you may require faster shutter speeds than what you're be used to, depending on what you were previously using.
p.1 #19 · Ming Thein on pixel density and sharpness
carstenw wrote:
Essentially, what Ming Thein is saying, I think, is that the downsampling process cannot match the amount of detail natively available at the lower level, unless it exceeds it by a certain amount. This makes sense, because it is not only about how much detail is captured, but also about how it is turned into discrete values when downsampling.
It does not make sense. If we assume a superior lens and technique, the original with the lower pixel count will be soft when there is an AA filter, or messed up when there is no AA filter. I would always pick the original with the largest pixel count as a starting point, provided the camera had a decent AA filter in place. Once you are in the digital domain, you have more flexible filters at your disposal than the hardware AA filter of the lower-pixel-count camera. There is nothing difficult or magic about fractional resampling factors if you use proper tools. There are also subjective aspects, because both direct capture and downsampling & sharpening are always a compromise between preserving original detail and minimizing artifacts.
p.1 #20 · Ming Thein on pixel density and sharpness
FlyPenFly wrote:
Certainly.
When I use my iPhone to take photos, I can be a bit more sloppy than I have to be with a DSLR.
Not just large swings like that, but even going from one DSLR to another can require a change in shutter speeds. When you become really familiar with shooting an event where things are same each time, it can really screw with you and leave you wondering what's going on. Pixel pitch isn't exactly the first thing that pops into mind!
