Do More Mega Pixels translate in a richer photo?

[deleteme]

Short answer: No

I started with a Canon 10D at 6MP and was astounded art the quality it could yield given certain subjects and good technique.
I now use 5Dmk3s and while they have a lot of resolution, that resolution is apparent only when I use the best technique and is more apparent on some subjects more than others.
The reality of high res cameras is that the majority of situations we find ourselves are less than optimal thus giving us files that are large without the detail we hoped would be there.
In studio and on location under tightly controlled conditions I can see good results but if I am shooting handheld there is no advantage over my 10D.

 

[sanj]

Short answer: No

I started with a Canon 10D at 6MP and was astounded art the quality it could yield given certain subjects and good technique.
I now use 5Dmk3s and while they have a lot of resolution, that resolution is apparent only when I use the best technique and is more apparent on some subjects more than others.
The reality of high res cameras is that the majority of situations we find ourselves are less than optimal thus giving us files that are large without the detail we hoped would be there.
In studio and on location under tightly controlled conditions I can see good results but if I am shooting handheld there is no advantage over my 10D.

Most of my work is handheld. Tele lenses, 800 odd ISO, f5.6, 1/500. I will not see ANY improvement? And why is this so? My first camera was 5d2. Now I am using 5d3 and 1dc - they all have similar mpixs. I do not have any experience with significant higher or lower mpixs. So do explain your theory. Pls.

 

[jeffa4444]

If someone show arguments proving that I'm wrong, I'll swallow my words, :-X but ...

Whereas the only variable is the amount of megapixel who jumped from 22 to 50, will not see any improvement on a computer screen without expand the viewing.
The additional resolution would NOT be noticeably in a 4K computer monitor. Only a 8K monitor would show a more sharp image in a 50 megapixel photo, or printing on paper larger than 1 meter.

Obviously, the new Canon 5DS / 5DSr may further improvements in color filter, bit depth, dynamic range, etc.

Not entirely true, oversampling still produces better images BUT the key will be the taking lens. The best monitor currently for the masses is the 27" 5K iMac for photo editing its a dream.

 

[ajfotofilmagem]

Short answer: No

In studio and on location under tightly controlled conditions I can see good results but if I am shooting handheld there is no advantage over my 10D.

Most of my work is handheld. Tele lenses, 800 odd ISO, f5.6, 1/500. I will not see ANY improvement? And why is this so? My first camera was 5d2. Now I am using 5d3 and 1dc - they all have similar mpixs. I do not have any experience with significant higher or lower mpixs. So do explain your theory. Pls.

I have not seen any controlled test of the new 5DS / 5DSr but I suppose it will not show real advantage over the sharpness of 1DX, while above ISO1600.

On the other hand, I hope that we will see a very noticeable improvement over 1DX, when in ISO100.

Time will tell.

 

[MaxFoto]

Shot this on Sunday with the low res 5D3 and the crappy 24-70L II.

 

[candc]

My response is completely based on an apples to apples comparison...same photographer, same glass,
same generation camera body, same post processing, same everything excepting...difference in MP.

If you take that 50 MP image and, for whatever reason, downsize it to a 22 MP size...you will have a
better IQ image than you would have had taking that same image with a 22 MP body.

Note that I did not use the term crop, but said...downsize.

That is what excites me, not the high mpx but the ability to downsample to a sharp clean image that is still big enough to be useful. At ful res it will have all kinds of cropping room and it may be a really good camera for an unintended market, bird photographers.

I don't think the high mpx on its own will make any difference for viewing at normal sizes but it gives you more to work with. Aside from bigger file sizes I don't see a downside.

 

[Dylan777]

Since I lost half of my Canon DSLR gear, I was forced to shoot with just 12MP camera + 55mm lens...

 

[Rocky]

My response is completely based on an apples to apples comparison...same photographer, same glass,
same generation camera body, same post processing, same everything excepting...difference in MP.

If you take that 50 MP image and, for whatever reason, downsize it to a 22 MP size...you will have a
better IQ image than you would have had taking that same image with a 22 MP body.

Note that I did not use the term crop, but said...downsize.

Do you have data to substantiate it?

 

[candc]

i think downsampling can give really good results. here is a downsampled d800 to match d3 image. and a downsampled one of mine shown next to the full rez one taken on a 70d. i think for downsampling to be really effective you need to use the bicubic sharper algorithm and go 50% in both directions.

 

[bmwzimmer]

I shoot a lot on landscape mode and occasionally like to crop to portrait and recompose the subject to give it a different look. I'm no professional and my humble 6d is no sports camera but I like to shoot a 4-5 shot bursts in landscape of a kid hitting a baseball or blowing out candles and I like to save or post on facebook 2 shots out of the burst. One in landscape and one in portrait or vica versa that I cropped that looks like it was shot with 2 different cameras. It gives that moment a different look... For me the extra resolution doesn't matter too much but I see lots of value there for professional or very serious sports/wildlife guys where they're shooting at fast shutter speeds so the images should be pretty sharp.

 

[pcdebb]

Interesting lesson for me here. I've battled with this question for ages, and battling it now as to if I am going to upgrade from a T3. Truthfully i want more glass :

 

[Ryan_]

i think downsampling can give really good results. here is a downsampled d800 to match d3 image. and a downsampled one of mine shown next to the full rez one taken on a 70d. i think for downsampling to be really effective you need to use the bicubic sharper algorithm and go 50% in both directions.

I'm confused about the buzzard image, do you mean you uprezed and then down rezed the buzzard image to get the one on the right(sharper image)? Because they look like exact same photo but that one on the right looks way more detailed. Just wondering how?

 

[candc]

The one on the right is just downsampled with dxo using the bicubic sharper algorithm shown at 100% The one on the left is full res viewed at 50%. Remember that sensors use a bayer array and each pixel only records either red green or blue. Then Bayer interpolation tries to combine that info into an image with the same resolution. Downsampling with the bicubic sharper algorithm basically combines 4 adjacent pixel which each contain separate color information into 1 pixel. In effect it's like having pixels that are twice the size that record all the color information. So if you start with a 50mpx sensor image and downsample it then it is better than one that uses that same starting resolution as the downsampled image.

 

[privatebydesign]

The one on the right is just downsampled with dxo using the bicubic sharper algorithm shown at 100% The one on the left is full res viewed at 50%. Remember that sensors use a bayer array and each pixel only records either red green or blue. Then Bayer interpolation tries to combine that info into an image with the same resolution. Downsampling with the bicubic sharper algorithm basically combines 4 adjacent pixel which each contain separate color information into 1 pixel. In effect it's like having pixels that are twice the size that record all the color information. So if you start with a 50mpx sensor image and downsample it then it is better than one that uses that same starting resolution as the downsampled image.

All you are doing there is demonstrating your video cards bad interpolation of a 50% view. You can only judge sharpness on a monitor at 100%, that is it. The reason your downsampled buzzard looks so sharp by comparison is because your graphics card is just passing the 100% signal through, it isn't dynamically resizing it like it is to the 50% view one.

Bayer arrays don't mess with detail or luminance of a pixel, they just, potentially, mess with the colour of a pixel. The AA filter messes with the detail, which is why all images from any one model camera (if it has an AA filter) all require the same level of input/capture sharpening.

 

[candc]

Right, and when you look at the two images we are really talking about Acutance. The cheetah images are really better at demonstrating what I am getting at which is that an image captured with a higher resolution sensor and downsampled is better than one captured with a sensor of same size as the downsampled image.

 

[privatebydesign]

Right, and when you look at the two images we are really talking about Acutance. The cheetah images are really better at demonstrating what I am getting at which is that an image captured with a higher resolution sensor and downsampled is better than one captured with a sensor of same size as the downsampled image.

I wouldn't necessarily disagree with that, just your buzzard illustration. But there are lots of caveats to 'better' and any relevant comparisons.

Here is a 1Ds MkIII and a 7D, both cropped to the same area, the 1Ds MkIII is at 100% the 7D is down sampled to match the resolution. Ignore the slight difference in focus and the dof is, actually, the same, they are both reproduced the same size so there is no dof difference between crop and ff in this specific situation. Same lens, same manual exposure and flash, heavy tripod, mirror up cable release, 10X manual Live View focus on both. The important bit about this comparison is that they are the same sensor area, the same generation sensor tech, just one has over twice the pixels as the other that have been downsampled.

The 7D crop is a downsampled 722,000px, the 1Ds MkIII is a native 323,000px, I cant see the differences at well over twice the pixels, can you?

The other thing with this comparison, I deliberately set it up to favour the crop camera, i'e 200iso (the 7D's optimal iso) and f5.6, again optimal for the 7D with a 300mm f2.8. At more taxing iso levels the differences could be quite different.

 

[candc]

they both look about the same but #1 seems to be a bit sharper and more contrasty. which one is which?

 

[privatebydesign]

they both look about the same but #1 seems to be a bit sharper and more contrasty. which one is which?

That is because the bottom one is back focused slightly, ignore the ruler and look at the paper towel and the bottom one is sharper with more contrast. When doing stuff like this focus is everything, and way beyond AF capabilities.

 

[candc]

I guess your right then. It seems this "more megapixels" business is over the top and reached its useful real world limit. If you have to look that hard then its not worth dealing with the huge files.

 

[privatebydesign]

I have been saying that for a long time

There are a few people who need or want these new sensors and fewer who will actually get the best out of them, but even me who is a full time pro and who's work often gets printed big and glossy will rarely see the difference when downsampled. Like I say, noise will be a different test, but the new sensors have been capped low anyway.

There are some very good uses for the new sensors, and I'd still get one if my work goes that way, but getting 'better' IQ from down sampling isn't a good enough reason.