Home

Please support mflenses.com if you need any graphic related work order it from us, click on above banner to order!
SearchSearch MemberlistMemberlist RegisterRegister ProfileProfile Log in to check your private messagesLog in to check your private messages Log inLog in

Sensor size , crop etc
Post new topic   Reply to topic
Home   Digital SLR Cameras
View previous topic :: View next topic  


PostPosted: Mon Dec 12, 2011 5:52 am    Post subject: Reply with quote
woodrim wrote:
eawhead: FYI, when I crop my image 100% I get an area about 25% the size that you did.


I believe you're doing it wrong, then. No offense (really). I know what I'm doing.

Remember, the "area" you get from your "100% crop" is entirely dependent on where & how you trim the image. Your 100% crop can be twice that rectangle, if you choose to trim it like that.

A 10MP image is 3648 pixels wide by 2736 pixels tall. The image of the Porche you uploaded is roughly this size, but slightly different aspect ratio: 3872 x 2592px. A 100% crop of that at 1024 x 768px would be this:




When you click on this, you may have to click on the image once, since your browser may be resizing it to fit your screen. I work on a 2560px X 1440px screen, so I tend to trim generously Very Happy


Here's a 100% crop of the same image at 480 x 360px





Note that the image "size" is smaller, but the "zoom factor" is the same as the above. It's just a smaller portion of the above trim.


PostPosted: Mon Dec 12, 2011 7:30 am    Post subject: Reply with quote
iangreenhalgh1 wrote:
I am following you now, but I'm still not sure of you're point, sorry if I'm being dense!

For me, it boils down to the resolution of the sensor vs the resolution of the lens, lines per inch is a good measure. Supposing the sensor has 100lpi resolution, the lens would need to have at least that resolution to produce images with the maximum sharpness the sensor is capable of. The higher the resolution of the sensor, the better the lens you need, using my comparison of 5DMkII to NEX-3, the lens would have to be 17% sharper to max out the sensor's resolution. In short, if a lens isn't really sharp on a 5DMkII, it will look even less sharp on a NEX-3.

In real world terms, this means that if someone tests a lens on a 5DMkII and it lack's sharpness to some degree, it will suck on a higher resolution sensor that packs a high MP count into a much smaller sensor area.



I understand the theory behind what you're saying, but let's just say that none of these sensors are going to out-resolve the lens, for the sake of argument. You have two cameras, same sensor size, one 10MP, one 20MP. You set them up in the same spot, same lens, and shoot at the same object.


Assuming a 3:2 aspect ratio sensor, a 10MP image is 3872 x 2584px, a 21MP image is 5616 x 3744px. Remember, you're seeing the same image when you're zoomed out (same FOV).

Now, when you compare the "100% crop" of these two images "at the same image size", say 800 x 600px, you are going to be comparing a portion of the image that is half the area for the 20MP sensor, blown up to the same size as the 10MP image. An easy way to think about this is comparing the 100% crops at 3872 x 2584px. For the 21MP image, you are going to crop the 5616 x 3744px image to 3872 x 2584px, a rectangle that is half the area. So, as an example, in a shot like the following, you will be comparing the entire image with the portion of the image in the red rectangle blown up to the same size as the whole.





What does that mean? You will increase the size of every little fault to twice the size. Any motion/camera shake blur: twice the size. Circle of confusion: twice the size. Chromatic aberrations: twice the size (width). Am I making sense?


PostPosted: Mon Dec 12, 2011 7:42 am    Post subject: Reply with quote
I haven't paid any attention to Photodo for many years. The first few times I looked at it it differed too much from my own experience so I heve never bothered with it.

Last edited by jjphoto on Sun Mar 17, 2013 9:24 pm; edited 1 time in total


PostPosted: Mon Dec 12, 2011 11:39 am    Post subject: Reply with quote
I have only one 28mm - The Olympus 28mm f/3.5 Absolutely love it, and for 20€ it was a very good buy. Great colors and a super small and light lens.
There's Pentacon 29mm f/2.8 waiting for me in the post office, but I still have to pick that one up Razz


PostPosted: Mon Dec 12, 2011 1:33 pm    Post subject: Reply with quote
iangreenhalgh1 wrote:
I am following you now, but I'm still not sure of you're point, sorry if I'm being dense!

For me, it boils down to the resolution of the sensor vs the resolution of the lens, lines per inch is a good measure. Supposing the sensor has 100lpi resolution, the lens would need to have at least that resolution to produce images with the maximum sharpness the sensor is capable of. The higher the resolution of the sensor, the better the lens you need, using my comparison of 5DMkII to NEX-3, the lens would have to be 17% sharper to max out the sensor's resolution. In short, if a lens isn't really sharp on a 5DMkII, it will look even less sharp on a NEX-3.

In real world terms, this means that if someone tests a lens on a 5DMkII and it lack's sharpness to some degree, it will suck on a higher resolution sensor that packs a high MP count into a much smaller sensor area.


Your quite right...Take the Sigma SD1 for example...It has a 1.5x crop sensor, with three stacks of 15 milllion photosites crammed onto it. Current Sigma lenses have proved to be totally inadequate and most are being outresolved by its sensor, rather than the other way round.
But not only that, you also have you factor in the law of quantum physics...The smaller the photosites, the lower their dynamic range and the noisier they are. Sigma have simply crammed too many photosites onto the SD1's sensor, so that whilst it may have good resolution, everything else has suffered and its clear from the samples that SD1 images dont have the inherrantly good image quality of previous Sigma DSLR's, probably due to intrusive noise reduction.
I would have prefered the SD1 sensor to be a full frame 15mp x3 Foveon sensor...Then it would have had the best image quality of any DSLR currently available, one of the highest dynamic ranges of any DSLR currently available, and far better high ISO performance that any previous Sigma DSLR to date, plus a much wider FOV per lens.
Sadly it was'nt to be, so the only sensible upgrade route for those currently using Sigma DSLR's is the 5D MkII or the upcoming 5D MkIII.


PostPosted: Mon Dec 12, 2011 1:47 pm    Post subject: Reply with quote
woodrim wrote:
[





Wow! really nice from the Vivitar CF.

I got a bunch of them

Minolta 2.8 (meh)
Komine 2.8 (meh a bit better than the Minolta)
Komine 2.8 close focus (haven't use it yet! Embarassed)
Yashica ML 2.8 (nice but a bit cold)
Komine 2.0 close focus (very nice but big)
Rokkor 2.0 (really nice but big and heavy)

Me being me, I really don't know how to handle this FL very well on m4/3 (56mm equiv)


PostPosted: Mon Dec 12, 2011 2:46 pm    Post subject: Reply with quote
rawhead wrote:

You do have to remember that my image is a 100% crop from a 21.2MP Full-Frame RAW file, not a 10MP APS-C JPEG.


I know, but I am not comparing images, I'm evaluating what I see in your crop.


PostPosted: Mon Dec 12, 2011 3:13 pm    Post subject: Reply with quote
DSG wrote:

Your quite right...Take the Sigma SD1 for example..


To be honest, the SD1 is such a silly thing that I refuse to consider it as an example for sensible development. Wink Laughing
Just look at the price! Shocked


PostPosted: Mon Dec 12, 2011 3:43 pm    Post subject: Reply with quote
LucisPictor wrote:
DSG wrote:

Your quite right...Take the Sigma SD1 for example..


To be honest, the SD1 is such a silly thing that I refuse to consider it as an example for sensible development. Wink Laughing
Just look at the price! Shocked


I feel the same way about the Pentax Q, manufacturers have managed to trick consumers into thinking that the more megapixels the better, which is totally false. The Q uses software to create bokeh, that is ludicrous to me.

I worry that the NEX-3 I have won't be replaced by a better model in terms of IQ as I worry that Sony will just keep packing more megapixels into the same size sensor and using software tricks instead of actual improvements in IQ. Ever higher pixel densities is stupid, there is a limit to the resolution a lens can provide and once you pass a certain point, no lens, even the best ones will be able to produce a sharp image. Look at the Samsung NV15, it's so noisy at all ISO levels as to be a crap camera. The NV8 it replaced which had less megapixels is a better camera, this is retrograde movement in IQ.

Probably when Canon make a successor to the 5DMkII it will have no better IQ as they will just pack more pixels into the same sensor size as Sigma have done with the SD1.


PostPosted: Mon Dec 12, 2011 4:08 pm    Post subject: Reply with quote
Ian, you are making assumptions again Wink The Nex-c3, Nex-5n and Nex-7 all have better image quality than the original Nex-3 and Nex-5. They have better high ISO performance, better colour reproduction and higher dynamic range. Plus, they seem to have countered the issues with wide angled rangefinder lenses, whilst upping the sensor resolution at the same time Smile


PostPosted: Mon Dec 12, 2011 4:22 pm    Post subject: Reply with quote
Well, I respectfully disagree, noise and dynamic range are largely functions of the pixel density, ever higher pixel densities make it harder and harder to control noise and produce a large dynamic range, there is a limit to what can be done with the APS-C sized sensor and we aren't for off reaching that limit.

I should have said that I worry that future NEX cameras will only be small improvements in IQ over the original NEX-3 and it is quite likely they will reach a point in the near future where they can't get the same level of IQ with more megapixels.

Samsung have crammed 20MP onto an APS-C sensor for their new NX200, we shall have to see how that performs, I doubt you can go much above 20MP and maintain a high level of IQ.

Just look at small sensor cameras, they have long passed the point where IQ suffers because of too many pixels in too small a space, this will also happen with APS-C cameras and my worry is the NEX-3 with a 4.77 pixel pitch represents as far as they should push pixel densities.


PostPosted: Mon Dec 12, 2011 9:03 pm    Post subject: Reply with quote
Orio wrote:
rawhead wrote:

You do have to remember that my image is a 100% crop from a 21.2MP Full-Frame RAW file, not a 10MP APS-C JPEG.


I know, but I am not comparing images, I'm evaluating what I see in your crop.


I was getting at the fact, in a not so straightforward way, that you were quick to opine that the "resolution" of the 100% crop you saw in woodrim's Porche/trunk image was great, while suggesting that the 100% crop of my image is "muddy". What I'm pointing out is that a 100% crop of a 21.1MP image is *always* going to look poorer (muddier, noisier, less in-focus, everything) than a 100% crop of a 10MP camera's image, everything else being equal.

In other words, if all you're doing is evaluating wha you saw in my crop, all I'm doing is pointing out the fact that your observation isn't as useful, unless you try to do some kind of comparison in a fair way, which is what I thought what some of were trying to do by uploading 100% crops of sample images. Maybe I was mistaken Smile


PostPosted: Tue Dec 13, 2011 10:12 pm    Post subject: Reply with quote
Therefore I think I still have the perfect camera actually.

Pentax K100D, 6megapixels on APS-C

Every lens looks good !

OK, maybe the original Canon 5D is a bit better at 12.8 megapixels on full frame. The K100D still probably beats it at high ISO though.

But anyway, the point is that the peak of technology was reached back in 2006-2007 !


PostPosted: Wed Dec 14, 2011 5:10 pm    Post subject: Reply with quote
I agree with Luis, there is a 'sweet spot' for pixel density and if you exceed it, you run into issues such as excessive noise, dynamic range limitations etc etc.

Packing more and more pixels onto small sensors is stupid, manufacturers are doing this though because consumers seem to think more MP the better.

I just shake my head when I see the latest compacts with 12MP on a tiny sensor, they have so much noise and crappy dynamic range.

Manufacturers are trying software remedies, I don't want extra processing of my images to achieve decent IQ!


PostPosted: Wed Dec 14, 2011 5:45 pm    Post subject: Reply with quote
rawhead wrote:

I was getting at the fact, in a not so straightforward way, that you were quick to opine that the "resolution" of the 100% crop you saw in woodrim's Porche/trunk image was great, while suggesting that the 100% crop of my image is "muddy". What I'm pointing out is that a 100% crop of a 21.1MP image is *always* going to look poorer (muddier, noisier, less in-focus, everything) than a 100% crop of a 10MP camera's image, everything else being equal.
In other words, if all you're doing is evaluating wha you saw in my crop, all I'm doing is pointing out the fact that your observation isn't as useful, unless you try to do some kind of comparison in a fair way, which is what I thought what some of were trying to do by uploading 100% crops of sample images. Maybe I was mistaken Smile


I only said that I appreciate the clarity and definition of the crops of the Vivitar lens. And that I found your crop not to be sharp in any point, and muddy overall.
I stand with my judgement, but don't read more into it. I am not comparing systems. In fact I am not even comparing the lenses, because it is impossible to compare two lenses from two different photos.
I only evaluated the images, i.e. the results.

Reg. your assertion that a crop of a 21 MP image is always looking poorer than a crop from a 10 MP camera, I disagree, because I have my personal experience from the 300D and 400D and 50D, and none of these cameras could make crops as good as the 5D or the 5DII.
I also disagree because you are forgetting a main point: the full frame 21MP camera makes bigger crops because it has more sensor real estate. Taking the real world object's size, the shooting distance, and the captured object's proportions on the frame as fixed parameters, the 21 MP camera has a native resolution that is bigger than the 10MP camera, and this is why the crop is larger. But bigger resolution is an advantage, not a disadvantage.
Another advantage of full frame cameras is that the sensor cells are usually not as densely packed as in an APS-C camera. Even if the full frame camera has a higher number of cells, once spread on the full frame space they are less dense, which results in a better signal to noise ratio. Which is another advantage. I could measure the advantages with no doubt when I had the 50D, which is a camera that was released almost at the same time as the 5DII, so they are technologically comparable. The 50D was a very good tool, but the 100% crops of the 5D Mark II simply blew it away, every time.


PostPosted: Thu Dec 15, 2011 1:40 pm    Post subject: Reply with quote
iangreenhalgh1 wrote:


I worry that the NEX-3 I have won't be replaced by a better model in terms of IQ as I worry that Sony will just keep packing more megapixels into the same size sensor and using software tricks instead of actual improvements in IQ. Ever higher pixel densities is stupid, there is a limit to the resolution a lens can provide and once you pass a certain point, no lens, even the best ones will be able to produce a sharp image.


Both nex-5n and nex-7 are much better then the old 14,2 Mpx sensor, in terms of colors, noise, DR. So, while I agree that higher Mpx count is a marketing trick, sometimes it works.


PostPosted: Thu Dec 15, 2011 4:36 pm    Post subject: Reply with quote
Orio wrote:
rawhead wrote:

I was getting at the fact, in a not so straightforward way, that you were quick to opine that the "resolution" of the 100% crop you saw in woodrim's Porche/trunk image was great, while suggesting that the 100% crop of my image is "muddy". What I'm pointing out is that a 100% crop of a 21.1MP image is *always* going to look poorer (muddier, noisier, less in-focus, everything) than a 100% crop of a 10MP camera's image, everything else being equal.
In other words, if all you're doing is evaluating wha you saw in my crop, all I'm doing is pointing out the fact that your observation isn't as useful, unless you try to do some kind of comparison in a fair way, which is what I thought what some of were trying to do by uploading 100% crops of sample images. Maybe I was mistaken Smile


I only said that I appreciate the clarity and definition of the crops of the Vivitar lens. And that I found your crop not to be sharp in any point, and muddy overall.
I stand with my judgement, but don't read more into it. I am not comparing systems. In fact I am not even comparing the lenses, because it is impossible to compare two lenses from two different photos.
I only evaluated the images, i.e. the results.

Reg. your assertion that a crop of a 21 MP image is always looking poorer than a crop from a 10 MP camera, I disagree, because I have my personal experience from the 300D and 400D and 50D, and none of these cameras could make crops as good as the 5D or the 5DII.
I also disagree because you are forgetting a main point: the full frame 21MP camera makes bigger crops because it has more sensor real estate. Taking the real world object's size, the shooting distance, and the captured object's proportions on the frame as fixed parameters, the 21 MP camera has a native resolution that is bigger than the 10MP camera, and this is why the crop is larger. But bigger resolution is an advantage, not a disadvantage.
Another advantage of full frame cameras is that the sensor cells are usually not as densely packed as in an APS-C camera. Even if the full frame camera has a higher number of cells, once spread on the full frame space they are less dense, which results in a better signal to noise ratio. Which is another advantage. I could measure the advantages with no doubt when I had the 50D, which is a camera that was released almost at the same time as the 5DII, so they are technologically comparable. The 50D was a very good tool, but the 100% crops of the 5D Mark II simply blew it away, every time.


So if I have two cameras, both full frame, one 12mp one 21mp, and two '100%' crops, each 1600x1200 pixels, the two images will display on screen at exactly the same size, but in the 12mp image the area shown is 1.414 times larger than in the 21mp sensor image. To display the same size areas at the same size on screen, make a 1.414x larger crop of the 21mp image, reduce to 1600x1200 pixels.

Really? A '100% crop' of 21mp ff is 'better' than '100% crop' of 12mp ff? 1600x1200 pixels of 21mb sensor is 'better' than 12mb sensor? This is against intuition saying the least magnified image should be best.


PostPosted: Thu Dec 15, 2011 5:32 pm    Post subject: Reply with quote
iangreenhalgh1 wrote:

Probably when Canon make a successor to the 5DMkII it will have no better IQ as they will just pack more pixels into the same sensor size as Sigma have done with the SD1.


Maybe yes, maybe no. I think Canon is already very much aware of the limitations of resolution vs. sensor size and how it affects IQ. Take a look at the G10, G11, and G12. Of the three cameras, the G10 has the highest megapixel count: 14mp, compared to the G11 and G12's 10mp. I think most folks will agree that, for a P&S, the G-series is outstanding. But it's because Canon wanted to clean up the noise at higher ISOs that they went ahead and dropped the pixel count, thus bowing out of the "my sensor has a higher pixel density than yours" wars.

Given that the 5D Mk-whatevers are FF sensors, however, it seems to me that there is still plenty of room for increased pixel count before running into serious IQ problems. But hey, if Canon wants to draw the line at the Mk II's 21 mp, I don't have a problem with that. 21 mp translates into 5616 x 3744 pixels, which means that a 36mm x 24mm sensor is almost 4000 ppi (3962 ppi to be precise), which is essentially the same count as the Nikon CoolScan 9000. I just don't see much of a need to go past this in terms of resolution, unless one wants to make oversize enlargements from tiny portions of an image frame. But given that the market can be irrational and that a manufacturer is often required to stay abreast of demand, regardless of whether the demand is rational or not, who's to say where it will all end up?


PostPosted: Thu Dec 15, 2011 5:48 pm    Post subject: Reply with quote
All a "100% crop" means is that you are viewing an unreduced image. If a digicam produces a 1000 pixel wide image and I determine that I want to display it at 1000 pixels, then it is already at a "100% crop." If it produces a 2000 pixel wide image and I want to display it at 1000 pixels, then out of necessity I will have to crop away half the image. Thus, the higher the pixel count, the smaller the area will be used in the frame of a 100% crop, depending on the number of pixels one wants in the final crop.

Given the same scene recorded by an APS-C sensor and a FF sensor, and assuming similar pixel densities between sensors, the FF sensor's 100% crop will show a smaller area of the image than the APS-C's sensor will. But if we have a high-density APS-C sensor and a relatively low density FF sensor, then the situation could possibly be reversed. It's all about pixel density.


PostPosted: Thu Dec 15, 2011 5:55 pm    Post subject: Reply with quote
cooltouch wrote:
All a "100% crop" means is that you are viewing an unreduced image. If a digicam produces a 1000 pixel wide image and I determine that I want to display it at 1000 pixels, then it is already at a "100% crop." If it produces a 2000 pixel wide image and I want to display it at 1000 pixels, then out of necessity I will have to crop away half the image. Thus, the higher the pixel count, the smaller the area will be used in the frame of a 100% crop, depending on the number of pixels one wants in the final crop.

Given the same scene recorded by an APS-C sensor and a FF sensor, and assuming similar pixel densities between sensors, the FF sensor's 100% crop will show a smaller area of the image than the APS-C's sensor will. But if we have a high-density APS-C sensor and a relatively low density FF sensor, then the situation could possibly be reversed. It's all about pixel density.


Yes! That is my point, it is all about pixel density!


PostPosted: Mon Dec 19, 2011 8:40 pm    Post subject: Reply with quote
cooltouch wrote:
[...]Given the same scene recorded by an APS-C sensor and a FF sensor, and assuming similar pixel densities between sensors, the FF sensor's 100% crop will show a smaller area of the image than the APS-C's sensor will.[...]


assuming similar pixel densities, 100% crops will show a similar area...

iangreenhalgh1 wrote:
Yes! That is my point, it is all about pixel density!


I think its all about magnification! Laughing


PostPosted: Mon Dec 19, 2011 8:57 pm    Post subject: Reply with quote
visualopsins wrote:
cooltouch wrote:
[...]Given the same scene recorded by an APS-C sensor and a FF sensor, and assuming similar pixel densities between sensors, the FF sensor's 100% crop will show a smaller area of the image than the APS-C's sensor will.[...]

assuming similar pixel densities, 100% crops will show a similar area...
iangreenhalgh1 wrote:
Yes! That is my point, it is all about pixel density!

I think its all about magnification! Laughing

you can talk about it another 10 years but the FF sensor will always be bigger that the crop
Post new topic   Reply to topic
Home   Digital SLR Cameras
Page 1 of 1