brainiac wrote:
I hardly see any degradation of the image here, and certainly not enough to influence a comparison between two cameras. Remember, this is the result of two consecutive interpolations, not one. A modest amount of sub-pixel USM would mitigate any barely visible softening still further. Thanks for the example - it's a win for common sense and maths, and a loss for superstition. Your concern with the detrimental effects of uprezzing is entirely academic.
Whether or not you see it is irrelevant. Just because you can't see it (or choose to ignore it) doesn't mean that it doesn't exist. Thus the argument that upresing an image doesn't destroy the image is false. Opinions based on a false premise are dubious, at best.
Again, the other points I've made are ignored. The only reason to uprez an image is to print it! But as I've mentioned, time and again, the print you make is different than what you see on the screen. Thus, such a comparison is dubious. It only shows that one image may be better in those circumstances only, but doesn't seem to touch upon any other use of the image. Printing images isn't the only use of photographs.
The first image shows a picture with 25% monochrome noise added at 800x800. The second image shows 25% monochrome noise added at 400x400 and enlarged to 800x800. The grain is magnified. Thus, despite the fact that both images had the exact same type of noise added, the enlarged image shows the noise far more prominently, giving the advantage to any image that has a higher resolution. Given the upresolution test, I've rarely seen a camera with lower resolution "win" over a camera with greater resolution.
If the purpose of the test is to test grain on a COMPUTER MONITOR, you need to treat the images as though they're going to be viewed on a computer monitor. If you want to compare grain of images in print, the only way you can do that is to actually PRINT the images. At least Alundeb printed and scanned images for comparison. If the only reason you want to upres an image is to test how they would look when you print, then print the images, scan them in and compare those images together. As I've mentioned, time and again, printed images look different than what you have on screen.
brainiac wrote:
...which is why you are even less likely to see the 'detrimental' effects of uprezzing in a print than on screen. You have made my point: any damage done is below the threshold of what a print will resolve, so ignore it and uprez at will.
But you're coming to conclusions based on a digital upresing of images. This upres will look different than a print. If the only reason to upres is to print, then you should print and compare the prints, not just the digital upresing of images. The only way looking at a digital upres on your computer monitor makes sense is if you print at < 100 dpi and I doubt many people do that.
brainiac wrote:
I hardly see any degradation of the image here, and certainly not enough to influence a comparison between two cameras. Remember, this is the result of two consecutive interpolations, not one. A modest amount of sub-pixel USM would mitigate any barely visible softening still further. Thanks for the example - it's a win for common sense and maths, and a loss for superstition. Your concern with the detrimental effects of uprezzing is entirely academic.
brainiac wrote:
...which is why you are even less likely to see the 'detrimental' effects of uprezzing in a print than on screen. You have made my point: any damage done is below the threshold of what a print will resolve, so ignore it and uprez at will.
kakomu wrote:
Whether or not you see it is irrelevant. Just because you can't see it (or choose to ignore it) doesn't mean that it doesn't exist. Thus the argument that upresing an image doesn't destroy the image is false. Opinions based on a false premise are dubious, at best.
Again, the other points I've made are ignored. The only reason to uprez an image is to print it! But as I've mentioned, time and again, the print you make is different than what you see on the screen. Thus, such a comparison is dubious. It only shows that one image may be better in those circumstances only, but doesn't seem to touch upon any other use of the image. Printing images isn't the only use of photographs.
The first image shows a picture with 25% monochrome noise added at 800x800. The second image shows 25% monochrome noise added at 400x400 and enlarged to 800x800. The grain is magnified. Thus, despite the fact that both images had the exact same type of noise added, the enlarged image shows the noise far more prominently, giving the advantage to any image that has a higher resolution. Given the upresolution test, I've rarely seen a camera with lower resolution "win" over a camera with greater resolution.
If the purpose of the test is to test grain on a COMPUTER MONITOR, you need to treat the images as though they're going to be viewed on a computer monitor. If you want to compare grain of images in print, the only way you can do that is to actually PRINT the images. At least Alundeb printed and scanned images for comparison. If the only reason you want to upres an image is to test how they would look when you print, then print the images, scan them in and compare those images together. As I've mentioned, time and again, printed images look different than what you have on screen.
But you're coming to conclusions based on a digital upresing of images. This upres will look different than a print. If the only reason to upres is to print, then you should print and compare the prints, not just the digital upresing of images. The only way looking at a digital upres on your computer monitor makes sense is if you print at < 100 dpi and I doubt many people do that. ...Show more →
Where to start with the errors in this post? You need to forget pretty much everything you think you know.
Whether or not you see it is irrelevant. Just because you can't see it (or choose to ignore it) doesn't mean that it doesn't exist.
I hadn't realised that we were arguing about stuff we can not see in photographs. Now that I understand that, I agree with everything you say. If anyone reading this thread is concerned, on the other hand, with what can in fact be seen in photographs, look no further than the uprezzing advice I gave a few pages ago.
kakomu wrote:
Also, as I've mentioned time and again, the grain is magnified when you uprez an image. This gives an unfair advantage to any camera that has a higer resolution. Here's an example of two images. http://kakomu.com/images/large.png http://kakomu.com/images/enlarge.png
The first image shows a picture with 25% monochrome noise added at 800x800. The second image shows 25% monochrome noise added at 400x400 and enlarged to 800x800. The grain is magnified. Thus, despite the fact that both images had the exact same type of noise added, the enlarged image shows the noise far more prominently, giving the advantage to any image that has a higher resolution. Given the upresolution test, I've rarely seen a camera with lower resolution "win" over a camera with greater resolution.
alundeb wrote: kakomu wroteAlso, as I've mentioned time and again, the grain is magnified when you uprez an image. This gives an unfair advantage to any camera that has a higer resolution. Here's an example of two images. http://kakomu.com/images/large.png http://kakomu.com/images/enlarge.png
The first image shows a picture with 25% monochrome noise added at 800x800. The second image shows 25% monochrome noise added at 400x400 and enlarged to 800x800. The grain is magnified. Thus, despite the fact that both images had the exact same type of noise added, the enlarged image shows the noise far more prominently, giving the advantage to any image that has a higher resolution. Given the upresolution test, I've rarely seen a camera with lower resolution "win" over a camera with greater resolution.The grain magnification is not unfair. It just shows that smaller grain (higher res camera) actually is better, given the same noise level.
I'm afraid kakomu has this the wrong way around entirely. The higher rez image, when printed at any size, will be less enlarged than the lower rez image, and therefore its grain pattern will be less amplified. This is precisely why I advocate uprezzing the lower rez image to the higher rez image's dimensions. Only then can you compare grain as if each camera had made a print of a given size. Kakomu has enlarged the wrong grain pattern. That kind of enlargement is what happens when you print the image from the lower rez camera, and that is why uprezzing is crucial to onscreen comparison.
This is one of many errors in kakomu's post which show that as yet he has an extremely tenuous grasp of this subject. I would advise anyone to be wary of this kind of confused advice which is doled out across hundreds of web sites. Even if you struggle to understand the spatial relationships in imaging, the rule is simple. To compare the performance of any two cameras, uprez the lower rez images to the dimensions of the higher before comparing 100% crops. Feel free to add USM to either image in order to match the results as closely as possible, and feel free to adjust that USM to suit the picture.
brainiac wrote:
I'm afraid kakomu has this the wrong way around entirely. The higher rez image, when printed at any size, will be less enlarged than the lower rez image, and therefore its grain pattern will be less amplified. This is precisely why I advocate uprezzing the lower rez image to the higher rez image's dimensions. Only then can you compare grain as if each camera had made a print of a given size. Kakomu has enlarged the wrong grain pattern. That kind of enlargement is what happens when you print the image from the lower rez camera, and that is why uprezzing is crucial to onscreen comparison....Show more →
Showing the pictures on screen doesn't show how they look in print. This is what you're ignoring. If you print the images and scan them, then you'll have an idea of how much of a role the increased resolution actually plays and whether or not the grain even matters. Otherwise, magnifying and exaggerating defects in an image is rather unethical if there's no basis to this claim.
brainiac wrote:
This is one of many errors in kakomu's post which show that as yet he has an extremely tenuous grasp of this subject. I would advise anyone to be wary of this kind of confused advice which is doled out across hundreds of web sites. Even if you struggle to understand the spatial relationships in imaging, the rule is simple. To compare the performance of any two cameras, uprez the lower rez images to the dimensions of the higher before comparing 100% crops. Feel free to add USM to either image in order to match the results as closely as possible, and feel free to adjust that USM to suit the picture....Show more →
My "grasp" on the subject matter is hardly "tenuous". The fact that you barely even respond to my points and resort to name calling or belittling just shows that you can barely defend it.
None of your suggestions have any real-world value. If you want to compare prints, then print the images out and scan them (as alundeb did). Then you can show whether or not your upres tests have any real-world value.
Otherwise, your comparisons are outside of what people actually do. People don't typically upres an image with the intent of keeping it in a digital form. This is the point that you are ignoring. You need to go full-nine in your endeavors. If you want to discuss real-world print tests, then you need to actually print the images.
If you want to discuss real-world DIGITAL value, consider online news sources (e.g. CNN, Sports Illustrated, etc) which have a wealth of images on their websites at < 1000px as the longest dimension. This shows that downsizing the image is the norm as far as digital representation of images is concerned.
Even in print news and magazines, the resolution of those images is far below what people use with color inkjet printers.
Moreover, any on-screen comparison at anything but 100% shows only the extent to which you can manipulate a file to look better or worse. It's pretty easy to make a 7D file look far worse than a 5D file at high ISO if you scale it down to the 5D size.
Interestingly who actually prints nowadays? I don't think I've printed an image in 12 months I find most clients just want digital files nowadays for Facebook or emailing to family or whatever
Dawei Ye wrote:
Interestingly who actually prints nowadays? I don't think I've printed an image in 12 months I find most clients just want digital files nowadays for Facebook or emailing to family or whatever
This is another point which seems to flying over a lot of people's heads. The only people I know of who print images are those on this forum. All the people I know who take photographs just throw their images online using Flickr, facebook, etc. Every person or group who has asked me to photograph things just wants them online. If they're going to print them, they're B&W photocopies on a newsletter. Printing images on photo paper is outside of the norm, it seems.
I don't get it either how upresing could be detrimental to a comparison. The argument that upresing changes the pixel data falls on itself, because pixel data are always changed before any normal presentation: web, projector, TV screen, print, etc. If the "holy unchangeable pixel" argument was vaild, it would be impossible to compare the quality of any final output, inluding prints and web resized images.
alundeb wrote:
I don't get it either how upresing could be detrimental to a comparison. The argument that upresing changes the pixel data falls on itself, because pixel data are always changed before any normal presentation: web, projector, TV screen, print, etc. If the "holy unchangeable pixel" argument was vaild, it would be impossible to compare the quality of any final output, inluding prints and web resized images.
The point is that upresing an image exaggerates the noise in an image. Exaggerating a defect is dishonest.
And, as I've stated numerous times, people don't typically upres an image with the intent of keeping it in a digital form (unless they plan on heavily manipulating it). We all know that upresing an image makes it look worse than having an intrinsically higher resolution image. This is a no-brainer. Thus there are multiple reasons why upresing an image will always make an image look worse than one with a higher resolution sensor.
The problem is that the conclusions drawn from this exaggeration are dubious. If the point of upresing an image is to show that one camera produces a better print than another camera, simply upresing an image doesn't show this at all.
Any comparison of the upresed images in digital form is deceptive, considering that digital image will look different than the printed image.
I've already enumerated that prints look different than what you see on screen and, thus far, no one has refuted this point (to any strong degree, at least). Thus, it would stand to suggest that if you want to compare prints, you need to ACTUALLY PRINT THE IMAGE. If the printed image will hide resize artifacts (as brainiac already conceded), then it would stand to suggest that the problems that we see in digital form MAY NOT SHOW IN PRINTED FORM! Thus, we cannot make any reasonable conclusion about a digital image's noise unless we print. Once we print an image, we can make conclusions about noise, resolution and detail in a print.
Thus, we come full circle: The only reason to upres an image is to print the image. Printed images will look different than digital images on-screen. Thus, any conclusion about an upresed image on-screen DOES NOT apply to the actual printed image. Thus, making on-screen comparisons with an upresed image HAS NO MERIT.
kakomu wrote:
Thus, we come full circle: The only reason to upres an image is to print the image. Printed images will look different than digital images on-screen. Thus, any conclusion about an upresed image on-screen DOES NOT apply to the actual printed image. Thus, making on-screen comparisons with an upresed image HAS NO MERIT.
When printing, there is always a relationship between the quality of the file and the quality of the print. The fact that the print looks different, does not remove the fact that the quality difference between prints is proportional to the quality diffrence between upresed files. Inspection of upresed files is very important to acheive a large print with fine detail.
For smaller output, it is still essential to compare at the same magnification.
In addition, a 100% crop is already interpolated by the RAW converter, so there is no such thing as visual comparison of uninterpolated images.
alundeb wrote:
In addition, a 100% crop is already interpolated by the RAW converter, so there is no such thing as visual comparison of uninterpolated images.
A 100% crop of the RAW file by a converter is "interpreted". However, interpolation is the insertion of information outside of the original information. This doesn't really occur at 100%
alundeb wrote:
When printing, there is always a relationship between the quality of the file and the quality of the print. The fact that the print looks different, does not remove the fact that the quality difference between prints is proportional to the quality diffrence between upresed files. Inspection of upresed files is very important to acheive a large print with fine detail.
Sure. However, the comparison between two images before printing has no merit.
kakomu wrote: alundeb wroteIn addition, a 100% crop is already interpolated by the RAW converter, so there is no such thing as visual comparison of uninterpolated images.A 100% crop of the RAW file by a converter is "interpreted". However, interpolation is the insertion of information outside of the original information. This doesn't really occur at 100%
Yes it does, because the bayer matrix makes it theoretically impossible to extract data with the same spatial frequency as the pixel frequency at 100%. It is closer to 70%.
Interpolation is not insertion of data "outside". It is insertion of data "between".
Insertion of data outside is called extrapolation.
alundeb wrote:
Yes it does, because the bayer matrix makes it theoretically impossible to extract data with the same spatial frequency as the pixel frequency at 100%. It is closer to 70%.
The in-camera jpegs tend to look pretty much the same at 100% (save for color or compression differences). Unless you have access to technical data to present nothing but the R, G or B channels of each photo site of the RAW data, this point is moot. You have to treat the image as a whole.
alundeb wrote:
Interpolation is not insertion of data "outside". It is insertion of data "between".
Insertion of data outside is called extrapolation.
When I said "outside" I meant that there is a data set that comprises the "image" and any information that isn't part of the image is outside of this data set.
kakomu wrote: alundeb wroteYes it does, because the bayer matrix makes it theoretically impossible to extract data with the same spatial frequency as the pixel frequency at 100%. It is closer to 70%.The in-camera jpegs tend to look pretty much the same at 100%. I don't find much truth in this statement.
alundeb wroteInterpolation is not insertion of data "outside". It is insertion of data "between".
Insertion of data outside is called extrapolation.When I said "outside" I meant that there is a data set that comprises the "image" and any information that isn't part of the image is outside of this data set.
In-camera JPG is just an in-camera RAW converter. The same applies here.
Insertion of data between original samples does not invalidate the image.
You need to learn everything from start about interpolation.
brainiac wrote:
Since we're on the subject of how bicubic uprezzing turns a picture of a unicorn into a picture of a pantomime dromedary, one more thing before I go to bed: a web site wishing to overcome these silly objections to uprezzing could do all three of the following.
hmm the mysterious pantomime dromedary rears it's head once again in your posts. Is this some sort of mythical beast only known to those in the UK? (or perhaps only to yourself haha, web searches mostly point only to a series of your posts haha and then only a handful of others total, all from people in the UK. most searches bring up thousands to hundreds of thousands of hits, but your pantomime dromedary brings up almost nothing other than your posts haha)
so just what the heck is this 'pantomime' dromedary camel of yours
kakomu wrote: alundeb wroteYes it does, because the bayer matrix makes it theoretically impossible to extract data with the same spatial frequency as the pixel frequency at 100%. It is closer to 70%.The in-camera jpegs tend to look pretty much the same at 100%. I don't find much truth in this statement.
[snip]
I don't think you understand what alundeb is saying. The bayered sensor captures data at a particular spatial frequency. The debayering process introduces interpolation (regarless of whether you are shooting RAW and then debayering in your RAW converter or whether the camera is doing it to create a JPEG) the result of which is that the image that gets generated does not have the same spatial frequency.
thedigitalbean wrote:
I don't think you understand what alundeb is saying. The bayered sensor captures data at a particular spatial frequency. The debayering process introduces interpolation (regarless of whether you are shooting RAW and then debayering in your RAW converter or whether the camera is doing it to create a JPEG) the result of which is that the image that gets generated does not have the same spatial frequency.
Please read my edited post. Unless people here are able to deal with the individual R, G and B channels of specific photosites, the point is moot. All we can deal with is what we receive from the camera. To start digging into the technical minutiae of the camera's technology belies the fact that none of it is relevant to looking at the final product. What we have are jpeg and RAW files and unless we can utilize another conversion method or someone actually purchases a Sigma body, any discussion is (as was said elsewhere) purely academic.
Moreover, the method used to demosaic an image is irrelevant, so long as it's consistently used to test images. It's a necessary evil that cannot be circumvented as far as digital photography is concerned, save for sensors that are similar to film emulsions with several transparent layers that are, each, reactive to different colors. However, in the end, the final image that we get from the raw converter is what's important to the testing.
Though, simply because an image is demosaiced does not mean that it's plausible to degrading an image further with other interpolating algorithms. Rather, all it will show is that if you use a certain process, one image will compare in this fashion to another image.