reply to post by free_spirit
free_spirit, your question is pertinent, but please let me point out that mine wasn't a tutorial: there are tutorials all over the internet, and even
wikipedia entries about bilinear, trilinear, bicubic etc interpolations make the point
Whenever i'd want to publish some tutorial, you would be
able to read somewhere
TUTORIAL: did you read "tutorial" somewhere? No, you didn't, because that was just a post which purpose was
to make the point about image resizing related to THIS thread: if it happens to you to ignore the purpose of some of my post, then please be so kind
to
ask rather than guessing and trying to pass your guesses as facts, thank you in advance for that.
First and most important: From what source did you get the capture? Or should I say
these are images from a video, where did you get the exact frame you showed?
This is the image from which i've started
I took it from
your post,
this one.
I took only that, because i never use enhancements, enlergements etc made by others.
What software you used to make the closeup, Photoshop or any other brand.
Paint Shop Pro.
What tool you used to make your close up.
Image resize.
What was the original size and resolution of the source where you captured the frame.
You already know the answer to this question since you have posted it.
You talk about resize, did you resized the video itself or just the frame you captured from that source.
Ive resized just a part of the frame: question marks when you pose questions no, huh?
My image was captured from the high quality footage at NTSC standard size(not the Youtube copy) and the still frames preserved the original
resolution. This is why my images are sharp and clear even at zooming. Haktan Akdogan did the same from the original tapes. The images I posted are
courtesy of him and Mr. Yalman.
I don't know what you mean for "sharp" and "clean": in resizing, what does matter is not how nice some image looks, but how close it is to the
reality:
for example, in the following image, you can try by yourself to do this experiment: calculate the ratio of these segments: in order to don't be
"biased" in my calculation, ive even reduced the differences of the ratios:
Regardless each single value, in the "Ratio" column you should have the same values, or light variations: you should have a series of all 1, or all
1,6 or all 1,8.
And this is a quick comparison between image and zoom: the differences are obvious even to the naked eye: and here, (as said i've kept the
differences VERY low) you have ratios from 1 to almost 1,8 (meaning, 80% of ERROR). So either the zoom is not of the same image (and it says all about
the accuracy of whomever made the enhancement) or the Zoom does NOT show the actual resized image, but something ELSE.
Now let me try to do a zoom with pixel resize:
I will zoom about the same area:
Measures:
Ratio:
You can check the accuracy of my measurements here
Of course it's just a
coincidence that the ratio is the same, i know, i know.
You say your image is what the camera recorded. I have to disagree with that. The
close up images from the original footage prove it. Remember the big difference is
the source wich results in better resolution and quality.
It will be necessary to know those answers internos to understand the process you did
to get that results. Because if not I'm afraid your tutorial and image are not correct.
Yes, i've said that THIS
is what was actually caught on camera, of course according to the image that you have posted.
This enlargement shows this area:
I didn't take the photo personnally so i have no ide: as far as i know, it could be anything, but this doesn'ty change the things at all.
Now, if we are done with your third degree, we can talk about your enhancements: you say:
Now, we all can see a big difference in both images, one is unclear due to pixelization and the other is more clear even at high zoom, both of
them same size and values.
Yes, we can see a big difference, but the reason is NOT the one that you are trying to pass as truth, free_spirit.
Simply, i've resized the base image with the only method which preserves the actual image: PIXEL resize. You didn't use the pixel resize method, but
a smart resizing method: my image is an enlargement of the actual one (IF the one that you have posted can be called actual one). Your image is a
Computer Generated one, with a bicubic algorithm: in that way the result that you get is is frequently used for scaling images and video for display
(see bitmap resampling). It preserves details better than the bilinear algorithm, but unavoidably, bicubic interpolation causes overshoot, which
increases
acutance.
In photography, acutance is the edge contrast of an image. Acutance is related to the amplitude of the derivative of brightness with respect to
space. Due to the nature of the human visual system, an image with higher acutance appears sharper even though an increase in acutance does not
increase real resolution.
However, due to the negative lobes on the kernel, it causes overshoot (haloing). This can cause clipping, and is an artifact (see also ringing
artifacts), but it increases acutance (apparent sharpness), and can be desirable.
en.wikipedia.org...
In the example image, two light gray lines were drawn on a gray background. As the transition is instantaneous, the line is as sharp as can be
represented at this resolution. Acutance in the left line was artificially increased by adding a 1 pixel wide darker border on the outside of the line
and a 1 pixel wide brighter border on the inside of the line. The actual sharpness of the image has been decreased because the transition takes place
across 4 pixels, but the apparent sharpness is increased because of the greater acutance.
So, basically, the method that you are applying is NOT reliable, but can (and SHOULD) be used whenever you need to resize any image for all the
purposes, except.... ANALYSIS.
It will be necessary to know those answers internos to understand the process you did
to get that results. Because if not I'm afraid your tutorial and image are not correct.
As said, mine was NOT a tutorial, and with all the due respect, i don't think that you are entitled to make any judgement in imageering, while i DO
urge you to read some tutorial, then try again.
Thank you for your post.
[edit on 7/6/2009 by internos]
Is it from the same hoaxer as last time? 
